Movement-responsive adaptive deep brain stimulation

Abstract

Devices, systems, software, and methods are provided for treating bradykinesia with deep brain stimulation (DBS). In particular, DBS is performed with a neural recording device that records brain electrical signals from neural activity associated with intended movement of the subject and automatically adjusts deep brain stimulator settings and / or delivers electrical stimulation to the brain when pre-specified patterns of neural activity associated with the intended movement are detected. Machine learning computational models are used to detect and classify patterns of neural activity associated with intended movement. The neural signatures of "intended movement" are used to assist with DBS programming to determine therapeutic stimulation parameters that boost movement when a patient starts to prepare for movement and during movements but stop when the patient is resting.

Description

MOVEMENT-RESPONSIVE ADAPTIVE DEEP BRAIN STIMULATIONCROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims benefit of U.S. Provisional Patent Application No. 63 / 531 ,506, filed August 8, 2023, which application is incorporated herein by reference in its entirety.BACKGROUND OF THE INVENTION

[0002] Parkinson’s disease (PD) is a common neurodegenerative movement disorder characterized by motor symptoms (e.g., bradykinesia, rest tremor, rigidity, postural instability) that are highly disabling and significantly impair quality of life (Fasano, A. et al. BMC Neurol. 19, 1 -1 1 (2019); Sanchez-Luengos et al. Qual. Life Res. 31 , 3241 -3252 (2022); Bloem, B. R., Okun, M. S. & Klein, C. Parkinson’s disease. Lancet 397, 2284-2303 (2021 )). Deep brain stimulation (DBS) has grown to widespread use as an effective treatment for advanced Parkinson’s Disease (PD) when medication alone is ineffective. Conventional DBS (eDBS) treats motor symptoms by delivering electrical stimulation to the basal ganglia at a fixed current and frequency. Notably, however, this doesn’t account for the dynamic clinical needs of the patient, which fluctuate across a range of timescales. These fluctuations arise from sources including medication wash-in / out, activity level, and physiological motor state. The unresponsive nature of conventional DBS can result in involuntary movements (dyskinesias) when stimulation is too high with respect to the current clinical state, or slowness of movement, the cardinal symptom of PD, when stimulation is too low.

[0003] Thus, there remains a need for better methods of delivering DBS to patients with movement disorders such as Parkinson’s disease to relieve motor symptoms and provide assistance with voluntary movement.SUMMARY OF THE INVENTION

[0004] Devices, systems, software, and methods are provided for treating bradykinesia with DBS. In particular, DBS is performed with a neural recording device that records brain electrical signals from neural activity associated with intended movement and automatically adjusts deep brain stimulator settings and / or delivers electrical stimulation to the brain when pre-specified patterns of neural activity associated with intended movement are detected. Machine learning computational models are used to detect and classify patterns of neural activity associated with intended movement. The neural signatures of “intended movement” are used to assist with DBS programming to determine therapeutic stimulation parameters that boost movement when a patient starts toprepare for movement and during movements but stop when the patient is resting. The subject methods are designed to counteract Parkinsonian slowness during movement and prevent excessive movement (dyskinesia) during rest. The inventors have demonstrated proof of principle of a fully automated, end-to-end, machine learning supported, data driven algorithm for parameterization of adaptive DBS, which was shown to increase naturalistic movement speed and reduce dyskinesia side effects and total electrical stimulation needed for a patient with Parkinson’s disease. The devices and methods can be used remotely and self-directed by patients in their own homes.

[0005] In one aspect, a method for treating bradykinesia in a subject using deep brain stimulation is provided, the method comprising: positioning a stimulation electrode at a first location in a subthalamic nucleus region or a second location in a globus pallidus internus region of the brain of the subject to deliver electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region; positioning a first neural recording electrode at a third location in the subthalamic nucleus region of the brain of the subject and a second neural recording electrode at a fourth location in a contralateral cortex region of the brain of the subject to record brain electrical signal data associated with an intended movement of the subject; detecting a brain electrical signal associated with the intended movement using the first neural recording electrode or the second neural recording electrode; and applying electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the stimulation electrode in a manner effective to provide a prokinetic effect on the intended movement of the subject when the brain electrical signal associated with the intended movement is detected using the first neural recording electrode or the second neural recording electrode.

[0006] In certain embodiments, the subject has Parkinson’s disease, Parkinsonism, Lewy body dementia, progressive supranuclear palsy, corticobasal degeneration, Huntington's disease, dystonia, stroke, or multiple system atrophy.

[0007] In certain embodiments, the contralateral cortex region comprises a precentral gyrus region, a postcentral gyrus region, or both the precentral gyrus region and the postcentral gyrus region.

[0008] In certain embodiments, the contralateral cortex region comprises the central sulcus.

[0009] In certain embodiments, the electrical stimulation reduces bradykinesia and increases vigor of the intended movement compared to in absence of the electrical stimulation.

[0010] In certain embodiments, the electrical stimulation is applied unilaterally or bilaterally.

