Closed-Loop Deep Brain Stimulation for Parkinson's Disease: Mechanisms, Clinical Translation, and Future Directions

Abstract

Closed-loop deep brain stimulation (aDBS) represents a paradigm shift in the treatment of Parkinson's disease (PD), moving from continuous, open-loop stimulation to dynamic, biomarker-driven neuromodulation. This article synthesizes recent clinical evidence and technological advancements for a research and drug development audience. It explores the foundational principles of aDBS, focusing on neural oscillopathies like beta-band oscillations as control signals. The review details methodological frameworks for implementing aDBS, including biomarker selection and programming strategies, and addresses core challenges in troubleshooting system optimization. Finally, it examines validation data from long-term outcomes and comparative effectiveness, contextualizing aDBS within the broader therapeutic landscape for PD. The integration of artificial intelligence and multi-modal data streams is highlighted as a key driver for the next generation of personalized, adaptive neuromodulation therapies.

The Principles and Neural Basis of Adaptive DBS

Deep brain stimulation (DBS) has transformed Parkinson's disease (PD) treatment, yet conventional continuous DBS (cDBS) is limited by its static nature, potentially leading to residual motor fluctuations, dyskinesias from overstimulation, or inadequate symptom control from understimulation [1]. Adaptive DBS (aDBS) represents a paradigm shift toward personalized neuromodulation by dynamically adjusting stimulation parameters based on real-time neurophysiological feedback [2]. This closed-loop system primarily utilizes subthalamic beta band (13-35 Hz) oscillations as a control signal, which correlates with bradykinesia and rigidity severity [1]. This application note defines aDBS and its core mechanisms, provides structured clinical data and programming protocols, and outlines essential methodological frameworks for research implementation.

Conceptual Framework and Definitions

Adaptive Deep Brain Stimulation (aDBS) is a closed-loop neuromodulation system that automatically adjusts stimulation parameters in response to neural biomarker feedback. Unlike static cDBS, which delivers constant stimulation, aDBS dynamically titrates therapy to match fluctuating symptom severity and medication states [1].

The core mechanism of aDBS involves a continuous cycle of sensing, analysis, and modulation. Implanted electrodes sense local field potentials (LFPs), from which specific biomarkers are extracted. The system algorithms then translate biomarker characteristics into tailored stimulation parameters, creating a real-time feedback loop [2].

Neural Biomarkers in aDBS

The subthalamic beta band (13-35 Hz) oscillation power serves as the primary control signal for current aDBS systems in PD. Evidence establishes that elevated beta power correlates with bradykinesia and rigidity severity, while successful dopaminergic medication or cDBS suppresses beta power [1]. This inverse relationship between beta power and motor performance provides the physiological basis for beta-guided aDBS algorithms, which increase stimulation amplitude during high beta power periods and decrease it during low beta power states [1].

Clinical Outcomes and Quantitative Evidence

Recent clinical studies demonstrate both the efficacy and programming complexities of chronically implanted aDBS systems. The following tables summarize key quantitative findings from real-world aDBS implementations.

Table 1: Clinical Outcomes of Adaptive vs. Continuous DBS from Home-Based Ecological Momentary Assessments (EMA)

Clinical Measure	Continuous DBS (Mean ± SD)	Adaptive DBS (Mean ± SD)	Statistical Significance (p-value)	Effect Size (β)
Overall Well-being (points)	5.92 ± 1.01	6.73 ± 1.33	p = 0.007	β = 0.8
General Movement	5.47 ± 1.22	6.20 ± 1.44	p = 0.058 (trend)	β = 0.7
Dyskinesia Severity	3.12 ± 1.70	3.18 ± 1.76	p = 0.988	β = 0.03
Tremor Symptoms	2.54 ± 1.51	2.56 ± 1.67	p = 0.988	β = 0.002
Favorable Clinical Status (% time)	81.5 ± 25%	87.0 ± 16%	p = 0.23	β = 0.37

Table 2: aDBS Programming Parameters and Patient Outcomes

Programming Parameter	Value (Mean ± SD; Range)	Clinical Implementation Notes
Number of Programming Visits	7.8 ± 3.7 (4-13)	Until satisfactory configuration achieved
Upper Stimulation Limit (mA)	2.28 mA	+0.23 mA vs. cDBS setting
Lower Stimulation Limit (mA)	1.71 mA	-0.33 mA vs. cDBS setting
Stimulation Amplitude Range	0.58 ± 0.19 mA (0.3-1.0 mA)	Individualized based on symptom fluctuation
Upper LFP Threshold (percentile)	75th percentile	Manufacturer guidance for beta power
Lower LFP Threshold (percentile)	25th percentile	Manufacturer guidance for beta power
Patient Retention on aDBS	6/8 patients	After 2-week trial period

Experimental Protocols and Programming Methodologies

Three-Step Clinical Programming Protocol

Step 1: Biomarker Identification and Contact Selection

• Procedure: Conduct "Signal Test" local field potential (LFP) recordings in OFF-medication state (approximately 12 hours after last levodopa dose) to maximize beta peak visibility. Record from all available electrode contacts.
• Biomarker Analysis: Identify oscillatory peaks within beta frequency range (13-35 Hz). In cases of double beta peaks, perform test stimulations and assess medication-induced beta power modulation to identify the most physiologically responsive peak [1].
• Contact Selection: Prioritize contacts with optimal signal-to-noise ratio for beta activity. If sensing-compatible contacts demonstrate suboptimal clinical efficacy, consider enabling unilateral sensing. In 50% of clinical cases, unilateral sensing was implemented due to suboptimal signals in the contralateral hemisphere [1].

Step 2: Threshold Definition and Stimulation Limit Setting

• Timeline Data Collection: Enable continuous LFP data acquisition over several days (typically 3-7 days) to capture natural beta power fluctuations across medication cycles and daily activities.
• LFP Threshold Calculation: Set initial LFP thresholds to the 25th (lower) and 75th (upper) percentiles of daytime beta power following manufacturer guidance. Expect strong inter-individual variance in absolute threshold values (upper threshold range: 225-3160; lower threshold range: 100-2970) [1].
• Stimulation Limits: Determine minimum effective stimulation amplitude in OFF-medication state to prevent hypokinetic episodes. Set upper amplitude limit to avoid stimulation-induced side effects (dyskinesia, dysarthria). Refine limits based on individual symptom patterns [1].

Step 3: Parameter Optimization and Adverse Event Management

• Adaptation Verification: Verify stimulation amplitude appropriately tracks beta power fluctuations. Address "stuck" stimulation (remaining at upper or lower limits) by adjusting LFP thresholds.
• Symptom Control Refinement: If symptoms persist despite proper adaptation, adjust amplitude limits. For hypokinetic episodes, raise lower amplitude limit; for hyperkinetic episodes, lower upper amplitude limit.
• Long-Term Management: Monitor for multi-day beta power trends that may require periodic threshold adjustments. Address artifact-related maladaptation through sensing contact reselection or algorithm refinement [1].

Data-Driven Research Protocol for Neural Decoding

Objective: Develop comprehensive aDBS algorithms leveraging artificial intelligence to decode Parkinson's motor symptoms from multimodal neural and behavioral signals [2].

Neural Signal Acquisition

• Data Collection: Simultaneously record LFPs from implanted DBS leads, kinematic data from wearable sensors (accelerometers, gyroscopes), and clinical symptom scores.
• Signal Processing: Preprocess neural signals with bandpass filtering (specifically 13-35 Hz for beta), movement artifacts removal, and power spectral density analysis.

Feature Engineering and Model Development

• Feature Extraction: Calculate time-frequency representations of neural signals, movement kinematics (velocity, amplitude, timing), and symptom severity scores.
• Algorithm Training: Implement machine learning models (linear regression, random forests, neural networks) to establish relationships between neural features and motor symptoms. Train models to predict UPDRS scores from neural signatures.

Closed-Loop Integration

• Control Policy Development: Translate model outputs into stimulation parameters using threshold-based or continuous modulation policies.
• System Validation: Conduct in-clinic testing with expert neurologist oversight to verify safety and efficacy before real-world implementation [2].

Visualization of Core aDBS Mechanism

The Scientist's Toolkit: Research Reagent Solutions

Table 3: Essential Research Materials and Analytical Tools for aDBS Investigation

Research Tool	Specifications	Primary Research Application
DBS System with Sensing Capability	Commercial systems with LFP recording (e.g., Activa PC+S, Percept); Sampling rate ≥ 250 Hz	Chronic neural signal acquisition; Real-time biomarker monitoring
Beta Oscillation Analysis Suite	Custom MATLAB/Python scripts for power spectral density; Time-frequency analysis	Beta power quantification; Biomarker validation
Motion Capture System	Wearable inertial sensors (accelerometers, gyroscopes); Clinical rating scales (UPDRS)	Objective motor symptom assessment; Kinematic-neural correlation
Adaptive Stimulation Algorithm	Dual-threshold control policy; LFP percentiles (25th/75th) for amplitude modulation	Closed-loop stimulation control; Parameter optimization studies
Data-Driven Modeling Platform	Machine learning libraries (scikit-learn, TensorFlow); Feature extraction pipelines	Neural decoding model development; Multi-modal signal integration

Adaptive DBS represents a significant advancement in neuromodulation, transitioning from static stimulation to personalized, biomarker-driven therapy. The core mechanism of beta-guided aDBS effectively addresses limitations of conventional DBS by dynamically adjusting stimulation to match fluctuating symptom severity [1]. While current systems demonstrate improved patient well-being and motor function, ongoing research into multi-modal biomarkers and artificial intelligence-driven decoding promises to enhance therapeutic precision for Parkinson's disease symptoms beyond the motor domain [2]. The structured protocols and experimental frameworks provided herein offer researchers comprehensive tools for implementing and advancing closed-loop DBS technologies.

Parkinson's disease (PD) is a progressive neurodegenerative disorder characterized by the loss of dopaminergic neurons and consequent dysregulation of the basal ganglia-thalamocortical (BGTC) network [3] [4]. The cardinal motor symptoms of PD—bradykinesia, rigidity, and tremor—are strongly correlated with pathological neural synchrony within this distributed network [3]. Among the various oscillatory disturbances, or "oscillopathies," observed in PD, exaggerated beta-band activity (13-35 Hz) has emerged as a particularly robust biomarker of the parkinsonian state [4]. This aberrant synchronous activity manifests across the BGTC network and exhibits a striking relationship with clinical status: it becomes elevated in the dopamine-depleted state, correlates with motor impairment severity, and is suppressed by both dopaminergic medication and therapeutic deep brain stimulation (DBS) [4] [5].

The investigation of beta-band oscillations has transitioned from a purely scientific pursuit to a foundation for clinically actionable biomarkers that directly inform therapeutic strategies. The most significant application lies in the development of adaptive deep brain stimulation (aDBS) systems, which aim to deliver stimulation only when needed by using real-time physiological feedback to guide therapy [4] [2]. This article details the experimental protocols and analytical frameworks essential for quantifying beta-band activity and leveraging it to optimize closed-loop neuromodulation for Parkinson's disease.

Quantitative Features of Beta-Band Biomarkers

Research has identified several distinct but interrelated features of beta-band activity that serve as valuable biomarkers. These features can be quantified from local field potential (LFP) recordings and are summarized in the table below.

Table 1: Key Beta-Band Biomarkers in Parkinson's Disease

Biomarker Feature	Description	Measurement Technique	Clinical/Experimental Correlation
Spectral Beta Power	Magnitude of oscillation in the 13-35 Hz frequency band [4]	Power spectral density analysis of STN LFPs [5]	Correlates with bradykinesia and rigidity severity; suppressed by dopaminergic medication and DBS [4] [5]
Beta Bursting	Transient, short-duration episodes of elevated beta oscillation [4]	Time-frequency analysis to detect and characterize burst duration and amplitude [4]	Long-duration bursts (>400 ms) are specifically correlated with motor impairment [6]
Phase-Amplitude Coupling (PAC)	Coupling between the phase of beta oscillations and the amplitude of high-frequency oscillations (HFOs, 200-500 Hz) [5] [7]	Calculation of modulation index between low-frequency phase and high-frequency amplitude [5]	More specific to the motor STN than beta power alone; reflects medication "off" state and more affected side [5]
Inter-regional Synchrony	Synchronization of beta activity across different nodes of the BGTC network [4]	Coherence or phase-locking value between LFPs from different structures (e.g., STN and cortex) [4]	Reflects pathological hypersynchrony of the motor network; disrupted by effective therapy [4]

These biomarkers are not uniformly present across all patients. Beta power is not a universally optimal signal for aDBS, with some patients showing atypical oscillatory reactivity [5]. This heterogeneity underscores the need for a multi-faceted approach to biomarker identification, with beta-HFO PAC emerging as a promising and highly specific alternative [5] [7].

Table 2: Biomarker Characteristics and Clinical Utility

Characteristic	Spectral Beta Power	Beta-HFO Phase-Amplitude Coupling
Sensitivity to Dopaminergic Medication	Suppressed [5]	Cluster size significantly reduced in medication "on" state [5]
Specificity to Motor STN	Moderate [5]	High; more confined to motor STN contacts [5]
Lateralization with Symptoms	Not consistently significant across patient groups [5]	Stronger in the more affected STN [5]
Utility for DBS Contact Selection	Predictive in some cases [5]	Higher concordance with clinically effective contacts [5]

Experimental Protocols for Beta-Band Oscillation Analysis

Protocol 1: Recording and Analysis of Subthalamic LFPs in PD Patients

This protocol outlines the procedure for acquiring and analyzing local field potentials from the subthalamic nucleus in human Parkinson's disease patients, typically during a temporary externalization period following DBS lead implantation [5].

Materials and Reagents

• DBS Electrodes: Clinical-grade DBS leads (e.g., directional or conventional leads) implanted in the STN [5].
• Neural Signal Amplifier: A high-resolution, multichannel electrophysiology amplifier system.
• Data Acquisition System: Computer with analog-to-digital conversion capability, sampling rate ≥1000 Hz.
• Electrode Localization Software: Surgical planning software (e.g., Lead-DBS) for anatomical localization of recording contacts [5].
• Analysis Software: MATLAB with toolboxes (e.g., FieldTrip, Chronux) or custom scripts for spectral analysis.

Procedure

• Patient Preparation: Recordings are typically performed 2-7 days post-electrode implantation, before internal pulse generator placement. Patients should be in the practically defined "off" medication state (≥12 hours after last dopaminergic medication) [5].
• Signal Acquisition: Connect the externalized DBS leads to the amplifier system. Record bipolar LFPs from adjacent electrode contact pairs. Ensure a minimum sampling rate of 1000 Hz. Record during a resting, awake state for at least 5 minutes to ensure data stability [5].
• Pre-processing:
  • Apply a band-pass filter (e.g., 1-500 Hz).
  • Remove line noise (50/60 Hz) using a notch filter.
  • Visually inspect data and remove artifacts.
• Spectral Analysis:
  • Compute power spectral density using Welch's method or multitaper spectral analysis.
  • Identify the individual beta peak frequency for each patient within the 13-35 Hz range [5].
• Beta Burst Analysis:
  • Band-pass filter the signal around the individual beta peak frequency.
  • Apply the Hilbert transform to extract the amplitude envelope.
  • Define a burst threshold (typically based on a percentile of the amplitude distribution).
  • Detect bursts and quantify their duration and rate [6].
• Phase-Amplitude Coupling Analysis:
  • Filter the signal in the beta frequency band and extract the phase using the Hilbert transform.
  • Filter the signal in multiple high-frequency bands (e.g., 50-100 Hz, 100-200 Hz, 200-500 Hz) and extract their amplitude envelopes.
  • Compute the modulation index to quantify the coupling between beta phase and HFO amplitude [5].
• Anatomical Correlation: Reconstruct the electrode location using Lead-DBS software. Correlate biomarker magnitude with contact location relative to the motor STN [5].

Protocol 2: Computational Modeling of Closed-Loop DBS Control

This protocol describes a computational framework for developing and testing closed-loop DBS control algorithms in silico before preclinical testing [6].

Materials and Software

• Network Model: A biophysically based model of the cortico-basal ganglia-thalamic network, incorporating key nuclei (STN, GPe, GPi, cortex, thalamus) [6] [8].
• DBS Electric Field Model: A model simulating the extracellular electric field generated by DBS.
• LFP Simulation Module: A component that simulates LFP recordings from non-stimulating contacts on the DBS electrode.
• Control Algorithm Platform: Software environment (e.g., MATLAB/Simulink, Python) for implementing and testing control policies.

Procedure

• Model Configuration:
  • Implement the network model using conductance-based biophysical neuron models.
  • Set parameters to simulate the parkinsonian state (e.g., increased striatal inhibition to GPe, enhanced cortical drive to STN) [8].
  • Validate the model by confirming it generates elevated beta-band oscillations.

• DBS and LFP Simulation:

  • Incorporate a model of the DBS electric field and its effects on neural elements (axons, cell bodies).
  • Implement a forward model of LFP generation that captures synaptic activity in the STN.
  • Simulate the LFP recorded from bipolar electrode contacts [6].

• Control Algorithm Implementation:

  • On-Off Control: Implement a threshold-based controller that turns stimulation on when beta activity exceeds a set threshold and off when it falls below [6].
  • Dual-Threshold Control: Implement a controller that increases stimulation amplitude when beta is above an upper threshold, decreases it when below a lower threshold, and maintains it when within the target range [6].
  • Proportional-Integral (PI) Control: Implement a controller that adjusts stimulation parameters (amplitude or frequency) based on both the instantaneous error (proportional) and the accumulated error (integral) between measured beta and a target value [6].

• Performance Evaluation:

  • Simulate the closed-loop system and quantify beta suppression.
  • Compare power consumption to open-loop DBS.
  • Assess the controller's ability to track dynamic changes in beta activity.
  • Evaluate robustness to noise and model uncertainties.

Signaling Pathways and System Workflows

Pathological Beta Synchronization in the BGTC Network

The following diagram illustrates the mechanisms within the basal ganglia-thalamocortical network that lead to the generation and amplification of pathological beta oscillations in Parkinson's disease.

This diagram illustrates the critical shift in network dynamics that occurs with dopamine depletion. In the parkinsonian state, enhanced inhibition from the striatum to the GPe reduces GPe firing, which disinhibits the STN. Combined with cortical beta inputs, this creates a feedback loop where the STN and GPe become susceptible to synchronized bursting at beta frequencies [8]. The network exhibits resonance properties, selectively amplifying exogenous beta inputs, particularly when striatal inputs promote anti-phase firing between cortex and GPe [8].

Closed-Loop DBS Control Workflow

The following diagram outlines the complete workflow for an adaptive deep brain stimulation system that uses beta-band activity as a control signal.

This workflow demonstrates the core principle of aDBS: continuous real-time adjustment of stimulation parameters based on biomarker feedback. The proportional-integral (PI) controller has demonstrated superior performance in computational models, effectively suppressing pathological beta activity while reducing power consumption to approximately 25% of open-loop DBS [6]. This system creates a dynamic control loop that can adapt to fluctuating pathological states throughout the day.

Table 3: Key Research Reagents and Resources for Beta Oscillation Studies

Resource Category	Specific Examples	Research Application
Recording Electrodes	Directional DBS leads; Conventional DBS leads; Macro/microelectrodes [5]	Recording local field potentials from deep brain structures; Enabling spatially specific biomarker mapping [5]
Signal Acquisition Systems	Neural signal amplifiers (e.g., Blackrock Microsystems, TDT); Analog-to-digital converters [5]	High-fidelity recording of neural signals with appropriate sampling rates and minimal noise [5]
Computational Modeling Tools	Biophysically detailed network models (e.g., STN-GPe models); DBS electric field models [6] [8]	Testing control algorithms in silico; Investigating mechanisms of oscillation generation and suppression [6] [8]
Analysis Software/Toolboxes	Lead-DBS (electrode localization); FieldTrip, Chronux (signal analysis); Custom MATLAB/Python scripts [5]	Anatomical localization; Spectral analysis, burst detection, and PAC calculation [5]
Control Algorithm Platforms	MATLAB/Simulink; Python with control libraries; Real-time operating systems [6]	Implementing and testing closed-loop control policies (on-off, dual-threshold, PI) [6]

Beta-band oscillations have firmly established their role as a key biomarker for guiding therapy in Parkinson's disease, particularly in the development of closed-loop DBS systems. The experimental protocols and analytical frameworks presented here provide researchers with standardized methodologies for quantifying these pathological rhythms and leveraging them for therapeutic purposes. The future of this field lies in moving beyond single biomarkers toward multi-modal approaches that incorporate various neural signals and even kinematic data [2]. Furthermore, the integration of artificial intelligence and machine learning for neural decoding promises to enhance the precision of symptom estimation and facilitate more personalized adaptive stimulation paradigms [2]. As these technologies mature, biomarker-driven neuromodulation will continue to evolve, offering increasingly effective and efficient therapy for individuals with Parkinson's disease.

The evolution of Deep Brain Stimulation (DBS) for Parkinson's disease (PD) is marked by a pivotal shift from open-loop to closed-loop systems, known as adaptive DBS (aDBS). Traditional aDBS paradigms have predominantly relied on beta-band oscillations (13-35 Hz) from subcortical structures as feedback signals. While beta-driven aDBS provides significant therapeutic benefit, its limitations are increasingly apparent; it does not fully address the complex, multi-dimensional nature of PD symptoms, particularly axial motor deficits and non-motor symptoms [2].

Innovations in neurotechnology and data science now facilitate a more comprehensive approach. This Application Note details the expansion of the biomarker landscape beyond beta oscillations by integrating multi-modal inputs, including other neural signals and behavioral kinematics. This data-driven, precision neurology framework aims to bolster therapeutic efficacy and personalize neuromodulation therapy for PD [2] [9].

The Multi-Modal Biomarker Framework for aDBS

The rationale for moving beyond a single biomarker stems from the recognition of PD as a multi-system network disorder. Pathologies involving α-synuclein aggregation, mitochondrial dysfunction, and neuroinflammation lead to distributed alterations in neural circuit function [9] [10]. Capturing this complexity requires a multi-modal approach that interrogates the disease state from complementary angles.

The proposed framework integrates three primary data streams to inform aDBS control policies:

• Expanded Neural Signals: Moving beyond subcortical beta to include gamma oscillations, phase-amplitude coupling, and cortical signals from motor and associative regions.
• Kinematic & Behavioral Signatures: Using wearable sensors and algorithmic decoding to quantify motor states (e.g., bradykinesia, tremor, freezing of gait) in real-time.
• Neurophysiological Proxies: Employing non-invasive tools like Transcranial Magnetic Stimulation (TMS) to track disease progression through metrics like the cortical silent period (CSP), a marker of GABAergic inhibition that progressively lengthens with motor decline [2] [11].

The confluence of these data streams through machine learning and artificial intelligence enables the development of sophisticated symptom estimation and neural decoding models, creating a more holistic and responsive closed-loop system [2] [12].

Quantifying the Biomarker Landscape

The tables below summarize key biomarker categories and their correlation with clinical outcomes, providing a reference for experimental design.

Table 1: Electrophysiological & Neurophysiological Biomarkers for aDBS

Biomarker Category	Specific Metric	Measurement Technique	Association with PD State/Progression	Potential aDBS Application
Oscillatory Activity	Beta-band (13-35 Hz) power	Subthalamic Local Field Potentials (LFPs)	Correlates with rigidity and bradykinesia; suppressed by levodopa [2]	Standard aDBS control signal
	Gamma-band (>60 Hz) power	Cortical & Subcortical LFPs	Associated with movement initiation; potentially anti-akinetic [2]	Control signal for bradykinesia
Cross-Frequency Coupling	Beta-gamma Phase-Amplitude Coupling (PAC)	STN LFP / Cortical EEG	Exaggerated in PD; linked to motor impairment [2]	Composite control signal
Cortical Inhibition	Cortical Silent Period (CSP)	Transcranial Magnetic Stimulation (TMS)	Progressively lengthens over time; associated with motor decline on MDS-UPDRS [11]	Biomarker for disease progression & therapy tuning
Network Connectivity	Resting Motor Threshold (rMT)	Transcranial Magnetic Stimulation (TMS)	Deteriorates longitudinally; changes pronounced post-environmental stressors [11]	Tracks cortical excitability changes

Table 2: Kinematic & Digital Biomarkers for aDBS

Biomarker Category	Measured Parameter	Sensor Modality	Clinical Correlation	Potential aDBS Application
Gait & Posture	Stride Length, Velocity, Trunk Sway	Inertial Measurement Units (IMUs), Force Plates	Quantifies bradykinesia, postural instability; predicts fall risk [12] [13]	Control for axial symptoms
Tremor	Limb Acceleration (4-6 Hz)	Wrist/Ankle-worn IMUs	Direct measure of resting and action tremor severity [12]	On-demand tremor suppression
Bradykinesia	Movement Speed, Amplitude	Gyroscopes, Magnetometers	Correlates with MDS-UPDRS bradykinesia scores [12]	Control signal for akinesia
Freezing of Gait (FOG)	High-frequency leg movement prior to gait arrest	Shank-worn IMUs	Identifies FOG episodes with high sensitivity/specificity [13]	Triggers high-frequency stimulation to abort FOG

Experimental Protocols for Multi-Modal Biomarker Acquisition

Protocol 4.1: Ambulatory Kinematic Data Collection and Motor State Decoding

Objective: To capture high-fidelity movement data for developing machine learning models that decode motor symptoms in real-world environments.

Materials:

• Inertial Measurement Unit (IMU) sensors (e.g., containing tri-axial accelerometer, gyroscope, magnetometer).
• Data logger or Bluetooth-enabled device for real-time streaming.
• Secure mounting straps for limbs and trunk.
• Laptop/PC with Python/R and machine learning libraries (e.g., scikit-learn, TensorFlow/PyTorch).

Procedure:

• Sensor Placement: Affix IMUs bilaterally to the wrists (for tremor and bradykinesia), shanks (for gait and FOG), and the sternum (for postural control).
• Data Synchronization: Synchronize all sensors to a common time source. Perform a brief calibration sequence (e.g., repeated jumps).
• Concurrent Clinical Annotation: While the participant performs the MDS-UPDRS Part III motor examination or a structured protocol (e.g., timed up-and-go, gait), record the sensor data. Annotate the data stream with the onset/offset of each task and the clinician's severity scores.
• Free-Living Data Collection: Instruct the participant to wear the sensors during activities of daily living for a prescribed period (e.g., 6-8 hours), using a patient diary to log specific activities and symptom events.
• Data Preprocessing:
  • Apply signal filtering (e.g., high-pass >0.1 Hz to remove drift, band-pass for tremor-specific frequencies).
  • Segment data into epochs corresponding to annotated tasks or fixed-time windows.
  • Extract features from each epoch: time-domain (e.g., RMS, variance), frequency-domain (e.g., spectral power in relevant bands), and domain-specific features (e.g., FOG index).
• Model Training: Train a classifier (e.g., Support Vector Machine, Random Forest) or regression model (e.g., LASSO, Convolutional Neural Network) using the extracted features to predict clinical scores or classify motor states (e.g., "ON"/"OFF," "FOG"/"non-FOG") [12] [13].

Protocol 4.2: Intraoperative and Chronic Neural Signal Recording for Biomarker Discovery

Objective: To identify patient-specific neural biomarkers from implanted DBS leads and/or cortical electrodes for personalized aDBS control policies.

Materials:

• Implantable DBS system capable of chronic sensing (e.g., Medtronic Percept RC, Boston Scientific Vercise).
• Intraoperative neural recording system.
• Electroencephalography (EEG) caps with high-density arrays.
• Neural signal processing software (e.g., FieldTrip, EEGLAB, custom Python/MATLAB scripts).

Procedure:

• Intraoperative Mapping: During DBS lead implantation, record LFPs from the target region (e.g., STN, GPi) and simultaneously from scalp EEG. Have the patient perform simple motor tasks (e.g., hand gripping) to identify movement-related neural correlates.
• Chronic Ambulatory Sensing: Post-implantation, configure the sensing-enabled implantable pulse generator (IPG) to continuously or periodically stream LFP data from the implanted DBS leads.
• Symptom Correlations: Synchronize neural data streams with patient-reported outcomes (e.g., electronic diaries for pain, mood) and/or kinematic data from wearable sensors (Protocol 4.1).
• Biomarker Identification:
  • Preprocess neural signals: apply notch filters for line noise, band-pass filters for frequency bands of interest.
  • Compute features: spectral power, cross-frequency coupling, network coherence.
  • Use machine learning (e.g., LASSO regression, Linear Discriminant Analysis) to derive a model that predicts symptom severity from neural features. For example, train a model to distinguish high vs. low pain states or to predict tremor amplitude from LFP features [14].
• Control Policy Implementation: Translate the validated model into an embedded algorithm for the aDBS system. For instance, define thresholds for beta power or the output of a multi-feature classifier that triggers or titrates stimulation intensity [2] [14].

Visualization of Multi-Modal aDBS Workflow

The following diagram illustrates the integrated workflow for a multi-modal, closed-loop DBS system.

The Scientist's Toolkit: Research Reagent Solutions

Table 3: Essential Materials and Tools for Multi-Modal aDBS Research

Item/Category	Example Product/Specification	Function in Research
Sensing-Capable DBS System	Medtronic Percept RC, Boston Scientific Vercise Gevia	Chronic recording of local field potentials (LFPs) from implanted DBS leads; enables correlation of neural activity with symptoms and therapy.
Wireless IMU Sensors	APDM Opal, DyCare FreeSense, Shimmer3	Ambulatory, high-frequency recording of kinematic data (acceleration, angular velocity) for quantifying tremor, bradykinesia, and gait.
TMS System	MagVenture MagPro, Deymed DuoMag	Non-invasive assessment of cortical excitability (rMT) and GABAergic inhibition (CSP) to track disease progression and cortical involvement [11].
Neural Signal Processing Toolbox	FieldTrip (MATLAB), MNE-Python, EEGLAB	Open-source software for preprocessing, feature extraction (e.g., spectral power), and analysis of LFP and EEG data.
Machine Learning Library	Scikit-learn (Python), TensorFlow/PyTorch	Provides algorithms for developing classification and regression models to decode symptom states from multi-modal data [12] [14].
Digital Biomarker Platforms	Kinesia ONE (Great Lakes Neurotech), cloud-based analysis portals	Integrated solutions for collecting and analyzing sensor-based motor data, often with validated algorithms for specific symptoms like tremor.

Adaptive Deep Brain Stimulation (aDBS) represents a significant evolution in the treatment of Parkinson's disease (PD), moving beyond static continuous DBS (cDBS) to a closed-loop system that dynamically adjusts stimulation parameters based on real-time physiological feedback [2]. Parkinson's disease is characterized by dysfunctional neural dynamics within the basal ganglia-thalamocortical circuit, particularly pathological oscillatory activity in the beta frequency band (12-30 Hz) that correlates with core motor symptoms such as bradykinesia and rigidity [15] [16]. The fundamental innovation of aDBS lies in its ability to detect these circuit-level abnormalities and deliver precisely-timed neuromodulation to normalize pathological network dynamics, thereby offering more personalized therapy with potentially fewer side effects than conventional approaches [2] [1]. This application note details the experimental frameworks and methodologies for investigating how aDBS modulates these pathological network dynamics at the circuit level, providing researchers with practical tools for advancing closed-loop neuromodulation therapies.

Pathological Network Dynamics in Parkinson's Disease

In the parkinsonian state, the cortex-basal ganglia-thalamocortical circuit exhibits characteristic aberrant activity patterns that can be quantified through various neurophysiological metrics. Understanding these dynamics is crucial for developing effective aDBS strategies.

Key Circuit-Level Abnormalities

• Excessive Beta Synchronization: A hallmark of PD pathophysiology is exaggerated beta-band (12-30 Hz) oscillatory activity, particularly in the subthalamic nucleus (STN) [16]. This pathological synchrony correlates with motor symptom severity and can be suppressed by both dopaminergic medication and DBS [17] [16].

• Altered Brain Dynamics: EEG microstate analysis reveals that PD slows brain dynamic changes on a sub-second timescale, evidenced by longer mean microstate duration with reduced occurrence per second compared to healthy controls [17]. These parameters significantly correlate with movement disorders assessed by UPDRS-III [17].

• Pathological Connectivity Patterns: The triple network model (default mode, salience, and executive/frontoparietal networks) exhibits abnormal topology in PD, with impaired functional connectivity in cortical-striatal loops and related neural circuits [17] [16].

Table 1: Key Neurophysiological Biomarkers of Parkinsonian Network Dynamics

Biomarker	Measurement Technique	Circuit Manifestation	Clinical Correlation
STN Beta Power (12-30 Hz)	Local Field Potential (LFP) recording	Exaggerated oscillatory power in STN	Correlates with bradykinesia and rigidity [16]
Cortex-STN Coherence	Imaginary coherency between ECoG and STN-LFP	Enhanced beta-band synchronization between cortex and STN	Associated with motor impairment [16]
EEG Microstate Duration	High-density EEG (256-channel)	Slowed brain dynamic changes (80-120 ms typically)	Correlates with UPDRS-III scores [17]
Hyperdirect Pathway Activity	Granger causality, bispectral time delay analysis	Increased information flow from cortex to STN	Disrupts normal motor control [16]

Core Mechanisms of aDBS Action

aDBS operates through multiple complementary mechanisms to normalize pathological network dynamics. The system typically utilizes subthalamic beta power as a control signal, with stimulation amplitude automatically adjusted according to predetermined thresholds [1].

Biomarker-Guided Stimulation

The foundational mechanism of aDBS involves continuous monitoring of pathological biomarkers, primarily subthalamic beta power, with real-time adjustment of stimulation parameters. In commercially available Dual Threshold aDBS systems, stimulation amplitude modulates between upper and lower limits based on whether beta power exceeds the 75th percentile or falls below the 25th percentile of daytime beta power [1]. This approach allows the system to respond to dynamic changes in pathological activity throughout the day.

Pathway-Specific Network Modulation

At the circuit level, aDBS mimics key mechanisms of dopaminergic therapy by suppressing excessive interregional network synchrony associated with indirect and hyperdirect cortex-basal ganglia pathways [16]. Research combining multisite intracranial recordings with normative MRI-based whole-brain connectomics demonstrates that both dopamine and DBS reduce pathological oscillatory coupling between cortex and STN, though they exert distinct mesoscale effects on local neural population activity [16].

Diagram 1: Circuit mechanisms of aDBS. The diagram contrasts normal basal ganglia-thalamocortical circuitry with the parkinsonian state characterized by increased beta synchrony and shows how aDBS detects and normalizes these pathological dynamics.

Experimental Protocols for Investigating aDBS Mechanisms

In Vivo Human Electrophysiology Protocol

This protocol enables direct investigation of aDBS effects on cortex-basal ganglia circuits in human participants, combining invasive neural recordings with stimulation parameter modulation.

Table 2: Key Research Reagents and Equipment

Item	Specification	Experimental Function
DBS System with Sensing Capability	Medtronic Percept RC16, Boston Scientific Cartesia X	Simultaneous neural recording and stimulation delivery [1] [16]
Externalized DBS Leads	4-16 contact electrodes (e.g., 1.5 mm spacing)	Acute postoperative intracranial EEG and LFP recording [16]
Electrocorticography (ECoG) Arrays	Subdural grid/strip electrodes	Cortical population activity recording [16]
Beta Band Analysis Software	Custom MATLAB/Python scripts with multitaper spectral analysis	Quantification of 12-30 Hz oscillatory power [16]
Connectivity Analysis Tools	Imaginary coherency, Granger causality algorithms	Measuring directed and undirected cortex-STN coupling [16]

Procedure:

• Participant Preparation: Recruit PD patients undergoing bilateral STN-DBS implantation. Record resting-state neural activity through externalized leads 2-5 days postoperatively [16].

• Multimodal Recording Setup:

  • Acquire STN local field potentials (LFP) from DBS electrode contacts
  • Record cortical activity using electrocorticography (ECoG) targeted at sensorimotor cortex
  • Maintain electrode impedance <30 kΩ throughout recording session [17]

• Experimental Conditions:

  • OFF Therapy: Withdraw dopaminergic medication for ≥12 hours before recording
  • ON Levodopa: Administer standard levodopa dose following OFF state recording
  • ON STN-DBS: Apply therapeutic high-frequency stimulation (typically 130-180 Hz) [16]

• Signal Processing:

  • Apply 1-40 Hz band-pass filter to remove DBS artifact (130 Hz) [17]
  • Perform independent component analysis to remove ballistocardiogram, myographic, and oculomotor artifacts
  • Compute spectral power using multitaper Fourier analysis
  • Calculate connectivity metrics: imaginary coherency (undirected) and Granger causality (directed) [16]

• Data Analysis:

  • Compare beta power (12-30 Hz) across conditions in cortex and STN
  • Assess cortex-STN coherence and directionality of information flow
  • Correlate neurophysiological changes with clinical symptom severity (UPDRS-III)

Chronic Ambulatory aDBS Programming Protocol

This protocol outlines the clinical programming approach for investigational chronic aDBS systems, enabling evaluation of long-term therapeutic efficacy and biomarker stability.

Diagram 2: aDBS programming workflow. The three-step clinical protocol for implementing chronic adaptive deep brain stimulation shows key decision points and parameter selection considerations.

Procedure:

• Preparation Phase: Biomarker Identification

  • Record STN-LFP in OFF medication state to maximize beta peak visibility
  • Identify sensing-compatible contacts with optimal signal-to-noise ratio
  • Select responsive beta peaks through continuous test stimulation and medication-induced beta power modulation [1]

• Initial Setup: Parameter Configuration

  • Collect continuous Timeline data over several days to establish baseline beta dynamics
  • Set LFP thresholds to 25th (lower) and 75th (upper) percentiles of daytime beta power
  • Define stimulation amplitude limits based on therapeutic window established during cDBS
  • Typical parameters: Upper limit 2.28 mA (mean), Lower limit 1.71 mA (mean), Range 0.58 ± 0.19 mA [1]

• Optimization Phase: Parameter Refinement

  • Monitor for inadequate adaptation (stimulation stuck at limits)
  • Adjust LFP thresholds if stimulation fails to track beta dynamics
  • Refine amplitude limits to address persistent hypokinetic or hyperkinetic symptoms
  • Resolve artifact-related maladaptation through sensing parameter adjustment [1]

• Ambulatory Assessment

  • Deploy ecological momentary assessments (EMA) for home-based symptom tracking
  • Collect data over 2-week periods for both cDBS and aDBS conditions
  • Monitor stimulation adaptation patterns through device Timeline data [1]

Computational Modeling Protocol for Controller Design

This in silico protocol enables development and testing of novel control algorithms for aDBS using biophysically-based network models before human implementation.

Procedure:

• Model Configuration:

  • Implement cortex-basal ganglia-thalamus network model with Hodgkin-Huxley type neurons [18]
  • Include representations of cortex (CTX), striatum (STR), STN, globus pallidus (GPe/GPi), and thalamus (TH)
  • Configure 10 single-compartment model neurons per nucleus with appropriate synaptic connectivity [18]

• Parkinsonian State Simulation:

  • Induce exaggerated beta oscillations (12-30 Hz) through altered synaptic weights and dopaminergic depletion
  • Quantify beta band power in GPi LFP signals as feedback control signal [18]

• Controller Implementation:

  • Develop Radial Basis Function (RBF) network-driven supervisory control algorithm
  • Train inverse model of the plant to automatically adjust stimulation parameters
  • Compare with traditional proportional-integral-derivative (PID) controllers [18]

• Performance Validation:

  • Test controller ability to track target beta power during dynamic changes in parkinsonian state
  • Evaluate stimulation efficacy in suppressing pathological oscillations while minimizing energy delivery
  • Assess robustness to noise and parameter variations [18]

Quantitative Outcomes and Clinical Correlations

Research findings demonstrate significant physiological and clinical effects of aDBS compared to conventional continuous stimulation.

Table 3: Quantitative Outcomes of aDBS in Parkinson's Disease

Parameter	cDBS Performance	aDBS Performance	Measurement Context
Overall Well-being (EMA scale 1-10)	5.92 ± 1.01	6.73 ± 1.33*	Home ecological momentary assessment [1]
General Movement (EMA scale 1-10)	5.47 ± 1.22	6.20 ± 1.44†	Home ecological momentary assessment [1]
Stimulation Amplitude Range	Fixed: ~2.04 mA	Adaptive: 0.58 ± 0.19 mA range	Clinical programming optimization [1]
Cortical High Beta Power (20-30 Hz)	-	Significant suppression with dopamine; No change with DBS	Resting-state ECoG recording [16]
STN Low Beta Power (12-20 Hz)	-	Significant suppression with dopamine; Variable with DBS	Resting-state LFP recording [16]
EEG Microstate Duration	Increased in PD vs. HC	Partial normalization toward healthy levels	256-channel EEG analysis [17]

*p = 0.007, †p = 0.058 (trend)

Advanced Methodological Considerations

Multi-Modal Biomarker Integration

Future aDBS systems are evolving beyond sole reliance on beta oscillations to incorporate multiple physiological and behavioral signals. Artifical intelligence approaches can decode Parkinson's motor symptoms from neural signals by integrating subcortical beta oscillations with other neural and kinematic signals [2]. This multi-modal approach may enhance therapeutic efficacy for symptoms not adequately addressed by beta-driven aDBS alone.

Phase-Locked Stimulation Strategies

Emerging research indicates that stimulation phase relative to ongoing oscillations significantly impacts therapeutic efficacy. In Alzheimer's disease models, phase-locked fornix-DBS can bidirectionally modulate hippocampal theta power, with effects significantly larger than open-loop stimulation [19]. Similar phase-dependent approaches may enhance aDBS for Parkinson's disease by targeting specific phases of pathological oscillations.

Dynamic Network Mapping

Combining fully invasive neural multisite recordings with normative MRI-based whole-brain connectomics enables precise mapping of pathway-specific network mechanisms [16]. This approach reveals that dopamine and DBS share suppression of excessive interregional synchrony in indirect and hyperdirect pathways, despite distinct mesoscale effects on local population activity.

Implementing aDBS: From Surgical Targeting to Clinical Programming

Application Note: Core Principles for Patient Selection

Optimal patient selection is the cornerstone of successful Deep Brain Stimulation (DBS) therapy for Parkinson's disease (PD). A comprehensive interdisciplinary evaluation is the critical first step to assessing risks, benefits, and establishing appropriate, patient-centered goals [20]. The selection process must balance disease-related characteristics with individual patient demographics and social factors to ensure positive long-term outcomes.

Key Clinical and Demographic Selection Factors

Levodopa Responsiveness: A positive response to a levodopa challenge test is a well-established, unified predictor of positive motor outcomes after DBS across all disease durations [21]. The response indicates that a patient's primary motor symptoms (bradykinesia, rigidity, tremor) are amenable to dopaminergic modulation, and by extension, to neuromodulation. It remains the most reliable positive prognostic factor for motor improvement.

Disease Duration and Symptom Profile: The "5-2-1 criteria" have been proposed as a specific marker for advanced PD, prompting a discussion about device-aided therapies like DBS. This criteria is met when a patient requires ≥5 oral levodopa doses per day, yet still experiences ≥2 hours of "off" symptoms per day and ≥1 hour of troublesome dyskinesia per day [20]. DBS should be considered for individuals with uncontrollable tremors despite adequate medication trials, significant motor fluctuations on dopaminergic therapies, or bothersome levodopa-induced dyskinesias [20].

Biological vs. Chronological Age: While many DBS centers consider patients under 70-75 years of age, biological factors are more critical than chronological age. Comorbidities, cognitive status, brain atrophy, and the burden of levodopa-resistant symptoms are the primary considerations. Studies have shown that older patients can still derive significant benefit from DBS without a substantially higher 90-day complication risk [20].

Neuropsychiatric and Cognitive Status: A thorough evaluation by a psychiatrist and neuropsychologist is mandatory. Significant untreated depression, anxiety, or cognitive decline are relative contraindications for DBS, as they can be exacerbated by stimulation and impair the patient's ability to participate in post-operative care and programming [20] [22].

Social Support Structure: The presence of adequate and reliable social support is essential for a successful outcome. Care partners help manage the anxiety of the procedure, ensure timely attendance at appointments, recognize and report side effects, and support recovery. Integrating a social worker into the multidisciplinary team can strengthen care and navigation through the surgical process [20].

Table 1: Key Factors in DBS Patient Selection and Their Clinical Implications

Factor	Clinical Implication	Evidence/Notes
Levodopa Response	Unified positive predictor of motor outcome.	Consistent factor across short, mid, and long disease durations [21].
5-2-1 Criteria	Triggers discussion for device-aided therapy.	≥5 levodopa doses/day, ≥2h "off" time/day, ≥1h dyskinesia/day [20].
Biological Age	More important than chronological age.	Considers comorbidities, cognitive decline, brain atrophy [20].
Axial Symptoms	Poor response to DBS.	Gait freezing, postural instability, and dysarthria often show limited improvement.
Cognitive Status	Critical exclusion criterion.	Significant dementia or cognitive impairment is a contraindication [22].

Addressing Disparities and Optimizing Counseling

Evidence shows significant disparities in DBS access. Women are significantly underrepresented in DBS referrals, comprising less than 25% of those evaluated, despite reporting greater psychological distress and self-perceived disability [20]. Similarly, Black and Hispanic patients are less likely to undergo DBS than White patients [20]. These findings highlight the need for enhanced, culturally sensitive preoperative counseling. Underrepresented patients may benefit from more time devoted to discussing risks, benefits, and postoperative expectations, as well as comprehensive education on DBS technology [20].

Application Note: Quantifying the Impact of Surgical Timing

The timing of DBS surgery relative to disease duration is a key determinant of therapeutic outcome. Evidence from a large-scale multicenter cohort study (n=1,717) provides quantitative insights into the optimal surgical window and its impact on motor, neuropsychological, and quality-of-life metrics [21].

Multicenter Cohort Data on Disease Duration and Outcomes

A 2025 analysis of patients who underwent subthalamic DBS classified them into three groups based on pre-surgical PD duration: short (<5 years), mid (5–10 years), and long (≥10 years). All groups showed significant improvements in motor function, mood, and quality of life after two years. However, the magnitude of benefit was not uniform [21].

Patients in the mid-duration group (5-10 years) experienced the most substantial and comprehensive benefits. They achieved the greatest improvements in motor function, significantly outperforming both the short- and long-duration groups. Furthermore, they showed superior outcomes in neuropsychological evaluations (anxiety and depression) and quality of life compared to the long-duration group [21]. This suggests a potential "optimal window" for DBS intervention.

Patients with short disease duration (<5 years) demonstrated significant improvements, but their motor outcomes were less favorable than the mid-duration group. A critical finding was that a higher baseline motor severity (MDS-UPDRS-III off-medication) was a negative prognostic factor for this group, which contrasts with its positive predictive value in longer-duration patients. This may indicate a more aggressive or malignant disease subtype in these rapidly progressing patients, warranting caution in selection [21].

The long-duration group (≥10 years) benefited from DBS, but to a lesser extent than the mid-duration group, particularly for non-motor symptoms and quality of life [21].

Table 2: Two-Year DBS Outcomes by Pre-Surgical Disease Duration (Multicenter Cohort, n=1,717)

Outcome Measure	Short Duration (<5 yrs)	Mid-Duration (5-10 yrs)	Long Duration (≥10 yrs)
MDS-UPDRS-III (Off-Med)	Significant Improvement	Greatest Improvement (vs. short +8.0%; vs. long +5.6%)	Significant Improvement
HAM-A (Anxiety)	Significant Improvement	Significant Improvement	Significant Improvement
HAM-D (Depression)	Significant Improvement	Greatest Improvement (vs. long +19.1%)	Significant Improvement
PDQ-39 (Quality of Life)	Significant Improvement	Greatest Improvement (vs. long +7.6%)	Significant Improvement
Key Prognostic Factor	Higher baseline score = Negative outcome	Higher baseline score = Positive outcome	Higher baseline score = Positive outcome

Experimental Protocols

Protocol 1: The Interdisciplinary DBS Candidate Assessment

This protocol outlines the standardized pre-operative evaluation for potential DBS candidates, as derived from established clinical practice [20].

Objective: To comprehensively evaluate a patient's suitability for DBS surgery through a multidisciplinary assessment of neurological, neuropsychiatric, and social factors, ensuring an optimal risk-benefit ratio.

Materials:

• Unified Parkinson's Disease Rating Scale (UPDRS) or MDS-UPDRS
• Levodopa challenge test equipment
• Neuropsychiatric battery (e.g., MoCA, BDI)
• Brain MRI
• Social worker assessment form

Procedure:

• Neurological Examination:
  • Perform a detailed MDS-UPDRS assessment in the practical "off" state (≥12 hours after last medication) and again in the "on" state after a supervised levodopa challenge.
  • Calculate the percentage improvement in motor scores to quantify levodopa responsiveness.
  • Document the presence and severity of motor fluctuations, dyskinesias, and drug-resistant tremor.

• Neuroimaging:

  • Acquire a high-resolution structural MRI to rule out contraindications (e.g., significant brain atrophy, vascular disease, atypical parkinsonism) and to aid in surgical planning.

• Neuropsychiatric Evaluation:

  • Conduct a formal assessment by a psychiatrist or neuropsychologist.
  • Screen for significant depression, anxiety, apathy, and impulse control disorders.
  • Administer cognitive tests to rule out dementia (e.g., MoCA). Mild cognitive impairment may not be an absolute contraindication but requires careful counseling.

• Interdisciplinary Team Conference:

  • The neurologist, neurosurgeon, psychiatrist, and other team members (e.g., nurse coordinator, social worker) review all collected data.
  • A collective decision is made on patient candidacy, weighing benefits against risks.
  • The optimal surgical target (STN vs. GPi) is selected based on patient-specific symptom priorities.

• Patient and Caregiver Education:

  • The team discusses the procedure, realistic goals, potential risks, and the post-operative management process with the patient and their care partner.
  • Ensure that expectations are aligned with probable outcomes.

Protocol 2: Electrophysiology-Guided Contact Selection for DBS Programming

This protocol details a machine learning approach to optimize DBS programming by predicting the therapeutic window of electrode contacts using electrophysiological signatures [23].

Objective: To accelerate the post-operative DBS programming process by using a data-driven model to identify the electrode contact with the largest therapeutic window, based on local field potentials (LFPs) and magnetoencephalography (MEG).

Materials:

• Sensing-capable DBS system (e.g., Medtronic Percept PC)
• MEG system
• Software for signal processing (e.g., MATLAB, Python)
• Pre-trained extreme gradient boosting (XGBoost) model

Procedure:

• Data Acquisition:
  • Post-operatively, record resting-state neural signals. This involves simultaneous LFP recordings from the implanted DBS leads and MEG recordings from the cortex.
  • Ensure the patient is in the off-medication state (e.g., overnight withdrawal) for a standardized baseline.
  • Note on Timing: Recordings should be performed after the stabilization of the post-operative "microlesion effect," which typically occurs 22-29 days after implantation, as electrophysiological signals (beta power, complexity) are unstable before this period [24].

• Feature Extraction:

  • For each electrode contact, calculate power spectral density from the LFPs to extract power features in key frequency bands: θ (4-8 Hz), α (8-12 Hz), β (13-30 Hz), γ (>35 Hz), and high-frequency oscillations (HFOs).
  • Compute STN-cortex coherence between the LFP signals and the MEG cortical signals across the same frequency bands for multiple brain regions.

• Model Prediction:

  • Input the extracted electrophysiological features (STN power and STN-cortex coherence) into the pre-trained XGBoost model.
  • The model will output a predicted "normalized therapeutic window" for each electrode contact.

• Contact Selection:

  • Rank all available contacts based on the model's predicted therapeutic window.
  • Initiate the clinical monopolar review starting with the contact predicted to have the largest therapeutic window, thereby reducing the time to find the optimal settings.

• Validation:

  • Clinically validate the selected contact through standard testing, assessing symptom control and the threshold for side effects.

Signaling Pathways and Workflows

The Scientist's Toolkit: Research Reagent Solutions

Table 3: Essential Tools for Advanced DBS Research

Tool / Technology	Function in Research	Example Use-Case
Sensing-Capable DBS Implant	Chronic recording of local field potentials (LFPs) from the implanted target.	Capturing β-oscillations from the STN for biomarker discovery and adaptive control [25] [23].
Adaptive DBS Algorithm	Closed-loop software that adjusts stimulation based on a neural biomarker.	Real-time modulation of stimulation amplitude in response to LFP β-power fluctuations [25] [26].
Machine Learning Model (XGBoost)	Multivariate prediction of clinical outcomes from electrophysiological data.	Accelerating DBS programming by predicting the therapeutic window of electrode contacts [23].
Magnetoencephalography (MEG)	Non-invasive recording of cortical activity with high temporal resolution.	Investigating STN-cortex coherence as a feature for predicting optimal stimulation parameters [23].
Structured Clinical Database	Standardized collection of motor, neuropsychological, and quality-of-life outcomes.	Conducting large-scale, real-world cohort studies to determine optimal surgical timing [21].

Electrode Technology and Sensing Capabilities for Real-Time Signal Acquisition

Closed-loop adaptive deep brain stimulation (aDBS) represents a paradigm shift in the management of Parkinson's disease (PD), moving beyond conventional open-loop stimulation toward responsive, personalized neuromodulation. The efficacy of these intelligent systems hinges on their core component: advanced electrode technology capable of real-time signal acquisition. These sensing systems detect and interpret specific neurological biomarkers to dynamically adjust stimulation parameters, creating a feedback-controlled therapeutic system [27] [6].

The most established biomarker for aDBS in PD is pathological beta-band oscillatory activity (13-30 Hz) within the cortico-basal ganglia network. Exaggerated beta activity correlates strongly with motor impairment, and its suppression through medication or DBS associates with motor improvement [6]. This oscillatory activity manifests not as a continuous signal but as fluctuating bursts, with longer durations (>400 ms) particularly linked to motor symptoms [6]. Beyond local field potentials (LFPs), research explores complementary biosignals including muscle activity via surface electromyography (sEMG) and neurochemical concentrations, enabling increasingly comprehensive patient state monitoring [28] [29].

Electrode Technologies and Signal Acquisition Modalities

Neural Electrodes for Local Field Potential Recording

DBS electrodes equipped with sensing capabilities can record LFPs from the subthalamic nucleus (STN) or other target structures. These recordings are typically obtained from the same macroelectrodes used for stimulation or from dedicated contacts, enabling detection of oscillatory neural activity that serves as the control signal for aDBS algorithms [27] [6].

Table 1: Quantitative Performance of Featured Sensing Technologies

Sensing Technology	Target Signal	Sensitivity	Detection Range	Response/Recovery Time	Key Performance Metrics
LFP Recording Electrodes [27] [6]	Beta-band oscillations (13-30 Hz)	N/A	N/A	Rapid response to beta fluctuations	Correlation with motor symptoms (UPDRS); Beta burst duration (>400 ms) linked to impairment
sEMG Electrodes [29]	Muscle activation patterns	N/A	N/A	High temporal resolution	Accurately predicts UPDRS scores; Differentiates PD subtypes (tremor vs. rigidity-dominant)
Enzymatic L-dopa Biosensor [28]	L-dopa in sweat	0.0649 μA/μM·cm²	8 nM (LOD) - therapeutic range	Real-time continuous monitoring	High selectivity against interferents (uric acid); >93% cell viability; <2% hemolysis rate
Zinc-Nickel Pressure Sensor [30]	Physical pressure (motor symptoms)	0.32 V kPa⁻¹ (3-6 kPa)	0–10 kPa	95/80 ms (response/recovery)	Durability: >4500 cycles; Power: 1.2 V cm⁻², 150 mW; 95% PD severity classification accuracy

Complementary Sensing Modalities

• Surface Electromyography (sEMG): sEMG provides a non-invasive readout of muscle activation patterns through electrodes placed on the skin. It is particularly valuable for quantifying core PD motor symptoms like tremor, bradykinesia, and rigidity. When combined with inertial measurement units (IMUs) that capture motion kinematics, sEMG enables robust multimodal assessment of motor states. Advanced feature extraction from sEMG signals (time-frequency descriptors, Mini-ROCKET features) allows for objective UPDRS-compatible scoring of motor symptoms, facilitating continuous monitoring outside clinical settings [29].

• Electrochemical Biosensors: For therapeutic drug monitoring, wearable electrochemical biosensors can detect L-dopa concentrations in biofluids like sweat. These devices typically use flexible substrates and enzymatic elements (e.g., tyrosinase) or non-enzymatic composites for specific L-dopa recognition. Recent innovations integrate the biosensing platform with flexible micro-supercapacitors as power sources, creating fully integrated, self-powered systems for personalized medicine [28].

• Self-Powered Physical Sensors: Novel sensor designs can harvest energy from their operation, enhancing longevity. One example is a zinc-nickel electrochemical-based pressure sensor that generates a voltage signal in response to physical pressure changes. Such devices can detect motor symptoms like tremors or gait disturbances, with high biocompatibility and stable power output, making them suitable for wearable form factors [30].

Experimental Protocols for System Validation

Protocol 1: In Vivo Validation of aDBS Using LFP Biomarkers

Objective: To evaluate the efficacy of a closed-loop DBS system that modulates stimulation parameters based on real-time LFP beta power in a parkinsonian model [27] [31].

• Animal Preparation: Utilize hemiparkinsonian rat models (e.g., 6-OHDA lesioned) or non-human primate models (e.g., MPTP treated). Confirm successful model creation using apomorphine-induced rotation tests and post-mortem tyrosine hydroxylase immunocytochemistry.
• Electrode Implantation: Surgically implant a DBS lead with recording capabilities in the STN. Histologically confirm final electrode location.
• Signal Acquisition & Processing: In freely behaving animals, continuously record LFPs from the STN.
  • Filtering: Bandpass filter the raw LFP signal in the beta frequency band (13-30 Hz in humans; adjust for species).
  • Rectification and Averaging: Rectify the filtered signal and compute its average rectified value (ARV) over a short window (e.g., 200-500 ms) to create a control signal proportional to beta power.
• Closed-Loop Control: Feed the beta ARV into the chosen control algorithm.
  • On-Off Control: DBS is triggered ON when beta power exceeds a set threshold and OFF when it falls below it [31].
  • Proportional (P) Control: The amplitude or frequency of DBS is modulated linearly in proportion to the measured beta power [6] [31].
  • Proportional-Integral (PI) Control: The stimulation parameter is adjusted based on both the instantaneous value (proportional) and the accumulated history (integral) of the beta power error signal, offering superior regulation [6].
• Behavioral Assessment: Quantify motor function during aDBS, conventional continuous DBS, and no stimulation using standardized tests (e.g., cylinder test for forelimb use, stepping test).
• Data Analysis: Compare behavioral outcomes, stimulation energy consumption, and the degree of beta power suppression between the different stimulation paradigms.

Protocol 2: Multimodal sEMG-IMU for Objective Motor Symptom Quantification

Objective: To objectively quantify PD motor symptoms and UPDRS scores using a synchronized sEMG and IMU system [29].

• Participant Setup: Adhere to ethical approval and informed consent. Place sEMG electrodes on muscles relevant to tested tasks (e.g., forearm flexors/extensors for upper limb tasks; tibialis anterior for gait). Securely attach IMU sensors to corresponding body segments.
• Standardized Task Protocol: Participants perform a series of standardized motor tasks:
  • Seated Rest: Forearm relaxed on an armrest to assess rest tremor.
  • Upper-Limb Tasks: Include forearm pronation-supination, finger-to-nose movement, fist clench, and thumb-index finger pinch to assess bradykinesia.
  • Gait: Walking a short distance to assess gait impairment and potential freezing of gait.
• Data Synchronization & Preprocessing: Record sEMG and IMU data with synchronized timestamps. Filter sEMG signals (e.g., bandpass 20-450 Hz) to remove noise and movement artifacts.
• Feature Extraction:
  • Handcrafted Features: Calculate time-domain (e.g., RMS, variance), frequency-domain (e.g., median frequency), and time-frequency domain features from sEMG.
  • Representation Learning: Extract high-level features using models like Mini-ROCKET or InceptionTime.
  • IMU Features: Extract kinematic features from accelerometer and gyroscope data.
• Model Training & Scoring: Train a machine learning model (e.g., LDA-SV - Linear Discriminant Analysis with Soft Voting) on the extracted features to predict UPDRS item scores (0-3) for each motor task. Aggregate segment-level probabilities to assign final item-level ratings.
• Validation: Compare algorithm-generated scores against blinded clinician UPDRS ratings using intraclass correlation coefficient (ICC) and weighted kappa statistics.

Diagram 1: The core closed-loop feedback mechanism for aDBS, where recorded beta oscillations directly control stimulation parameters.

Diagram 2: Experimental workflow for objective quantification of PD motor symptoms using synchronized sEMG and IMU data.

The Scientist's Toolkit: Essential Research Reagents and Materials

Table 2: Key Reagents and Materials for aDBS Research

Item Name	Function/Application	Specific Example / Rationale
DBS Macroelectrode	Implanted device for stimulation and LFP recording.	Clinical-grade electrodes with multiple contacts (e.g., 4-8) to allow simultaneous recording from one pair and stimulation from another [27] [6].
sEMG Electrodes	Non-invasive detection of muscle electrical activity.	Disposable Ag/AgCl surface electrodes; Wireless sEMG bands for unrestricted movement during behavioral tests [29].
Inertial Measurement Unit (IMU)	Captures kinematic data (acceleration, rotation).	Wearable IMU sensors containing tri-axial accelerometers and gyroscopes, synchronized with sEMG for multimodal analysis [29].
Enzymatic Biosensor	Selective detection of L-dopa in biofluids.	Flexible electrode modified with tyrosinase enzyme or Au nanorods/PEDOT:PSS composite for non-enzymatic detection in sweat [28].
Flexible Micro-Supercapacitor	Power source for wearable sensors.	NiCo LDH//TRGO-based supercapacitors offering high specific capacitance and flexibility, enabling self-powered operation [28].
Computational Model (in silico)	Testing and tuning aDBS algorithms preclinically.	Biophysically grounded network models of cortico-basal ganglia-thalamic loop that simulate LFP beta dynamics and DBS effects [6].
Hemiparkinsonian Animal Model	Preclinical in vivo testing of aDBS systems.	6-OHDA lesioned rats or MPTP-treated non-human primates to model PD pathology and test efficacy and safety [31].

The advancement of closed-loop DBS for Parkinson's disease is intrinsically linked to progress in electrode technology and real-time sensing capabilities. The integration of direct neural signal acquisition (LFP) with complementary modalities like sEMG and biochemical sensing provides a multi-faceted view of the patient's disease state. The experimental protocols and tools detailed herein provide a framework for developing and validating these sophisticated systems. As these technologies mature, they pave the way for smarter, more personalized neuromodulation therapies that dynamically adapt to patient needs, promising superior symptom control with reduced side effects.

A Stepwise Clinical Programming Protocol for aDBS Activation

Adaptive Deep Brain Stimulation (aDBS) represents a significant advancement in the treatment of Parkinson's disease (PD), moving beyond traditional open-loop systems to a responsive, closed-loop approach. This protocol outlines a standardized procedure for activating and programming aDBS systems, framed within the context of developing more personalized and efficient neuromodulation therapies. Deep Brain Stimulation of the subthalamic nucleus (STN) has proven to be an effective long-term therapy for moderate to advanced Parkinson's disease, with studies demonstrating sustained improvement in motor function and activities of daily living over five years, alongside stable reduction of anti-parkinsonian medication [32] [33].

The fundamental mechanism of aDBS builds upon our understanding of conventional DBS, which appears to function through differential synaptic depression—specifically, a greater decrease in glutamate release compared to GABA at the stimulation site, effectively inhibiting the hyperactive STN neurons that characterize Parkinson's pathology [34]. The closed-loop aDBS system improves upon this mechanism by continuously monitoring neural biomarkers of symptom severity and adjusting stimulation parameters in real-time, potentially enhancing therapeutic efficacy while reducing side effects and energy consumption compared to continuous stimulation paradigms.

Quantitative Outcomes of Conventional DBS

Table 1: Five-Year Outcomes of STN-DBS for Parkinson's Disease (N=137) [32]

Assessment Measure	Baseline (Mean ± SD)	Year 1 (Mean ± SD)	Year 5 (Mean ± SD)	Relative Improvement
UPDRS-III (Motor)*	42.8 ± 9.4	21.1 ± 10.6	27.6 ± 11.6	51% at Y1 (P<.001), 36% at Y5 (P<.001)
UPDRS-II (ADL)*	20.6 ± 6.0	12.4 ± 6.1	16.4 ± 6.5	41% at Y1 (P<.001), 22% at Y5 (P<.001)
Dyskinesia Score	4.0 ± 5.1	1.0 ± 2.1	1.2 ± 2.1	75% at Y1 (P<.001), 70% at Y5 (P<.001)
Levodopa Equivalent Dose	Baseline	-28%	-28%	Stable 28% reduction (P<.001)

Assessed in medication-off state

Table 2: aDBS Programming Parameters and Neural Biomarkers [35] [34]

Parameter	Range	Default	Clinical Significance
Frequency	130-185 Hz	130 Hz	Higher frequencies may improve efficacy but increase side effect risk
Pulse Width	60-90 μs	60 μs	Wider pulses increase volume of tissue activation
Amplitude	1.0-4.0 V / 0.5-3.5 mA	1.5 mA	Titrated based on therapeutic window and side effects
Beta Band Threshold	75-85% peak amplitude	80%	Triggers stimulation when beta power exceeds this level
Stimulation Delay	0-500 ms	100-300 ms	Minimizes lag between biomarker detection and stimulation
mEP Score	0-100 arbitrary units	N/A	Quantitative biomarker for capsular activation side effects

Stepwise Clinical Programming Protocol

Phase 1: Pre-Activation Assessment (Day 0)

Step 1: Lead Localization Verification

• Confirm STN lead placement using postoperative CT or MRI fused with preoperative planning images
• Verify DBS lead coordinates relative to planned target and assess proximity to internal capsule
• Document specific electrode contacts available for programming based on final lead position

Step 2: Baseline Clinical Assessment

• Perform UPDRS Part III (motor examination) in medication-off state (≥12 hours overnight withdrawal)
• Assess tremor, rigidity, bradykinesia, and axial symptoms using standardized rating scales
• Document pre-existing neurological deficits and medication regimen
• Establish patient's typical diurnal symptom fluctuations

Step 3: System Integrity Check

• Verify impedance values for all electrode contacts (expected range: 500-1500 Ω)
• Check system integrity including connections, battery status, and telemetry functionality
• Ensure proper communication between implantable pulse generator and programming device

Phase 2: Initial Parameter Determination (Day 1)

Step 4: Biomarker Identification and Calibration

• Record local field potentials (LFPs) from all adjacent contact pairs in rest state
• Identify beta band (13-35 Hz) oscillatory activity characteristic of Parkinsonian state
• Determine individual patient's beta peak frequency and amplitude (typically 18-25 Hz)
• Set initial beta amplitude threshold at 80% of peak resting amplitude for aDBS triggering

Step 5: Therapeutic Window Mapping

• Begin with monopolar review of each electrode contact using fixed parameters (60 μs pulse width, 130 Hz)
• Gradually increase amplitude in 0.1 mA increments until therapeutic benefit observed (improvement in rigidity or tremor)
• Continue increasing until side effects emerge (muscle contraction, paresthesia, speech disturbance)
• Document therapeutic window for each contact (difference between threshold for benefit and side effects)
• Select contact with widest therapeutic window for initial programming

Step 6: Initial aDBS Parameterization

• Program initial aDBS parameters based on therapeutic window mapping:
  • Amplitude: Set at 50-70% of side effect threshold
  • Frequency: 130 Hz (default), may increase to 185 Hz if inadequate response
  • Pulse width: 60 μs (default), may increase to 90 μs if more widespread activation needed
  • Beta threshold: 80% of peak resting amplitude
  • Stimulation delay: 200 ms (default)

Phase 3: System Optimization and Validation (Weeks 1-4)

Step 7: Closed-Loop Validation

• Activate aDBS mode and record neural signals during stimulation
• Verify appropriate beta suppression with stimulation delivery
• Confirm return of beta activity when stimulation ceases
• Assess for appropriate triggering during volitional movements that normally suppress beta oscillations

Step 8: Clinical Effect Verification

• Perform UPDRS Part III in medication-off state with aDBS active
• Compare scores to preoperative and pre-aDBS baselines
• Assess specific symptom improvement (tremor, rigidity, bradykinesia)
• Document any stimulation-induced side effects

Step 9: Medication Adjustment

• Gradually reduce dopaminergic medication starting 1-2 weeks after aDBS activation
• Focus initially on reducing levodopa equivalent dose by 20-30% based on clinical response
• Monitor for emergence of non-motor symptoms during medication reduction
• Adjust aDBS parameters if necessary to compensate for medication reduction

Phase 4: Long-Term Management (Months 1-12)

Step 10: Progressive Parameter Optimization

• Schedule follow-up visits at 2 weeks, 6 weeks, 3 months, 6 months, and 12 months
• At each visit, assess clinical response and adjust aDBS parameters as needed
• Consider increasing stimulation amplitude if therapeutic effect wanes
• Fine-tune beta threshold based on patient's symptom fluctuations
• Utilize mEP scoring if capsular activation side effects occur [35]

Step 11: Remote Monitoring and Data Review

• Utilize remote programming capabilities when available
• Review stored LFP data to assess system responsiveness over time
• Correlate biomarker patterns with patient-reported symptom diaries
• Adjust parameters based on temporal patterns of symptom expression

Signaling Pathways and Experimental Workflows

Closed-Loop aDBS Activation Workflow

aDBS Mechanism of Action

Research Reagent Solutions

Table 3: Essential Research Materials for aDBS Studies [35] [34]

Reagent/Equipment	Function/Application	Specifications
Vercise DBS System	Clinical aDBS platform	Multiple independent current control, directional leads, sensing capability
GCaMP6f/GCaMP8f	Calcium indicator for neuronal activity monitoring	Genetically encoded calcium indicator expressed in STN neurons
SF-Venus-iGluSnFR.S72A	Glutamate release sensor	Fluorescent sensor for measuring extracellular glutamate levels
iGABASnFR.F102G	GABA release sensor	Fluorescent sensor for measuring extracellular GABA levels
AAV9-hSyn-DIO-tdTomato	Fluorescent control and anatomical marker	Red fluorescent protein for control measurements and surgical guidance
Hybrid Electrode-Optical Fiber Probe	Combined stimulation and recording	Enables simultaneous DBS delivery and fluorescence signal collection
Spectrally Resolved Fiber Photometry	Neural signal recording	Measures fluorescence signals from multiple sensors during DBS
mEP Detection Algorithm	Quantitative side effect assessment	Automated detection of motor evoked potentials indicating capsular activation

Discussion and Technical Considerations

The stepwise protocol outlined above provides a structured framework for aDBS activation that balances therapeutic efficacy with safety considerations. Several technical aspects require particular attention during implementation:

Biomarker Selection and Validation: While beta oscillations serve as the primary control signal for most current aDBS systems, individual variations in beta peak frequency and amplitude necessitate careful patient-specific calibration [35]. Additional biomarkers such as phase-amplitude coupling or network-level oscillations may provide complementary control signals for future systems.

Stimulation Parameter Interactions: The relationship between aDBS parameters (amplitude, frequency, pulse width) and closed-loop control characteristics (threshold, delay, duration) involves complex interactions that affect both efficacy and efficiency. Conservative initial settings with gradual escalation minimize side effect risk while determining optimal combinations.

Safety Monitoring: Continuous assessment for stimulation-induced side effects, particularly capsular activation evidenced by muscle contractions, remains essential throughout the programming process. The mEP score provides a quantitative biomarker for such side effects and can guide contact selection and parameter adjustment [35].

The progressive nature of Parkinson's disease necessitates ongoing parameter adjustments, as evidenced by the slight decline in motor scores between years 1 and 5 in long-term DBS studies, despite maintained overall benefit [32]. aDBS systems offer the potential to automatically adapt to disease progression through continuous biomarker monitoring and parameter optimization, potentially sustaining initial therapeutic gains more effectively than conventional open-loop stimulation.

Leveraging Artificial Intelligence for Neural Decoding and Symptom Estimation

This document provides application notes and detailed experimental protocols for leveraging artificial intelligence (AI) in neural decoding and symptom estimation, framed within the development of closed-loop deep brain stimulation (DBS) systems for Parkinson's disease (PD) research. Closed-loop DBS represents a paradigm shift from traditional continuous (open-loop) stimulation by automatically adjusting therapy in response to a patient's real-time neural and physiological state [36] [37]. This approach promises more personalized therapy, reduced side effects, and prolonged battery life compared to open-loop systems [36] [37].

AI and machine learning are the cornerstone technologies enabling this advancement. They allow researchers to decode complex neural signals and objectively quantify motor symptoms, facilitating the intelligent control algorithms required for adaptive DBS [38] [39]. These protocols are designed for researchers, scientists, and drug development professionals working to translate novel biomarkers and AI models into next-generation neuromodulation therapies.

AI for Neural Decoding in Brain-Computer Interfaces

A critical application of AI in closed-loop DBS is decoding neural signals to infer a patient's intended actions or clinical state. This is fundamental for developing brain-computer interfaces (BCIs) that could restore communication or improve neuromodulation.

Key Experimental Findings

Recent studies have demonstrated the high potential of this approach:

• Semantic Decoding from Intracranial Recordings: Researchers analyzed rare intracranial recordings from epilepsy patients thinking about words from 15 categories. A machine learning tool achieved up to 77% accuracy in predicting a word's category (versus 7% by chance) and 97% accuracy in distinguishing between living and non-living objects [38]. This represents the highest reported accuracy for this type of semantic decoding and suggests brain activity can robustly reveal word meaning.

• Decoding for Gait Disturbance Prediction: In the context of PD, a deep learning model was trained on data from patients to identify the unique neural signature of Freezing of Gait (FoG), a debilitating symptom. The model demonstrated the ability to predict FoG episodes before their onset, showing potential for real-time DBS intervention to prevent freezing [38].

Protocol: Decoding Semantic Content from Intracranial Signals

Objective: To train a machine learning model to decode the category of words a patient is thinking about from local field potential (LFP) or electrocorticogram (ECoG) recordings.

Materials:

• Patients with implanted electrodes (e.g., for epilepsy monitoring).
• Neural data acquisition system (e.g., Medtronic Summit RC+S or Activa PC+S for research [36]).
• Stimulus presentation software.
• Computing environment for signal processing and machine learning (e.g., MATLAB [40] or Python with Matplotlib [41]).

Methodology:

• Stimulus Presentation: Present participants with words from a pre-defined set of categories (e.g., tools, animals, food) visually or audibly. Each trial should involve the participant actively thinking about the presented word.
• Neural Data Acquisition: Simultaneously record multi-channel time-domain neural signals (LFPs/ECoG) from the implanted electrodes during the task [36].
• Signal Preprocessing: Apply necessary preprocessing steps: bandpass filtering, line noise removal, and artifact rejection.
• Feature Extraction: Transform the neural recordings to extract relevant features. Common features include:
  • Spectral Power: Calculate the average power within specific frequency bands (e.g., theta: 4-8 Hz, beta: 13-30 Hz) across multiple electrode contact pairs [36]. The optimal frequency bands and recording contacts are identified through a biomarker discovery phase.
  • Temporal Features: Extract features from the raw time-series signal.
• Model Training: Use a supervised machine learning approach. The features (neural data) are the inputs, and the word categories are the labels. A classifier, such as Linear Discriminant Analysis (LDA), is trained to map the neural features to the correct category [36].
• Model Validation: Evaluate the model's performance using cross-validation on held-out test data, reporting metrics like classification accuracy.

AI for Quantifying Parkinson's Disease Motor Symptoms

Accurate, continuous symptom tracking is essential for tuning and assessing closed-loop DBS efficacy. AI enables objective, home-based quantification of PD motor symptoms, moving beyond episodic clinic visits.

Key Symptom Domains and AI Applications

Table 1: AI-Assisted Quantification of PD Motor Symptoms

Symptom Domain	Characteristic Manifestations	AI/Data Modality	Quantifiable Features
Gait & Posture	Bradykinesia, Freezing of Gait (FoG), postural instability [39]	Smartphone video analysis [38], wearable sensors (lower back, legs) [42]	Stride length, cadence, swing time, trunk sway, FoG prediction from neural or kinematic signatures [38]
Tremor	Resting tremor (4-6 Hz), postural tremor [39]	Wearable sensors (wrist) [42], accelerometers, gyroscopes	Tremor frequency, amplitude, duration
Bradykinesia	Slowness of movement, reduced amplitude, difficulty with repetitive tasks [39]	Wearable sensors (wrist, fingers) [42], computer vision	Speed, rhythm, and amplitude decay in finger-tapping tasks; arm swing magnitude
Speech	Hypokinetic dysarthria (soft, monotone voice) [39]	Audio recording, speech signal processing	Vocal intensity, pitch variation, speech rate, articulation precision
Facial Expression	Hypomimia (reduced expression) [39]	Computer vision (facial landmark tracking)	Frequency and amplitude of eyebrow, cheek, and mouth movements

Protocol: Markerless Gait Analysis Using Smartphone Video

Objective: To use AI and computer vision to extract quantitative gait parameters from standard smartphone videos, matching the output of expert clinicians or 3D motion capture systems [38].

Materials:

• Smartphone with video recording capability.
• Well-lit, unobstructed space for walking.
• Computer vision software libraries (e.g., OpenPose, MediaPipe).
• AI model for gait parameter estimation (e.g., convolutional neural networks).

Methodology:

• Video Acquisition: Record participants walking back and forth over a pre-defined distance (e.g., 10 meters) using a smartphone. Ensure the entire body is visible throughout the walk cycle. A collection of hundreds of videos is needed for model development [38].
• Pose Estimation: Process the video frames using a pose estimation algorithm. This algorithm identifies and tracks key body joints (e.g., ankles, knees, hips, wrists, elbows, shoulders) in 2D or 3D space.
• Feature Extraction: From the time-series data of joint positions, extract spatiotemporal gait parameters:
  • Temporal Parameters: Stride time, stance time, swing time, double support time.
  • Spatial Parameters: Stride length, step width, walking velocity.
  • Dynamic Parameters: Joint angles, trunk sway, arm swing symmetry.
• Model Validation: Validate the AI-derived measurements against gold-standard references. This can be done by comparing the AI outputs to:
  • Scores from expert rehabilitation clinicians who review the same videos [38].
  • Data from a synchronized 3D motion capture system.
  • Output from wearable inertial sensors worn during the walk [42].

Implementing a Closed-Loop DBS System

The integration of neural decoding and symptom estimation enables the core function of a closed-loop DBS system: dynamic therapy adjustment.

The Closed-Loop DBS Pipeline

The development and operation of a closed-loop DBS algorithm can be broken down into a structured pipeline [36].

Parameter Tables for System Configuration

Configuring a closed-loop DBS system involves optimizing parameters from three main categories [36].

Table 2: Closed-Loop DBS Programming Parameters

Parameter Category	Specific Parameters	Description & Optimization Goal
Feature Selection	Recording Contacts, Frequency Bands, Averaging Window Size, Update Rate	Identify the optimal neural signal source and spectral feature (e.g., beta band power) that correlates with symptom severity. Goal: Maximize correlation between feature and clinical state [36].
Classifier	LDA Weights, State Thresholds	Configure the on-board classifier (e.g., Linear Discriminant Analysis) to map neural features to device states (e.g., "high symptom" vs. "low symptom"). Goal: Balance detection sensitivity and specificity to minimize false positives [36].
Stimulation Control	Onset Duration, Termination Duration, Stimulation Amplitude/Frequency	Define how stimulation is dynamically adjusted. Onset/Termination durations provide hysteresis to prevent rapid state switching. Goal: Deliver effective therapy only when needed, minimizing energy use and side effects [36].

The Scientist's Toolkit: Research Reagent Solutions

Table 3: Essential Materials and Tools for Closed-Loop DBS Research

Item	Function/Application in Research
Research DBS Implants (Medtronic Activa PC+S, Summit RC+S [36])	Implantable pulse generators capable of simultaneous neural sensing (local field potentials), on-board computation, and controlled stimulation.
Wearable Sensors (Inertial Measurement Units - IMUs [42])	Objective, continuous monitoring of motor symptoms (gait, bradykinesia, tremor) in home environments to correlate with neural biomarkers.
Linear Discriminant Analysis (LDA)	A simple, deterministic classifier embedded in research DBS devices for real-time classification of neural states into categories (e.g., symptom vs. non-symptom) [36].
Software Libraries (MATLAB [40], Python/Matplotlib [41])	Platforms for complex data analysis, signal processing, feature extraction, machine learning model development, and visualization of neural and kinematic data.
Computer Vision Tools (OpenPose, MediaPipe)	Enable markerless motion analysis from video for quantifying gait and bradykinesia, facilitating accessible, equipment-free assessment [38].

Workflow for Algorithm Development and Testing

A step-by-step workflow for bringing a closed-loop DBS algorithm from development to implementation is critical for clinical translation.

Protocol: Developing and Deploying a Patient-Specific Closed-Loop DBS Algorithm

Objective: To create and validate a personalized closed-loop DBS algorithm that automatically adjusts stimulation based on a patient's neural biomarker.

Methodology:

• Recording Phase (Weeks 1-12):
  • Long-Term Recording & Symptom Logging: Following device implantation, initiate a recording-only period. Collect continuous neural data (LFPs) while patients concurrently complete standardized symptom surveys multiple times daily [36]. Supplement with data from wearable sensors for 24/7 motor symptom monitoring [42].
  • Biomarker Discovery & Feature Selection: Analyze the collected data to identify a neural feature (e.g., beta band power in the subthalamic nucleus) that fluctuates with the patient's reported or sensor-quantified symptom severity [36].
  • Classifier & Control Parameter Fitting: Using the discovered biomarker, fit a classifier model (e.g., calculate LDA weights and thresholds) to separate "high-symptom" from "low-symptom" brain states. Define initial control parameters (onset/termination durations) [36].
• Stimulation Testing Phase:
  • Open-Loop Stimulation Testing: Before deploying closed-loop control, conduct a period of open-loop stimulation to establish baseline therapeutic efficacy and determine effective stimulation parameters [36].
• Closed-Loop Therapy Phase:
  • Closed-Loop Algorithm Deployment: Program the patient's DBS device with the tuned parameters: the neural feature, classifier weights/thresholds, and stimulation control logic. Activate the closed-loop mode.
  • In-Clinic & At-Home Validation: Monitor the system's performance and clinical outcomes closely. Validate that stimulation is being delivered appropriately (e.g., during "high-symptom" states) and that it provides superior or equivalent symptom control with less energy use compared to open-loop stimulation [36] [37]. Use continued wearable sensor data and patient reports for longitudinal assessment [42].

Overcoming Clinical and Technical Hurdles in aDBS Therapy

Deep brain stimulation (DBS) is an established therapy for Parkinson's disease, but optimizing stimulation parameters remains challenging. Within closed-loop DBS systems, two interrelated programming challenges are critical for achieving optimal therapeutic outcomes: selecting the most effective stimulation contact and accurately identifying pathological beta-frequency oscillations (13-35 Hz) in local field potentials (LFP) to guide therapy adjustment [43] [1]. This application note synthesizes current research findings and provides detailed protocols to address these challenges, enabling more efficient programming of adaptive DBS systems.

Quantitative Data Synthesis

Table 1: Performance Comparison of Contact Selection Prediction Methods

Prediction Method	Feature Used	Dataset	Accuracy (%)	Notes	Source
Decision Tree	Beta-band Max Power	Netherlands (68 STN)	86.5	Predicting top 2 contact-levels	[44]
Decision Tree	Beta-band Max Power	Switzerland (21 STN)	86.7	Predicting top 2 contact-levels	[44]
Decision Tree	Beta-band Max Power	Germany (32 STN)	75.0	Predicting top 2 contact-levels	[44]
Pattern Based	AUC_flat	Combined	84.6	Multi-center performance	[44]
DETEC Algorithm	Beta Power	Combined	<70.0	Consistently underperformed newer methods	[44]
Logistic Regression	Beta Peak Power Ratio	27 Patients	23.7% variance explained	Predictive value for therapeutic window	[43]

Table 2: Beta Peak Detection Algorithm Performance Comparison

Algorithm Type	Representative Algorithms	Accuracy Range (%)	Match to Expert Consensus	Key Characteristics	Source
Algebraic Dynamic Peak Amplitude Thresholding	III, IV, V, VII, VIII, IX	66-76%	No significant difference	Most accurate category	[45]
Elemental Decomposition	N/A	84-97%	High accuracy	Best performance overall	[46]
Other Methods	I, II, VI, X	30-75%	Significantly different	Variable performance	[45]

Experimental Protocols

Protocol 1: Local Field Potential Recording for Beta Peak Identification

Purpose: To acquire subthalamic local field potentials for beta peak detection and analysis in Parkinson's disease patients.

Materials:

• Implanted DBS electrodes with externalized extensions or sensing-capable implantable pulse generator
• Electrophysiology recording system (e.g., 306-channel magnetoencephalography system with integrated EEG amplifier)
• MATLAB with Brainstorm toolbox (version 12-Aug-2022 or later)
• Standardized medication protocol (levodopa)

Procedure:

• Patient Preparation:

  • Conduct recordings 1-3 days postoperatively with externalized DBS leads
  • Perform sessions both with (ON) and without (OFF) dopaminergic medication
  • Ensure resting state activity measurement with eyes open

• Data Acquisition:

  • Record three consecutive blocks of 10 minutes each per medication condition
  • Use monopolar referencing strategy for enhanced LFP resolution [45]
  • Sample at 1000 Hz with appropriate filtering
  • Apply notch filter (3-dB bandwidth of 1 Hz) to eliminate line noise
  • Implement 1 Hz high-pass filter to remove motion-related artifacts

• Signal Processing:

  • Re-reference channels to the mean of all LFP channels per patient
  • Generate power spectra using Welch's method with:
    • Window length: 4000 msec
    • Overlap ratio: 50%
  • Visually review and clean LFP data for artifacts
  • Exclude channels with flat, noisy activity or multiple artifacts

• Beta Peak Detection:

  • Use "fitting oscillations & one over f" Brainstorm implementation
  • Detect up to three peaks between 10-40 Hz in power spectra
  • Apply Gaussian peak model with parameters:
    • Peak width limit: 0.5-12.0 Hz
    • Minimum peak height: 3 dB (in log[Power])
    • Proximity threshold: two standard deviations
  • Calculate beta peak frequency, amplitude, standard deviation, and power for the highest peak amplitude in entire beta band and low beta band (12-20 Hz) separately [43]

Protocol 2: Data-Driven Contact Selection for DBS Programming

Purpose: To predict optimal stimulation contacts using local field potential biomarkers, reducing reliance on time-consuming monopolar reviews.

Materials:

• Chronically implanted neurostimulators with LFP recording capability (e.g., Medtronic, Abbott, Boston Scientific systems)
• Bipolar LFP recordings from DBS electrodes
• MATLAB or Python with signal processing toolbox
• Standardized clinical assessment scales (MDS-UPDRS-III)

Procedure:

• Data Collection:

  • Acquire bipolar LFP recordings from multiple contact pairs
  • Conduct recordings after overnight suspension of dopaminergic medications (OFF state)
  • Ensure adequate signal quality with visual inspection for ECG or other artifacts

• Feature Extraction:

  • Calculate beta-band power measures (13-35 Hz) for each recording channel
  • Extract both "Max" (maximum beta power) and "AUC_flat" (area under the curve with 1/f removal) features
  • Generate power spectral densities using Welch's method
  • Identify clear beta activity using predetermined "AUC_flat" threshold (median value: 2.64)

• Channel Ranking:

  • Rank recording channels based on beta power feature amplitude
  • For clinically chosen contact-levels 1 and 2, prioritize sensing channels surrounding the stimulation levels (0-2 and 1-3)

• Contact Prediction:

  • Apply "decision tree" method using "Max" feature for in-clinic use:
    • Selection tree: Identify contacts with highest beta power
    • Elimination tree: Exclude contacts with poor signal-to-noise ratio
  • Implement "pattern based" method for offline validation:
    • Analyze spatial patterns of beta power across contact pairs
    • Predict optimal monopolar stimulation contact from bipolar recordings
  • Validate predictions against clinically selected contacts from monopolar review [44]

• Performance Evaluation:

  • Calculate predictive accuracy for top two contact-levels
  • Compare algorithm performance against existing methods (e.g., DETEC algorithm)
  • Assess agreement with chronic contact choices at 6-12 months post-operatively

The Scientist's Toolkit: Research Reagent Solutions

Table 3: Essential Materials and Tools for DBS Programming Research

Item	Specification	Research Application	Notes
DBS Electrodes	Directional leads (e.g., Vercise Cartesia)	Precise stimulation targeting	Enables current steering to optimize VTA
Implantable Pulse Generator	Sensing-capable (e.g., Percept, Activa)	Chronic LFP recording	Enables biomarker tracking over time
Signal Processing Software	MATLAB with Brainstorm toolbox	Spectral analysis and peak detection	Customizable algorithms for research
Anatomical Targeting Software	GUIDE XT, SureTune	VTA visualization and lead localization	Facilitates image-guided programming
Beta Peak Detection Algorithms	Algebraic dynamic peak amplitude thresholding	Objective beta peak identification	Reduces subjective bias in peak selection
LFP Feature Extraction	"Max" and "AUC_flat" metrics	Contact selection prediction	Standardized measures for comparison studies
Adaptive DBS Programming System	Dual Threshold aDBS	Closed-loop stimulation adjustment	Enables beta-guided therapy optimization

Implementation Considerations for Closed-Loop DBS

Effective implementation of these protocols in closed-loop DBS systems requires addressing several practical challenges. For beta peak identification, medication state significantly affects detection reliability, with OFF medication states generally providing more robust beta peaks for aDBS programming [1]. Additionally, approximately 25% of hemispheres may exhibit double beta peaks, necessitating continuous test stimulation and medication-induced beta power modulation to identify the most clinically relevant peak [1].

For contact selection, the decision tree method using beta-band "Max" power feature has demonstrated superior performance (75-87% accuracy) compared to existing algorithms across multiple clinical centers [44]. This method particularly excels for contacts 0, 1, and 2, though performance decreases for contact 3, suggesting anatomical considerations may complement electrophysiological guidance in these cases.

Integration of these approaches enables more efficient aDBS programming, with studies showing successful implementation of beta-guided aDBS resulting in significant improvements in patient-reported overall wellbeing and general movement compared to continuous DBS [1]. The proposed protocols provide a foundation for standardized, objective DBS programming that can reduce clinical burden and improve therapeutic outcomes in Parkinson's disease research.

Optimizing LFP Thresholds and Stimulation Limits to Prevent Under- and Over-Stimulation

The following tables consolidate key quantitative findings from recent clinical studies on adaptive Deep Brain Stimulation (aDBS) to serve as a reference for research and development.

Table 1: Local Field Potential (LFP) Detection Rates and Characteristics

Parameter	OFF Medication State	ON Medication State	Notes
Participants with Detectable LFP Peak	91.5% (54/59 participants) [47]	84.8% (56/66 participants) [47]	Peak defined as 8–30 Hz, ≥1.2 µVp
Bilateral Signal Detection	78.0% (46/59 participants) [47]	65.2% (43/66 participants) [47]	Higher rate OFF medication facilitates setup
Average Peak Frequency	17.5 ± 5.9 Hz [47]	18.1 ± 5.5 Hz [47]	Predominantly in low-beta range
Average Peak Amplitude	1.96 ± 1.3 µVp [47]	Data not fully specified	Measured OFF medication
Frequency Band Distribution (Hemispheres)	Alpha: 31.0%; Low-beta: 34.0%; High-beta: 35.0% [47]	Alpha: 21.2%; Low-beta: 41.4%; High-beta: 37.4% [47]	Medication shifts activity to higher frequencies

Table 2: Stimulation Parameters and Clinical Outcomes from Chronic aDBS Studies

Parameter	Continuous DBS (cDBS)	Adaptive DBS (aDBS)	Clinical Correlation
Stimulation Amplitude	Fixed at 2.04 mA (mean) [48]	Dynamic range: 0.58 ± 0.19 mA (mean) [48]	Upper limit ~2.28 mA; Lower limit ~1.71 mA [48]
Overall Well-being (EMA Score)	5.92 ± 1.01 [48]	6.73 ± 1.33 [48]	Significant group-level improvement (p=0.007) [48]
General Movement (EMA Score)	5.47 ± 1.22 [48]	6.20 ± 1.44 [48]	Trend towards enhancement (p=0.058) [48]
Therapeutic Window (TW)	Not Applicable	Beta Peak Power Ratio (ON/OFF) explains 23.7% of TW variance [49]	Contact with highest beta activity often has largest TW [49]
Patient Preference	Not Applicable	6 out of 8 patients chose to remain on aDBS long-term [48]	Indicates tolerability and perceived benefit

Experimental Protocols for aDBS Setup and Optimization

This section provides detailed methodological workflows for implementing and refining aDBS systems in a research context.

Protocol 1: Initial Biomarker Identification and Sensing Setup

This protocol outlines the critical first steps for identifying a reliable control signal [48] [47].

• A. Pre-Recording Preparation:

  • Medication State: Conduct the initial Signal Test or BrainSense Survey with the patient in the practical OFF medication state (e.g., overnight suspension of dopaminergic drugs). This maximizes the probability of detecting a pathological beta peak, which can be obscured by medication [48] [47].
  • Lead Configuration: Configure the sensing amplifier to record bipolar Local Field Potentials (LFP) from multiple contact pairs on the chronically implanted DBS electrode [44].

• B. Data Acquisition and Peak Detection:

  • Recording: Acquire resting-state LFP data with the DBS therapy temporarily suspended.
  • Spectral Analysis: Generate power spectral density estimates from the LFP data using methods such as Welch's periodogram (e.g., 4000 ms window, 50% overlap) [49].
  • Peak Identification: Identify oscillatory peaks within the beta frequency range (12–35 Hz). A Gaussian peak model can be applied with constraints for peak width (e.g., 0.5–12.0 Hz) and a minimum amplitude threshold (e.g., 3 dB in log-power) [49]. The objective is to identify a peak that is modulated by medication and task.

• C. Contact and Biomarker Validation:

  • Medication Challenge: If possible, administer a dose of levodopa and repeat the recording in the ON medication state. A physiologically relevant beta peak should demonstrate a reduction in power [48] [49].
  • Test Stimulation: Perform short-term test stimulation on candidate contacts. A relevant beta peak should be suppressed with effective stimulation [48].
  • Contact Selection: The contact pair yielding the highest amplitude, medication-responsive beta peak is selected for chronic sensing. In cases where the peak is absent on one hemisphere (~16% of cases in one study) or the sensing-compatible contact is clinically suboptimal, unilateral sensing can be enabled [48].

Protocol 2: Setting Initial LFP Thresholds and Stimulation Limits

This protocol describes the process of defining the dynamic operating range for the aDBS algorithm [48].

• A. Chronic LFP Monitoring:

  • Data Collection: Activate continuous LFP data logging (e.g., the "Timeline" feature) over several days while the patient is on their regular cDBS therapy and medication schedule. This captures the natural, long-term fluctuations of beta power [48].
  • Data Analysis: Review the accumulated multi-day LFP data to understand the distribution of beta power, including its upper and lower extremes and the presence of any large outliers that could distort the scale.

• B. Threshold Calculation:

  • Quantile-Based Method: Calculate the 25th and 75th percentiles of the beta power recorded during waking hours. These values are often used as the initial lower (LFPL) and upper (LFPU) thresholds, respectively [48].
  • Threshold Adjustment: Note that final LFP thresholds show strong inter-individual variance. Be prepared to adjust these values iteratively during the optimization phase. For example, one study reported final upper thresholds ranging from 225 to 3160, and lower thresholds from 100 to 2970 [48].

• C. Stimulation Limit Determination:

  • Upper Stimulation Limit (AU): Set to the amplitude that provides optimal symptom control without inducing persistent side effects (e.g., dyskinesia, dysarthria) [48].
  • Lower Stimulation Limit (AL): Crucially, evaluate this limit in the OFF medication state. Setting the lower limit based only on the ON state can lead to undertreatment and OFF symptoms, as occurred in 4 out of 8 patients in one study. The lower limit should be the minimum amplitude that provides adequate symptom relief when the patient is in a hypodopaminergic state [48].

Protocol 3: In-Clinic and At-Home Optimization Phase

This protocol ensures the aDBS system is correctly adapting to the patient's physiology and providing stable symptom control [48].

• A. Verification of Algorithm Adaptation:

  • Review Adaptation Logs: At the first optimization visit, review device logs to verify that the stimulation amplitude is dynamically moving between the set upper and lower limits in response to beta power fluctuations, rather than remaining stuck at one limit.
  • Troubleshoot Static Stimulation: If stimulation is static, adjust the LFP thresholds. For example, if the amplitude is consistently at the upper limit, the upper LFP threshold may be set too low and needs to be increased [48].

• B. Refinement of Stimulation Limits:

  • Address Under-Stimulation: If the patient experiences hypokinetic episodes (e.g., re-emergence of tremor, bradykinesia) even when the algorithm is adapting correctly, the lower stimulation limit (AL) is likely too low and should be raised [48].
  • Address Over-Stimulation: If the patient experiences hyperkinetic episodes (e.g., worsened dyskinesia), the upper stimulation limit (AU) may be too high and should be lowered [48].

• C. Long-Term Outcome Assessment:

  • Ecological Momentary Assessment (EMA): Deploy structured electronic diaries to collect real-world, at-home data on symptom severity, overall well-being, and side effects during both cDBS and aDBS phases. This provides ecologically valid outcome measures beyond in-clinic tests [48].

Workflow Visualization

The following diagram illustrates the core closed-loop control logic and the iterative programming workflow for aDBS.

Figure 1: aDBS Control Logic and Implementation Workflow.

The Scientist's Toolkit: Research Reagent Solutions

Table 3: Essential Materials and Analytical Tools for aDBS Research

Tool / Reagent	Specification / Function	Research Application
Implantable Neurostimulator	Device with sensing capability (e.g., Percept PC). Must record bipolar LFP and support aDBS algorithms.	Enables chronic LFP data acquisition and closed-loop stimulation delivery in freely behaving subjects [48] [47].
Directional DBS Leads	Electrodes with segmented contacts (e.g., SenSight). Allows for current steering.	Investigate precision targeting of beta oscillations and its impact on therapeutic window and side-effect profile [49] [47].
Spectral Analysis Software	Tool for LFP analysis (e.g., MATLAB with Brainstorm toolbox). Implements Welch's method and Gaussian peak detection.	Critical for identifying and characterizing subject-specific beta peaks and their modulation by therapy [49].
LFP Beta Power Metric	Feature extraction (e.g., "Max" power in beta band, "AUC_flat").	Serves as the primary control signal for the aDBS algorithm. Used to rank recording channels for optimal contact selection [44] [49].
Ecological Momentary Assessment (EMA)	Digital patient diary for symptom reporting.	Provides real-world, subjective outcome measures to correlate with physiological data and assess therapy efficacy [48].

Addressing Artifact-Related Maladaptation and Nocturnal Symptom Control

Closed-loop adaptive deep brain stimulation (aDBS) represents a significant advancement in the management of Parkinson's disease (PD) by dynamically adjusting stimulation parameters based on neural feedback. Unlike conventional continuous DBS (cDBS), which provides fixed stimulation, aDBS systems modulate therapy in response to physiological biomarkers, such as subthalamic nucleus (STN) beta band power, which correlates with bradykinesia and rigidity severity [1] [27]. However, the practical implementation of aDBS faces two significant challenges: artifact-related maladaptation, where stimulation artifacts corrupt feedback signals and disrupt system control, and the need for effective nocturnal symptom control, as patients often experience persistent motor and non-motor symptoms during sleep [1] [50]. These application notes provide detailed protocols and strategies to address these challenges, ensuring optimal aDBS performance and improved patient outcomes across the sleep-wake cycle.

Quantitative Outcomes of aDBS and Sleep Improvement

Table 1: Clinical Outcomes of Chronic Adaptive DBS in Parkinson's Disease

Outcome Measure	cDBS Performance (Mean ± SD)	aDBS Performance (Mean ± SD)	Statistical Significance (p-value)	Notes
Overall Well-being (EMA score)	5.92 ± 1.01	6.73 ± 1.33	p = 0.007	Group-level significant improvement [1]
General Movement (EMA score)	5.47 ± 1.22	6.20 ± 1.44	p = 0.058	Non-significant trend toward improvement [1]
Dyskinesia Severity	3.12 ± 1.70	3.18 ± 1.76	p = 0.988	No significant group-level change [1]
Tremor Symptoms	2.54 ± 1.51	2.56 ± 1.67	p = 0.988	No significant group-level change [1]
Favorable Clinical State	81.5 ± 25%	87.0 ± 16%	p = 0.23	Shift towards more ON time [1]
Patient Preference	N/A	6 of 8 patients	N/A	Chose to remain on aDBS long-term [1]

Table 2: Impact of DBS on Sleep Parameters in Parkinson's Disease

Sleep Parameter	DBS Target	Findings	Evaluation Method	Source
Total Sleep Time	STN	Increase of up to 1 hour	Polysomnography (PSG), Actigraphy	[50]
Wake After Sleep Onset	STN	Reduction of 51-72 minutes	Polysomnography (PSG)	[50]
Nocturnal Motor Symptoms	STN & GPi	Improvement leading to less fragmentation	Parkinson's Disease Sleep Scale (PDSS)	[50]
Rapid Eye Movement Behavior Disorder (RBD)	STN	No significant impact	RBD Diagnostic Criteria	[50]
Restless Legs Syndrome (RLS)	STN	Reduction in severity; 27% complete resolution	IRLS Rating Scale	[50]
Sleep & Arousal Disturbances	STN (left)	Significant improvement associated with a specific "sweet spot"	Self-report measures, Sweet-spot analysis	[51]

Experimental Protocols for aDBS Configuration and Validation

Protocol for Initial aDBS System Configuration

This protocol outlines the steps for the initial programming of a commercially available dual-threshold aDBS system based on subthalamic beta power.

• Objective: To establish a stable aDBS configuration that effectively tracks symptom fluctuations while minimizing artifact-related maladaptation.
• Materials: Sensing-capable implantable pulse generator, programming console, electroencephalographic recording equipment.
• Procedure:
  • Patient Preparation: Conduct the initial programming visit with the patient in the OFF medication state (typically ≥12 hours overnight withdrawal) to maximize beta peak visibility [1].
  • Biomarker & Contact Selection:
    • Perform a "Signal Test" to record local field potentials (LFPs) from adjacent sensing contacts.
    • Identify and select the most prominent and clinically relevant beta peak (e.g., low-beta, 13-20 Hz; high-beta, 20-35 Hz) for the control signal. In cases of double beta peaks, use continuous test stimulation and assess medication-induced beta power modulation to identify the most responsive peak [1].
    • If no clear beta peak is found, review prior OFF-medication data or repeat the Signal Test OFF medication.
  • Threshold Definition:
    • Enable continuous "Timeline" data acquisition over several days to review long-term beta power modulation.
    • Set the initial lower (L) and upper (U) local field potential (LFP) thresholds to the 25th and 75th percentiles of daytime beta power, as per manufacturer guidance. Expect strong inter-individual variance (e.g., U: 1011 ± 924; L: 691 ± 843) [1].
    • Avoid using short-term "BrainSense Streaming" data alone for final threshold setting, as it may misrepresent long-term beta dynamics.
  • Stimulation Limit Calibration:
    • Determine the upper stimulation amplitude limit by identifying the intensity that induces acute side effects (e.g., muscle contractions, paresthesia).
    • Establish the lower stimulation amplitude limit by identifying the minimum effective amplitude that controls OFF-medication symptoms, particularly bradykinesia and rigidity, to prevent nocturnal OFF dystonia and morning OFF periods [1].

Protocol for Nocturnal Symptom Control Optimization

This protocol focuses on tailoring aDBS parameters to address sleep-related symptoms.

• Objective: To mitigate nocturnal OFF periods, dystonia, and other sleep-disturbing symptoms without inducing stimulation-related side effects like dyskinesia.
• Materials: aDBS system, ecological momentary assessment (EMA) tool or sleep diary, Parkinson's Disease Sleep Scale (PDSS-2).
• Procedure:
  • Baseline Assessment: Use the PDSS-2 or a detailed sleep diary for one week to establish a baseline of nocturnal symptom frequency and severity.
  • Parameter Adjustment:
    • For Nocturnal OFF Symptoms: If the patient experiences OFF dystonia or prolonged morning OFF periods, systematically increase the lower stimulation amplitude limit. Re-evaluate the minimum effective stimulation in the OFF medication state for a more robust setting [1].
    • For Sleep Fragmentation: If sleep disruption is linked to dyskinesia (potential over-stimulation), consider a slight reduction of the upper stimulation amplitude limit.
  • Validation: After parameter adjustment, repeat the sleep assessment with PDSS-2 or sleep diaries for one week to evaluate efficacy.
  • Long-term Management: Encourage patients to report specific nocturnal events. Use the device's "Timeline" event marker feature, if available, to correlate symptoms with LFP patterns and stimulation amplitudes for further fine-tuning.

Signaling Pathways and Workflow Diagrams

Diagram 1: aDBS closed-loop system with key challenges.

Diagram 2: Algorithm for removing stimulation artifacts.

The Scientist's Toolkit: Research Reagent Solutions

Table 3: Essential Materials and Tools for aDBS Research

Item	Function/Application	Specification Notes
Sensing-Capable Implantable Pulse Generator	Records local field potentials (LFPs) and delivers adaptive stimulation.	Commercial systems with sensing and stimulation capabilities on the same electrode, e.g., Medtronic Percept [1].
Periodic Artifact Removal Algorithm	Removes high-amplitude DBS stimulation artifacts from neural recordings to recover underlying signals.	Must function with aliased frequencies, low sampling rates (200-250 Hz), and missing data segments. Publicly available code can be found on Github [52].
Polysomnography (PSG) Equipment	Objective measurement of sleep architecture (sleep latency, wake after sleep onset, sleep stages).	Gold standard for validating subjective sleep improvements from DBS [50].
Parkinson's Disease Sleep Scale (PDSS-2)	Validated questionnaire for subjective assessment of sleep disturbance in PD.	Used to track changes in nocturnal symptoms before and after aDBS optimization [50].
Ecological Momentary Assessment (EMA)	Real-time, at-home collection of patient-reported outcomes on well-being, movement, and dyskinesia.	Provides ecologically valid data for comparing cDBS and aDBS efficacy outside the clinic [1].

Managing Adverse Events and Balancing Efficacy with Burdensome Programming Visits

Adaptive Deep Brain Stimulation (aDBS) represents a significant evolution in the treatment of Parkinson's disease (PD), moving beyond continuous, open-loop stimulation to a closed-loop system that modulates therapy in response to real-time neural signals [53] [54]. This approach theoretically improves symptom control, reduces stimulation-induced side effects, and increases battery longevity compared to conventional DBS (cDBS) [53]. However, the clinical implementation of aDBS introduces unique challenges, particularly in managing device-related adverse events and balancing the efficacy of personalized therapy against the burden of complex programming visits. This application note provides a structured framework for researchers and clinicians to navigate these challenges, ensuring that the benefits of aDBS can be realized without being offset by operational complexities or patient safety concerns.

Comparative Efficacy and Efficiency of aDBS

Table 1: Key Performance Metrics of Adaptive DBS vs. Conventional DBS in Parkinson's Disease

Study Reference	Sample Size	Input Signal	Key Efficacy Findings	Efficiency/Stimulation Metrics
Little et al., 2013 [53]	8 PD patients	STN LFP (β band power)	Greater improvement in motor scores with aDBS (66.2%) vs. cDBS (54.3%)	Significant reduction in total energy delivered with aDBS
Rosa et al., 2015 [53]	1 PD patient	STN LFP (β band power)	Superior control of bradykinesia and reduced dyskinesias vs. cDBS	Well-tolerated without eliciting side effects
Malekmohammadi et al., 2016 [53]	5 tremor-dominant PD patients	Wearable watch (4–8 Hz tremor power)	36.6% reduction in average tremor during aDBS	Mean voltage was 76.35% lower than clinical cDBS; stimulation delivered 51.5% of the time
Little et al., 2016 [53]	8 advanced PD patients	STN LFP (β band power)	Better speech intelligibility with aDBS (70.4%) vs. cDBS (60.5%)	aDBS delivered stimulation only 42.6% of the time

Adverse Event Profile in DBS Therapy

Table 2: Categorization and Management of DBS-Related Adverse Events

Category of Adverse Event	Specific Examples	Management and Troubleshooting Strategies
Surgery- & Hardware-Related	Intracerebral hemorrhage, hardware infection, lead fracture, IPG pocket hematoma [55] [56]	Surgical revision, wound debridement, antibiotics, hardware replacement or explantation in severe cases [55]
Stimulation-Induced Neurological	Dysarthria, imbalance, gait disturbance, speech problems [53] [56]	Parameter adjustment (esp. amplitude), switching to aDBS mode to reduce current spread [53] [57]
Stimulation-Induced Psychiatric	Apathy, depression, impulse control disorders, hallucinations [56]	Medication review, stimulation parameter adjustment, psychiatric care and counseling
Disease Progression	Axial symptoms (gait, postural instability) less responsive to stimulation [56]	Multimodal therapy (physiotherapy, medication adjustment), managing expectations

Experimental Protocols for aDBS Implementation and Safety

Protocol 1: Initial Implantation and System Setup

Objective: To surgically implant the aDBS system with precision and minimize immediate surgical adverse events.

Materials: Preoperative high-resolution volumetric MRI, stereotactic planning software (e.g., iPlan Stereotaxy), DBS leads, implantable pulse generator (IPG), microelectrode recording system.

Methodology:

• Preoperative Planning: Fuse multiple MRI sequences to identify the target (e.g., STN or GPi). Plan the lead trajectory to avoid blood vessels, sulci, and ventricles [55].
• Intraoperative Procedure:
  • Frame placement and registration via CT-MRI fusion.
  • Perform a linear skin incision and create a burr hole with a dual-floor design to prevent protrusion of the burr hole cover [55].
  • Conduct microelectrode recording to map the target structure.
  • Implant the DBS lead and perform macrostimulation to assess therapeutic windows and identify stimulation-induced side effects (e.g., muscle contractions, paresthesia) [55].
  • Fix the lead to the burr hole cover and close the incision.
• IPG Implantation: Implant the IPG in the subclavian area under general anesthesia, typically on the same day. The connector is buried under the parietal bone to reduce erosion risk [55].
• Postoperative Care: Obtain CT scans to verify lead position and rule out hemorrhage. Remove staples between postoperative days 7-10 [55].

Protocol 2: aDBS Programming and Parameter Optimization

Objective: To establish effective and efficient aDBS parameters that control symptoms while minimizing side effects and programming burden.

Materials: Clinical programming interface, software for sensing and control algorithms, objective motor symptom scoring system (e.g., UPDRS), patient diary.

Methodology:

• Biomarker Identification: For PD, the primary biomarker is often the β-band (13-35 Hz) oscillatory activity in the local field potentials (LFPs) recorded from the STN [53]. Calibrate the baseline and threshold for this biomarker.
• Adaptive Strategy Selection:
  • ON/OFF Control: Stimulation is triggered only when the biomarker crosses a predefined threshold [53]. This is simple and highly efficient in energy saving.
  • Continuous/Proportional Control: Stimulation amplitude is continuously modulated in proportion to the biomarker's magnitude [53]. This may provide smoother symptom control.
• Parameter Setting:
  • Define the sensing window and filter settings for the biomarker.
  • Set the adaptive strategy (ON/OFF or proportional).
  • Determine the stimulation parameters (frequency, pulse width) and the control law that maps biomarker amplitude to stimulation output (e.g., minimum/maximum amplitude, gain) [53].
• In-Clinic Titration and Validation:
  • Perform blinded, randomized assessments comparing aDBS to cDBS and off-states using standardized motor scores.
  • Specifically assess symptoms prone to stimulation-induced side effects, such as speech and gait [53].
  • Use patient-reported outcomes to gauge tolerability.
• Long-Term Follow-up and Adjustment: Schedule follow-up visits to fine-tune parameters as the disease progresses or the patient's environment changes. Explore remote monitoring technologies to reduce visit burden [58].

The following workflow diagrams the core aDBS operational logic and its clinical implementation pathway.

Diagram 1: The core closed-loop feedback mechanism of an adaptive DBS system. The system senses a neural biomarker, processes it through a control algorithm, and adjusts stimulation output accordingly in real time [53].

Diagram 2: The clinical pathway for implementing aDBS, from patient selection to long-term management, highlighting key stages for ensuring safety and managing the programming burden [55] [59].

The Scientist's Toolkit: Essential Research Reagents and Materials

Table 3: Key Research Reagent Solutions for aDBS Investigation

Item/Category	Specific Examples & Specifications	Primary Function in aDBS Research
DBS System with Sensing Capability	Implantable pulse generators (IPGs) with integrated sensing amplifiers (e.g., Activa PC+S, Percept)	Allows simultaneous recording of local field potentials (LFPs) and delivery of stimulation, enabling closed-loop control.
Biomarker Analysis Software	Custom algorithms in MATLAB, Python for signal processing (e.g., beta power, coherence)	Processes raw neural data to extract features that correlate with symptoms, serving as the control signal for aDBS.
Motion Sensing / Wearables	Tri-axial accelerometers, gyroscopes, commercial wearable watches [53]	Provides objective, continuous motor symptom tracking (e.g., tremor power) for control or validation of aDBS efficacy.
Clinical Rating Scales	Unified Parkinson's Disease Rating Scale (UPDRS), especially Part III (Motor)	Gold-standard for objective, in-clinic assessment of motor symptoms during aDBS parameter validation.
Microelectrode Recording System	Multi-channel recording systems, micro-drives	Used intraoperatively to precisely map target nuclei (e.g., STN, GPi) for optimal lead placement.
Stereotactic Planning Software	iPlan Stereotaxy (Brainlab), StealthStation (Medtronic)	Software for fusing MRI/CT images and planning the precise surgical trajectory for DBS lead implantation.

The successful integration of aDBS into clinical practice hinges on a proactive and systematic approach to managing its unique profile of adverse events and operational complexities. By adhering to the structured protocols and utilizing the toolkit outlined in this document, researchers and clinicians can mitigate risks, streamline the programming process, and ultimately harness the full potential of this transformative technology. Future developments in automated programming and the integration of artificial intelligence are anticipated to further alleviate the burden on clinicians and pave the way for the widespread adoption of aDBS as the standard of care within the next decade [57].

Evaluating Efficacy, Safety, and Position in the Treatment Paradigm

Application Notes

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) provides sustained, clinically significant improvements in motor function, activities of daily living, and medication reduction for patients with moderate to advanced Parkinson's disease over a five-year period. Data from the prospective, randomized INTREPID trial demonstrates that while some decline occurs likely due to disease progression, the core benefits of DBS remain robust, establishing it as a durable therapeutic intervention rather than a last-resort option [32] [60].

These long-term outcomes provide a critical foundation for the development of next-generation closed-loop adaptive DBS systems. The sustained efficacy of conventional DBS validates STN as a key therapeutic target, while the observed minor decline in certain metrics highlights the need for more responsive neuromodulation strategies that can adapt to progressive disease changes [27].

Quantitative Outcomes from 5-Year Follow-Up

Table 1: Sustained Motor and Functional Improvements with STN-DBS

Assessment Metric	Baseline (Mean)	Year 1 Improvement	Year 5 Improvement	Statistical Significance
UPDRS-III (Motor, off-med)	42.8	51% (to 21.1)	36% (to 27.6)	P < 0.001 [32]
UPDRS-II (ADL, off-med)	20.6	41% (to 12.4)	22% (to 16.4)	P < 0.001 [32]
Dyskinesia Scores	4.0	75% reduction (to 1.0)	70% reduction (to 1.2)	P < 0.001 [32]
Levodopa Equivalent Dose	Baseline	28% reduction	28% reduction (sustained)	P < 0.001 [32]

Table 2: Safety Profile and Patient Satisfaction (5-Year)

Parameter	Result	Context
Most Common Serious Adverse Event	Infection (9 participants)	Most required surgical intervention [32] [61]
Mortality	10 deaths reported	None related to DBS therapy [32]
Patient Satisfaction	94%	High satisfaction maintained [61]

Experimental Protocols

Protocol 1: INTREPID Trial Design and Long-Term Follow-Up

Objective: To evaluate the long-term (5-year) efficacy and safety of bilateral STN-DBS for Parkinson's disease using a rigorous clinical trial design [32].

Methodology:

• Design: Prospective, randomized (3:1), double-blind sham-controlled initial phase, transitioning to open-label 5-year follow-up [32].
• Setting: 23 movement disorder centers across the United States [32].
• Participants: 313 enrolled patients with bilateral idiopathic PD meeting strict criteria: >5 years of motor symptoms, >6 hours/day of poor motor function, modified Hoehn and Yahr Scale scores >2, and UPDRS-III score ≥30 in medication-off state. All participants demonstrated ≥33% improvement in UPDRS-III with medication [32].
• Intervention: Bilateral implantation of the Vercise DBS system targeting the STN [32].
• Primary Outcomes: Change from baseline in UPDRS parts II and III in the medication-off state, dyskinesia scores, quality-of-life measures, and systematic safety assessments [32].
• Statistical Analysis: Utilized appropriate statistical models to calculate percentage improvements with 95% confidence intervals and determine significance (P-values) [32].

Protocol 2: Transitioning to Closed-Loop Adaptive DBS Systems

Objective: To implement a responsive neuromodulation system that automatically adjusts stimulation parameters based on real-time biomarkers of Parkinsonian state [27].

Methodology:

• System Architecture: Employ a sensing-enabled implantable pulse generator capable of both recording and stimulating.
• Input Signal Acquisition: Continuously monitor local field potentials (LFPs) from the implanted DBS electrodes, with specific focus on beta-band oscillations (13-30 Hz) as a validated biomarker correlated with rigidity and bradykinesia [27] [62].
• Control Algorithm: Program a closed-loop controller to analyze the beta-band power in real-time. Set a threshold for excessive beta activity that triggers stimulation delivery.
• Adaptive Stimulation: When the biomarker threshold is exceeded, the system automatically initiates or increases stimulation amplitude. Stimulation is reduced or suspended when biomarker levels normalize [27].
• Validation: Compare motor outcomes (e.g., UPDRS-III), stimulation energy usage, and adverse effect profiles between adaptive DBS and conventional continuous DBS settings in a controlled crossover design [27].

Diagram 1: Open-loop vs. closed-loop DBS control. The closed-loop system uses real-time biomarker feedback to automatically adjust stimulation.

Diagram 2: Five-year clinical validation workflow. The protocol from patient selection to long-term outcome assessment.

The Scientist's Toolkit

Table 3: Essential Research Reagents and Materials for DBS Investigation

Reagent/Material	Function/Application	Example Use Case
Vercise DBS System	Implantable neurostimulator for bilateral STN stimulation; enables controlled current delivery [32].	Primary intervention device in the INTREPID clinical trial for long-term efficacy and safety data collection [32].
Unified Parkinson's Disease Rating Scale (UPDRS)	Gold-standard clinical tool for quantifying PD motor severity (Part III) and disability in activities of daily living (Part II) [32].	Primary outcome measure to assess percentage improvement from baseline at years 1 and 5 in the medication-off state [32] [60].
Local Field Potential (LFP) Sensing	Records aggregate neural oscillations from DBS electrodes; provides biomarker input for adaptive systems [27] [62].	Senses beta-band (13-30 Hz) power as a control signal for closed-loop DBS algorithms [27] [62].
Levodopa Equivalent Dose (LED)	Standardized calculation to aggregate total dopaminergic medication load across different drug types [32].	Quantifies the sustained reduction (28%) in anti-parkinsonian medication requirements post-DBS [32] [61].
Adaptive Control Algorithm	Software that interprets biomarker signals and dictates real-time adjustments to stimulation parameters [27].	Core of closed-loop DBS; triggers stimulation when beta power exceeds a threshold, optimizing therapy delivery [27] [62].

Conventional deep brain stimulation (cDBS) delivers continuous, open-loop electrical stimulation to deep brain structures such as the subthalamic nucleus (STN) or globus pallidus interna (GPi), providing an established therapy for Parkinson's disease (PD) motor symptoms when medications become insufficient [27]. Despite its proven efficacy, cDBS possesses inherent limitations: its static parameters cannot respond to dynamic fluctuations in symptom severity, potentially leading to periods of overtreatment (causing side effects like dyskinesia and speech impairments) or undertreatment, alongside rapid battery consumption [27] [57]. Adaptive DBS (aDBS) represents a transformative, closed-loop alternative that dynamically adjusts stimulation parameters, typically amplitude, in real-time based on feedback from physiological biomarkers. The most established control signal is oscillatory power in the beta band (13-30 Hz) of local field potentials (LFPs) recorded from the DBS leads themselves, as this signal correlates with bradykinesia and rigidity severity in PD [48] [27]. This article synthesizes comparative clinical outcomes from recent real-world studies and pivotal trials, providing researchers and clinicians with a detailed framework for the application and implementation of aDBS technology.

Comparative Clinical Outcomes: aDBS vs. cDBS

Recent clinical studies have moved beyond short-term laboratory demonstrations to investigate the chronic, at-home use of aDBS, yielding critical data on its efficacy, energy efficiency, and safety profile compared to the standard of care, cDBS.

Table 1: Summary of Key Real-World Studies Comparing aDBS and cDBS

Study (Citation)	Design	Key Efficacy Findings (aDBS vs. cDBS)	Energy Efficiency	Safety & Tolerability
ADAPT-PD Pivotal Trial [25]	International, open-label, 68 participants	91% (DT-aDBS) and 79% (ST-aDBS) met the primary performance goal for maintaining good "on" time without troublesome dyskinesia.	ST-aDBS reduced TEED by ~15% (p=0.01). DT-aDBS showed a non-significant reduction.	All but one stimulation-related AE resolved during setup. No serious device-related AEs.
Chronic aDBS Study [48]	Real-world, 8 patients, home-based EMA	Significant improvement in overall well-being (p=0.007); a trend toward enhanced general movement (p=0.058). 6 of 8 patients chose to remain on aDBS.	Not explicitly reported.	Programming challenges (e.g., biomarker selection) were identified and addressed.
Randomized Crossover Trial [63]	Double-blind, randomized crossover, 9 patients	No statistically significant differences in primary outcomes. Heterogeneous effects: aDBS favored for dyskinesia duration and UPDRS scores; cDBS favored for ON-duration.	Not a primary outcome.	Demonstrated feasibility of blinded chronic evaluation.

The ADAPT-PD trial, the first large-scale study of chronic at-home aDBS, demonstrated that the therapy is not only feasible but also effective and safe in a real-world context [25]. The trial evaluated two aDBS algorithms: single-threshold (ST-aDBS) and dual-threshold (DT-aDBS). Its success was measured by a performance goal showing that the vast majority of participants maintained well-controlled symptom time without troublesome dyskinesias, performing at least as well as on cDBS [64] [25]. Furthermore, the reduction in Total Electrical Energy Delivered (TEED) indicates a potential for extended battery longevity, a significant practical advantage [25].

Supporting evidence comes from a smaller real-world study that used ecological momentary assessments at home. It found aDBS significantly improved patients' overall well-being, with a strong trend toward better movement, leading most participants to prefer and stay on adaptive stimulation long-term [48]. Conversely, a rigorous double-blind, randomized crossover trial found no statistically significant differences between aDBS and cDBS on its primary outcomes [63]. This study highlights the presence of heterogeneous treatment effects, suggesting that patient-specific factors, such as individual clinical characteristics and baseline disease burden, may critically influence which stimulation modality yields superior outcomes [63]. Exploratory analyses indicated that patients with a higher baseline disease burden might see greater benefits from aDBS for overall motor severity [63].

Experimental Protocols & Programming Methodologies

The translation of aDBS from a research concept to a clinical tool requires standardized yet flexible experimental and programming protocols. The following workflow and detailed breakdown outline the core methodology derived from recent trials.

Biomarker Identification and Signal Validation

The initial critical phase involves identifying a reliable LFP control signal.

• Sensing Configuration: The process requires DBS leads and a pulse generator (e.g., Medtronic Percept PC) capable of sensing LFPs. The clinical programmer is used to review signals from different contact pairs to identify those with a high signal-to-noise ratio and those that are clinically effective for stimulation, a challenge that sometimes necessitates unilateral sensing [48] [65].
• Peak Detection: Clinicians screen for a distinct peak in the power spectral density within the 8-30 Hz range (alpha and beta bands) with an amplitude ≥1.2 µVpeak [65]. This screening should be performed both OFF and ON medication. In the ADAPT-PD trial, LFP peaks meeting these criteria were identified in 91.5% of participants off medication and 84.8% on medication [65]. Medication can obscure the beta peak, so an OFF-medication assessment is recommended if no clear peak is found initially [48].
• Peak Selection: In cases of double beta peaks, continuous test stimulation and observing medication-induced beta power modulation can help identify the most physiologically and clinically responsive peak [48].

aDBS Algorithm Programming and Parameter Configuration

Once a suitable biomarker is identified, the aDBS algorithm is configured. The commercially available dual-threshold mode is a common focus.

• Stimulation Amplitude Limits: The clinician defines upper and lower boundaries for stimulation amplitude. The upper limit is typically set just below the threshold for inducing persistent adverse effects. The lower limit is the minimum amplitude required for adequate symptom control, which should be evaluated in the OFF-medication state to prevent undertreatment [48]. The resulting therapeutic window is personalized; in one study, the average range was 0.58 ± 0.19 mA [48].
• LFP Thresholds: The system requires thresholds for the LFP beta power to trigger amplitude adjustments. Guidance often suggests setting the upper and lower LFP thresholds to the 75th and 25th percentiles of the patient's daytime beta power, respectively [48]. These thresholds show strong inter-individual variance and are not static. A common challenge is stimulation becoming "stuck" at an upper or lower limit, requiring threshold adjustment based on multi-day data trends [48].
• Adaptation Dynamics: The speed of amplitude adjustment is configurable. The dual-threshold aDBS typically uses slower, incremental changes (e.g., increasing over 2.5 minutes and decreasing over 5 minutes) to prevent rapid fluctuations, whereas single-threshold mode can adjust much faster (e.g., within 250 ms) [65].

Optimization and Chronic Evaluation Protocol

The final phase involves refining these settings and evaluating long-term outcomes.

• In-Clinic Titration: The system's response is verified in-clinic using tools like "BrainSense Streaming" to confirm that stimulation amplitude appropriately tracks changes in beta power during various tasks and medication states [48] [65].
• At-Home Optimization: Patients are sent home with aDBS activated. Clinicians review chronic data streams (e.g., "BrainSense Timeline," which provides 24-hour views of LFP power and stimulation amplitude) to identify and correct issues like nocturnal OFF-dystonia or persistent dyskinesia, further refining amplitude limits and LFP thresholds over several days or weeks [48] [25].
• Outcome Assessment: Efficacy is evaluated against cDBS using blinded or randomized crossover designs with outcomes such as self-reported motor diaries (tracking "ON" time without troublesome dyskinesia), standardized rating scales like UPDRS, and measures of energy consumption (TEED) over periods of one to three months [63] [25].

The Scientist's Toolkit: Essential Research Reagents & Platforms

The implementation and advancement of aDBS research rely on a suite of specialized technologies and methodological approaches.

Table 2: Key Research Reagents and Platforms for aDBS Investigation

Tool / Technology	Function & Application in aDBS Research
Sensing-Enabled IPG (e.g., Percept PC)	An implantable pulse generator that records LFPs and delivers stimulation. It is the core platform for chronic aDBS studies, enabling the collection of long-term neural data and the deployment of adaptive algorithms in real-world settings [25] [65].
Directional DBS Leads (e.g., SenSight)	Leads with segmented electrodes that allow for more precise current steering. Their integration with aDBS enables stimulation to be adapted not only in amplitude but also in spatial delivery, potentially targeting specific symptom-related neural pathways [65].
BrainSense Streaming & Timeline	Software features within the clinical programmer. "Streaming" provides real-time LFP data for in-clinic algorithm testing, while "Timeline" captures long-term trends in LFP power and stimulation amplitude, which is crucial for optimizing chronic aDBS settings [48] [65].
Closed-Loop Research Platforms (e.g., clDBS Platform)	Custom research platforms (like the one described by [66]) that address core challenges such as real-time stimulation artifact removal and low-latency feedback control. They facilitate the translation of novel aDBS strategies from animal models to human clinical trials.
Beta-Band LFP (8-30 Hz)	The primary biomarker used for control algorithms. Its power and burst duration correlate with bradykinesia and rigidity severity, are attenuated by dopaminergic medication and DBS, and serve as the input signal for the control system in most current aDBS paradigms [48] [27] [65].

Real-world evidence confirms that chronic aDBS is a tolerable, effective, and safe therapy that can provide symptom control comparable or superior to cDBS, with the added benefits of dynamic personalization and reduced energy consumption. A Delphi consensus among DBS experts predicts that aDBS will become clinical routine within the next 10 years [57]. Future research must focus on defining optimal patient subgroups, simplifying programming procedures, developing multimodal control signals that incorporate non-motor symptoms, and integrating aDBS with complementary technologies like directional leads and artificial intelligence to achieve fully personalized neuromodulation therapies [63] [57] [66]. The protocols and data synthesized herein provide a foundational framework for researchers and clinicians driving this evolution forward.

Quantitative Evidence for LED Reduction Following DBS

Table 1: Documented Reductions in Levodopa Equivalent Daily Dose (LEDD) following Deep Brain Stimulation.

Study & Intervention	Patient Cohort	LEDD Reduction (Mean)	Key Findings
STN-DBS + Opicapone [67]	PD patients with preoperative dyskinesias (n=16)	-36.4%	Patients with significant preoperative dyskinesias showed less LEDD reduction but benefited from more stable dopamine replacement via COMT inhibition.
STN-DBS (Control Cohort) [67]	PD patients without opicapone (n=16)	-46.2%	Represents the typical, more substantial LEDD reduction achievable with STN-DBS in standard patient phenotypes.
Directional STN-DBS [68]	PD patients, directional vs. omnidirectional leads	-50.5% (Directional)	Directional stimulation, enabling a more focused delivery, was associated with a significantly greater reduction in LEDD compared to omnidirectional leads.
Omnidirectional STN-DBS [68]	PD patients, directional vs. omnidirectional leads	-41.8% (Omnidirectional)	Conventional omnidirectional stimulation provided a smaller, though still substantial, reduction in dopaminergic medication.

Experimental Protocol for Measuring LEDD Changes in DBS Studies

Objective

To quantitatively evaluate the stable reduction in Levodopa Equivalent Daily Dose (LEDD) following subthalamic nucleus deep brain stimulation (STN-DBS) for Parkinson's disease.

Preoperative Assessment (Baseline)

• Patient Selection: Enroll idiopathic PD patients with motor fluctuations and/or dyskinesias, a significant levodopa response (≥30% UPDRS-III improvement), and no contraindications for DBS surgery or cognitive impairment [69].
• LEDD Calculation: Calculate the baseline LEDD using standardized formulas prior to surgery [67]. Record all antiparkinsonian medications, including levodopa, dopamine agonists, MAO-B inhibitors, and COMT inhibitors.
• Clinical Phenotyping: Document specific clinical features, such as the presence and severity of dyskinesias (e.g., UPDRS-IV subscore), as these may predict the degree of postoperative LEDD reduction [67].

Surgical Intervention & Postoperative Management

• Targeting and Implantation: Perform bilateral STN-DBS electrode implantation. The use of directional electrodes should be noted for subsequent analysis [68].
• Stimulation Programming: Initiate stimulation 2-4 weeks post-surgery. Optimize parameters (contact selection, voltage, pulse width, frequency) to achieve maximal therapeutic benefit with minimal side effects.
• Medication Titration: Begin gradual reduction of dopaminergic medication, particularly levodopa and dopamine agonists, based on clinical response. The goal is to minimize dyskinesias and OFF-time while maintaining motor control [67] [69].

Outcome Assessment

• Primary Endpoint: Percent change in LEDD from baseline to a stable postoperative time point (typically ≥6 months post-surgery) [67].
• Secondary Endpoints:
  • Change in MDS-UPDRS Parts III (motor examination) and IV (motor complications) scores.
  • Changes in daily ON time without troublesome dyskinesia and OFF time.
  • Quality of life measures (e.g., PDQ-39).

Data Analysis

• Compare LEDD and clinical scores pre- and post-operatively using paired t-tests or Wilcoxon signed-rank tests.
• Use multivariate regression analysis to identify patient factors (e.g., age, preoperative LEDD, clinical phenotype) associated with the degree of LEDD reduction.

Mechanistic Workflow for Medication Reduction in Closed-Loop DBS

The following diagram illustrates the conceptual pathway through which advanced DBS systems facilitate a stable reduction in dopaminergic medication.

The Scientist's Toolkit: Key Reagents & Materials

Table 2: Essential Research Materials for Investigating DBS and Medication Reduction.

Tool / Material	Function/Application in Research
Medtronic Activa PC+S / Summit RC+S [36]	Research-grade implantable pulse generators capable of both sensing local field potentials (LFPs) and delivering stimulation, enabling closed-loop algorithm development.
Directional DBS Leads [68]	Electrodes with segmented contacts that allow for more focused current delivery, used to investigate the effect of precise stimulation on therapeutic window and LEDD reduction.
Local Field Potential (LFP) Sensing [36] [6]	The primary method for recording subcortical neural signals (e.g., beta-band oscillations) to serve as a control biomarker for closed-loop DBS.
Linear Discriminant Analysis (LDA) Classifier [36]	An embedded machine learning algorithm used on-device to classify neural states (e.g., normal vs. pathological) based on sensed features, triggering adaptive stimulation.
Levodopa Equivalent Dose (LEDD) Formulas [67]	Standardized equations to convert doses of all dopaminergic medications into a total daily levodopa equivalent, essential for quantifying medication changes.
MDS-UPDRS Scale [67] [70]	The gold standard clinical rating scale for assessing motor and non-motor symptoms of Parkinson's disease in therapeutic trials.

Safety Profile and Ethical Considerations in Adaptive Neurotechnology

Application Note: Clinical Safety and Efficacy

Adaptive neurotechnology, particularly closed-loop deep brain stimulation (aDBS), represents a significant advancement in the management of Parkinson's disease (PD). This application note synthesizes the latest clinical evidence and ethical frameworks to guide its responsible development and implementation for researchers and clinicians.

Safety and Efficacy Profile of Adaptive DBS

Recent clinical trials provide robust quantitative evidence for the safety and efficacy of chronic, at-home aDBS in Parkinson's disease. The pivotal ADAPT-PD trial, a multinational study, offers high-quality data on the long-term use of this technology.

Table 1: Key Efficacy and Safety Outcomes from the ADAPT-PD Pivotal Trial (N=68) [25]

Outcome Measure	Dual Threshold aDBS (n=40)	Single Threshold aDBS (n=35)	Notes
Primary Endpoint: Performance Goal Met	91% of participants	79% of participants	Goal: On-time without troublesome dyskinesias, with <2 hours/day reduction vs. cDBS
Therapeutic Energy Delivery (TEED)	Not significantly different from cDBS	15% reduction vs. cDBS (p=0.01)	Suggests potential for improved battery longevity with single threshold mode
Stimulation-Related Adverse Events (AEs)	All but one AE resolved during setup phase	All but one AE resolved during setup phase	No serious device-related AEs through long-term follow-up
Patient Preference (Post-Study)	6 of 8 patients chose to remain on aDBS [48]	N/A	Demonstrates clinical tolerability and patient acceptance

The trial concluded that long-term aDBS was tolerable, effective, and safe in PD patients previously stable on continuous DBS (cDBS) [25]. A smaller real-world study corroborates these findings, showing that aDBS led to significantly improved overall well-being (p=0.007) and a trend toward enhanced general movement (p=0.058) compared to cDBS [48].

Ethical Considerations and Governance

The rapid progression of neurotechnology necessitates parallel development of strong ethical frameworks. A comprehensive scoping review of the literature identifies six core ethical themes and corresponding governance strategies for responsible innovation [71].

Table 2: Core Ethical Themes and Proposed Governance Strategies in Neurotechnology [71]

Ethical Theme	Description of Concerns	Proposed Governance Strategies
Justice	Equitable access, fair distribution of risks/benefits, avoidance of disparities	Interdisciplinary collaboration, public engagement, scientific integrity
Beneficence & Nonmaleficence	Maximizing benefit while minimizing harm (physical, psychological)	Social responsibility and accountability, legislation & neurorights
Privacy & Brain Data	Protection of neural data, which can reveal thoughts, emotions, and reactions [72]	Legislation & neurorights, public engagement, epistemic humility
Autonomy & Informed Consent	Capacity for self-determination, quality of consent processes, influence on behavior	Interdisciplinary collaboration, public engagement, scientific integrity
Identity & Dignity	Impact on sense of self, personality, and personal identity	Social responsibility and accountability, public engagement
Moral Status	Potential blurring of lines regarding moral agency and responsibility	Epistemic humility, interdisciplinary collaboration

A critical analysis reveals that while these themes are recognized, ethical engagement in clinical studies is often superficial. A review of 66 clinical studies on closed-loop systems found that ethical issues are typically addressed only implicitly or reduced to procedural compliance like Institutional Review Board (IRB) approval, rather than subjected to substantive, reflective analysis [73]. This highlights a significant gap between theoretical ethics and clinical practice.

Experimental Protocols

Protocol: Programming and Implementing aDBS for Chronic Use

This protocol outlines the methodology for programming and evaluating adaptive Deep Brain Stimulation (aDBS) systems, based on established clinical workflows [74] [25].

Objective: To safely transition a patient with Parkinson's disease from conventional continuous DBS (cDBS) to chronic, at-home adaptive DBS (aDBS), ensuring therapeutic efficacy and managing potential programming challenges.

Materials:

• Patient implanted with a sensing-enabled DBS system (e.g., Medtronic Percept PC/RC).
• Corresponding clinician programmer and tablet.
• Institutional review board (IRB) approved study protocol (for research use) or clinical guidelines.
• Standardized clinical assessment tools (e.g., MDS-UPDRS III, patient motor diaries).

Procedure:

• Patient Selection & Baseline Assessment:

  • Confirm patient has idiopathic PD with bilateral DBS leads in the subthalamic nucleus (STN) or globus pallidus internus (GPi).
  • Ensure patient is on stable cDBS parameters and medication regimen.
  • Conduct baseline neurological assessments (MDS-UPDRS I-IV, MoCA) and quality of life measures (e.g., PDQ-39, FOG-Q) [74].

• Sensing Setup & Biomarker Identification (Visit 1 - SETUP):

  • Verify device integrity (battery, impedance).
  • Withhold patient's dopaminergic medication overnight (cDBS-ON/Meds-OFF state).
  • Perform BrainSense Survey/Setup to identify optimal sensing contacts.
  • Visually inspect local field potential (LFP) recordings to select a patient-specific alpha-beta peak (typically 13-35 Hz) for control.
  • Exclude contacts contaminated by artifacts (e.g., cardiac, movement-induced).
  • Activate BrainSense Streaming to confirm beta power modulation in response to incremental stimulation amplitude changes.
  • Activate Home Monitoring (Timeline) to capture several days of continuous LFP data and patient-marked events (e.g., medication intake, sleep, symptoms).

• aDBS Activation & Parameter Configuration (Visit 2 - ACTIVATION):

  • Review multi-day Timeline data to understand the variability and trends of the patient's beta power.
  • Set the dual threshold algorithm parameters [74]:
    • Upper LFP Threshold: Typically set to the 75th percentile of daytime beta power. Stimulation amplitude increases when beta power exceeds this threshold.
    • Lower LFP Threshold: Typically set to the 25th percentile of daytime beta power. Stimulation amplitude decreases when beta power remains below this threshold.
    • Stimulation Amplitude Limits: Define the upper and lower bounds for current adjustment, based on the patient's therapeutic and side-effect thresholds. Assess the minimum effective stimulation in the OFF medication state to prevent undertreatment [48].

• Optimization & Long-Term Follow-up (Subsequent Visits):

  • Refine LFP thresholds and amplitude limits based on patient-reported outcomes and device data.
  • Troubleshoot common challenges:
    • Stimulation "stuck" at limit: Adjust LFP thresholds to better match the dynamic range of the patient's beta power [48].
    • Persistent symptoms despite adaptation: Adjust the corresponding stimulation amplitude limit (e.g., raise the lower limit if OFF symptoms persist) [48].
  • After optimization, collect ecological momentary assessments (EMAs) or motor diaries at home under both cDBS and aDBS conditions for comparison.
  • Conduct final blinded assessment (MDS-UPDRS III) and ascertain patient preference for stimulation mode.

Protocol: Integrating Patient Perspectives in Neurotechnology Development

Objective: To qualitatively capture and integrate the experiences and values of neurotechnology end-users into the research, development, and commercialization process, addressing identified ethical gaps [75] [73].

Materials: * IRB-approved interview guide and consent forms. * Audio recording equipment and transcription service. * Qualitative data analysis software (e.g., NVivo).

Procedure:

• Participant Recruitment:

  • Recruit a purposive sample of patients who have used neurotechnology devices in therapeutic or research settings.
  • Utilize clinical networks or registries (e.g., ResearchMatch) for recruitment [75].

• Data Collection:

  • Conduct semi-structured interviews (approx. 16-20 participants) exploring [75]:
    • Experiences using the neurotechnology and its impact on daily living.
    • Perspectives on industry-academia (IA) partnerships.
    • Preferences regarding neural data use, privacy, and long-term sharing.
    • Views on responsibility for long-term device care and post-trial access.
    • Advice for future users and developers.
  • Audio-record and transcribe interviews verbatim.

• Data Analysis:

  • Employ inductive thematic analysis.
  • Familiarize yourself with the data by reading transcripts repeatedly.
  • Generate initial codes from the data.
  • Search for themes by collating relevant codes.
  • Review and refine themes to ensure they accurately represent the dataset.
  • Define and name final themes (e.g., "Support for IA partnerships with caution," "Informational gaps in consent," "Advocacy for data sharing with protections").

• Integration and Dissemination:

  • Translate findings into concrete recommendations for policymakers, clinicians, and industry stakeholders.
  • Focus recommendations on areas such as bias management in research, informed consent processes, neural data sharing policies, and plans for long-term device upkeep [75].
  • Share results through peer-reviewed publications, stakeholder workshops, and regulatory briefs.

The Scientist's Toolkit

Table 3: Essential Research Reagents and Materials for aDBS Investigation

Item	Function/Application
Sensing-Enabled Implantable Pulse Generator (IPG) (e.g., Percept PC/RC)	The core device that delivers stimulation and simultaneously records local field potentials (LFPs) for closed-loop control [25].
DBS Leads (e.g., segmented directional leads)	Implanted in deep brain structures (STN/GPi); used for both delivering electrical stimulation and recording neural signals.
Clinician Programmer with aDBS Software	Interface for healthcare professionals to configure sensing parameters, set aDBS algorithms (e.g., dual threshold), and review neural data [74].
BrainSense Timeline	A feature that enables continuous, long-term recording of LFP data in the patient's home environment, crucial for understanding natural brain signal fluctuations [48].
Standardized Clinical Rating Scales (MDS-UPDRS III, MoCA)	Gold-standard tools for the quantitative assessment of motor and cognitive symptoms before and after intervention [74].
Patient-Reported Outcome Measures (e.g., PDQ-39, FOG-Q, Ecological Momentary Assessments)	Tools to capture the patient's perspective on symptom control, well-being, and quality of life in real-world settings [48] [74].

Visual Workflows and Signaling Pathways

aDBS Clinical Workflow

Ethics to Action Framework

Conclusion

Closed-loop DBS marks a significant advancement in the management of Parkinson's disease, offering a more physiological and personalized therapeutic approach. The synthesis of evidence confirms that aDBS provides sustained motor improvement, enhanced quality of life, and stable medication reduction, with the greatest benefit observed when implemented in the moderate stages of the disease. Future progress hinges on overcoming key technical challenges, including the refinement of biomarker detection beyond beta oscillations and the development of smarter, more automated devices. For biomedical and clinical research, the path forward involves validating multi-modal control signals, conducting large-scale randomized trials, and establishing standardized programming protocols to broaden access. The integration of AI and data-driven models will be crucial for creating fully personalized adaptive systems, ultimately solidifying aDBS as a cornerstone of precision neurology.

References

• 1. Chronic adaptive deep brain stimulation for Parkinson's ... [https://www.nature.com/articles/s41531-025-01124-7]
• 2. Closing the loop in DBS: A data-driven approach [https://www.sciencedirect.com/science/article/pii/S1353802025000896]
• 3. Alterations in beta oscillatory activity occur with the ... [https://www.sciencedirect.com/science/article/pii/S096999612500292X]
• 4. Oscillatory beta dynamics inform biomarker-driven ... [https://pubmed.ncbi.nlm.nih.gov/37198135/]
• 5. Coupling between beta band and high frequency ... [https://www.nature.com/articles/s41531-024-00656-8]
• 6. Simulation of Closed-Loop Deep Brain Stimulation Control ... [https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2020.00166/full]
• 7. Coupling between beta band and high frequency ... [https://pubmed.ncbi.nlm.nih.gov/38383550/]
• 8. Beta-Band Resonance and Intrinsic Oscillations in a ... [https://www.frontiersin.org/journals/computational-neuroscience/articles/10.3389/fncom.2019.00077/full]
• 9. Parkinson's Disease: Bridging Gaps, Building Biomarkers, and ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12346353/]
• 10. Understanding Parkinson's disease: current trends and its ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12488584/]
• 11. Longitudinal dynamics of clinical and neurophysiological ... [https://www.nature.com/articles/s41598-025-13558-3]
• 12. Machine learning for Parkinson's disease [https://www.nature.com/articles/s41531-025-01025-9]
• 13. Digital Biomarkers for Parkinson Disease: Bibliometric ... [https://www.sciencedirect.com/org/science/article/pii/S1438887125007186]
• 14. Personalized, closed-loop deep brain stimulation for ... [https://www.medrxiv.org/content/10.1101/2025.08.11.25333010v1.full]
• 15. Circuit Mechanisms of Parkinson's Disease [https://www.sciencedirect.com/science/article/pii/S0896627319302119]
• 16. Shared pathway-specific network mechanisms of ... [https://www.nature.com/articles/s41467-025-58825-z]
• 17. Dysfunctional Brain Dynamics of Parkinson's Disease and ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC8331088/]
• 18. Adaptive Parameter Modulation of Deep Brain Stimulation ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC8481598/]
• 19. Phase-dependent closed-loop deep brain stimulation of ... [https://www.sciencedirect.com/science/article/pii/S1935861X25001007]
• 20. Patient, target, device, and program selection for DBS in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12215258/]
• 21. Deep brain stimulation surgical timing, outcomes, and ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12342336/]
• 22. The safety profile of subthalamic nucleus and globus ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12006204/]
• 23. Electrophysiological signatures predict the therapeutic ... [https://www.nature.com/articles/s41746-025-02089-w]
• 24. Electrophysiological changes in the acute phase after deep ... [https://www.sciencedirect.com/science/article/pii/S1935861X25002992]
• 25. Long-Term Personalized Adaptive Deep Brain Stimulation ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12455485/]
• 26. Trial: Adaptive DBS Is Tolerable, Effective and Safe - Consult QD [https://consultqd.clevelandclinic.org/trial-adaptive-dbs-is-tolerable-effective-and-safe]
• 27. Closed-Loop Adaptive Deep Brain Stimulation in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC10357172/]
• 28. Real-time sweat-based L-dopa monitoring using flexible ... [https://www.sciencedirect.com/science/article/abs/pii/S1385894725113855]
• 29. Evaluation of Parkinson's Disease Motor Symptoms via ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12561371/]
• 30. Self-powered pressure sensor based on zinc–nickel ... [https://pubs.rsc.org/en/content/articlehtml/2025/ta/d5ta06807j]
• 31. On-Off and Proportional Closed-Loop Adaptive Deep Brain ... [https://www.sciencedirect.com/science/article/pii/S1094715923001526]
• 32. Five-Year Outcomes from Deep Brain Stimulation of the ... [https://pubmed.ncbi.nlm.nih.gov/40952750/]
• 33. UF-led study shows deep brain stimulation benefits for ... [https://ufhealth.org/news/2025/uf-led-study-shows-deep-brain-stimulation-benefits-for-parkinsons-patients]
• 34. Differential synaptic depression mediates the therapeutic ... [https://www.nature.com/articles/s41593-025-02088-w]
• 35. An algorithmic procedure for measuring deep brain ... [https://pubmed.ncbi.nlm.nih.gov/40909847/]
• 36. Practical Closed-Loop Strategies for Deep Brain Stimulation [https://pmc.ncbi.nlm.nih.gov/articles/PMC8716790/]
• 37. Advances in closed-loop deep brain stimulation devices [https://jneuroengrehab.biomedcentral.com/articles/10.1186/s12984-017-0295-1]
• 38. Decoding human thought, motion, and brain function with AI [https://www.eurekalert.org/news-releases/1106545]
• 39. Motor symptoms of Parkinson's disease: critical markers for ... [https://www.frontiersin.org/journals/aging-neuroscience/articles/10.3389/fnagi.2025.1602426/full]
• 40. Data Visualization with MATLAB [https://www.mathworks.com/discovery/data-visualization.html]
• 41. Matplotlib — Visualization with Python [https://matplotlib.org/]
• 42. Quantitative home-based assessment of Parkinson's symptoms [https://bmcneurol.biomedcentral.com/articles/10.1186/s12883-015-0343-z]
• 43. Subthalamic beta peak power ratio as an electrophysiological ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12570425/]
• 44. Online prediction of optimal deep brain stimulation contacts ... [https://www.nature.com/articles/s41531-025-01092-y]
• 45. Comparison of beta peak detection algorithms for data- ... [https://www.nature.com/articles/s41531-024-00762-7]
• 46. ID: 331466 Towards an Objective Beta Frequency Peak ... [https://www.sciencedirect.com/science/article/pii/S1094715924004951]
• 47. Sensing data and methodology from the Adaptive DBS ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC11408616/]
• 48. Chronic adaptive deep brain stimulation for Parkinson's ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12397205/]
• 49. Subthalamic beta peak power ratio as an electrophysiological ... [https://neurolrespract.biomedcentral.com/articles/10.1186/s42466-025-00441-9]
• 50. The Impact of Deep Brain Stimulation on Sleep in Parkinson's ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC7242854/]
• 51. Investigating the neural correlates of subjective sleep ... [https://www.sciencedirect.com/science/article/pii/S1353802025006285]
• 52. Periodic Artifact Removal With Applications to Deep Brain ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC9553325/]
• 53. Closed-Loop Adaptive Deep Brain Stimulation in Parkinson’s Disease: Procedures to Achieve It and Future Perspectives - PMC [https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10357172/]
• 54. Closed-Loop Adaptive Deep Brain Stimulation in Parkinson's Disease: Procedures to Achieve It and Future Perspectives - PubMed [https://pubmed.ncbi.nlm.nih.gov/37182899/]
• 55. Surgical management of adverse events associated with ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC7082866/]
• 56. Adverse events in deep brain stimulation: A retrospective long ... [https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0178984]
• 57. Will adaptive deep brain stimulation for Parkinson’s disease become a real option soon? A Delphi consensus study | npj Parkinson's Disease [https://www.nature.com/articles/s41531-025-00974-5]
• 58. Patient perspectives: The balancing act of trial participation [https://www.medable.com/knowledge-center/blog-patient-perspectives-the-balancing-act-of-trial-participation]
• 59. Site engagement for multi-site clinical trials - PMC [https://pmc.ncbi.nlm.nih.gov/articles/PMC7358177/]
• 60. Lasting relief of Parkinson's motor symptoms found with DBS [https://parkinsonsnewstoday.com/news/lasting-relief-parkinsons-motor-symptoms-found-dbs-us-study/]
• 61. DBS sustains motor improvements in Parkinson's for 5 years [https://www.icthealth.org/news/dbs-sustains-motor-improvements-in-parkinsons-for-5-years]
• 62. Adaptive DBS: A New Era in Parkinson's Disease Treatment [https://www.apdaparkinson.org/article/adaptive-deep-brain-stimulation-dbs/]
• 63. A Randomized Crossover Trial with Exploratory Analysis of ... [https://www.medrxiv.org/content/10.1101/2025.09.07.25335282v1.full]
• 64. Smart Brain Stimulation Shows Promise for Parkinson's ... [https://www.psychiatrist.com/news/smart-brain-stimulation-shows-promise-for-parkinsons-patients/]
• 65. Sensing data and methodology from the Adaptive DBS ... [https://www.nature.com/articles/s41531-024-00772-5]
• 66. Closed-Loop Deep Brain Stimulation Platform for ... [https://www.sciencedirect.com/science/article/abs/pii/S1094715924012248]
• 67. Levodopa / opicapone as a complement to STN-DBS in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC11488416/]
• 68. Antiparkinsonian Drug Reduction After Directional Versus ... [https://www.sciencedirect.com/science/article/pii/S1094715922000265]
• 69. An update on best practice of deep brain stimulation in ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC6440024/]
• 70. Safinamide as an adjunct to levodopa monotherapy in ... [https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2025.1591664/full]
• 71. Ethical Governance Strategies for the Responsible ... [https://link.springer.com/article/10.1007/s11673-025-10440-9]
• 72. Ethics of neurotechnology: UNESCO adopts the first global ... [https://www.unesco.org/en/articles/ethics-neurotechnology-unesco-adopts-first-global-standard-cutting-edge-technology]
• 73. Ethical gaps in closed-loop neurotechnology: a scoping ... [https://www.nature.com/articles/s41746-025-01908-4]
• 74. Chronic Adaptive Deep Brain Stimulation in Parkinson's ... [https://www.medrxiv.org/content/10.1101/2025.09.25.25336275v1.full-text]
• 75. Ethical implications of neurotechnology in industry ... [https://pmc.ncbi.nlm.nih.gov/articles/PMC12404399/]