[0011] In certain embodiments, the brain electrical signal comprises alpha frequency, beta frequency, gamma frequency, or theta frequency neural oscillations.

[0012] In certain embodiments, the theta frequency neural oscillations are in a range from 4 Hz to 7 Hz, the alpha frequency neural oscillations are in a range from 8 Hz to 12 Hz, the beta frequency neural oscillations are in a range from 13 Hz to 30 Hz, and the gamma frequency neural oscillations are in a range from 60 Hz to 90 Hz.

[0013] In certain embodiments, the brain electrical signal data comprises field potential data.

[0014] In certain embodiments, the method further comprises using a control algorithm to automate said applying electrical stimulation when the brain electrical signal associated with the intended movement is detected.

[0015] In certain embodiments, the control algorithm uses a machine learning algorithm for movement classification. In some embodiments, the machine learning algorithm is a supervised machine learning algorithm.

[0016] In certain embodiments, the movement classification distinguishes between when the subject is in an intended movement state or a stationary state.

[0017] In certain embodiments, the method further comprises training a machine learning model to predict whether the subject is in an intended movement state using accelerometry data for detected movement of the subject. In some embodiments, the accelerometry data is from a wearable accelerometer worn by the subject that detects natural undirected movement of the subject. In some embodiments, the accelerometry data is from a wearable accelerometer worn by the subject that detects movement during directed movement tasks selected from finger tapping, opening or closing of a hand, wrist pronation or supination, walking, and keyboard typing.

[0018] In certain embodiments, the method further comprises performing principal components (PC) decomposition on power spectral data to identify personalized power band features for distinguishing the intended movement state from the stationary state.

[0019] In certain embodiments, the method, further comprises: transforming a time-domain of a local field potential (LFP) into a spectrogram using a short-time Fourier transform (STFT); performing principal components analysis (PCA) on z-scored STFT outputs to identify coordinated bands of neural activity; and applying a peak-finding algorithm to PCA component weights to find a frequency range best approximating each principal component. In some embodiments, the top principal components for the contralateral cortex region or the subthalamic nucleus region correspond to personalized frequency ranges for beta frequency neural oscillations and gamma frequency neural oscillations for the brain electrical signal associated with the intended movement of the subject.

[0020] In certain embodiments, the electrical stimulation is selectively increased during the intended movement state and not during the stationary state.

[0021] In certain embodiments, the method further comprises using accelerometry data, gyroscope data, multi-view video recordings of the subject, keylogging data from a computer used by the subject, or any combination thereof in combination with the recorded brain electrical signal data to distinguish between the intended movement state and the stationary state.

[0022] In certain embodiments, a wearable device worn by the subject is used to collect the accelerometry data, the gyroscope data, or a combination thereof. In some embodiments, the wearable device is a smartwatch.

[0023] In certain embodiments, the control algorithm further uses linear discriminant analysis (LDA) to adjust stimulation amplitude or frequency of the electrical stimulation.

[0024] In certain embodiments, the intended movement is a finger tap, opening or closing of a hand, wrist pronation or supination, walking, or keyboard typing.

[0025] In certain embodiments, the electrical stimulation is optimized for each hand of the subject independently.

[0026] In certain embodiments, the stimulation electrode is placed on a surface of the subthalamic nucleus region or the globus pallidus internus region.

[0027] In certain embodiments, the first neural recording electrode is placed on a surface of the subthalamic nucleus region.

[0028] In certain embodiments, the second neural recording electrode is placed on a surface of the contralateral cortex region.

[0029] In certain embodiments, the stimulation electrode is a non-brain penetrating surface electrode array or a brain-penetrating electrode array.

[0030] In certain embodiments, the first neural recording electrode and / or the second neural electrode is a non-brain penetrating surface electrode array or a brain-penetrating electrode array.

[0031] In certain embodiments, the first neural recording electrode and / or the second neural electrode is an electroencephalogram (EEG) electrode array, a subgaleal or burrhole mounted or cranially mounted neurostimulator electrode, or an electrocorticogram (ECoG) electrode array. In some embodiments, the ECoG electrode array spans regions of the precentral gyrus and the postcentral gyrus region that control hand and arm movement. In some embodiments, the ECoG electrode array spans the central sulcus.

[0032] In certain embodiments, the subject is further administered dopaminergic medication.

[0033] In certain embodiments, the method further comprises assessing effectiveness of the treatment in the subject. In some embodiments, the assessing comprises using behavioral data obtained of the subject. In some embodiments, the behavior data is accelerometry data, gyroscopedata, video-based pose kinematic data for the subject, or keylogging data from a computer used by the subject, or a combination thereof. In some embodiments, the assessing comprises using a Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) or a Hoehn and Yahr (HnY) scale, a Parkinson’s Disease Composite Scale (PDCS), or a Schwab and England Activities of Daily Living (ADL) Scale.

[0034] In certain embodiments, the method is performed at a site remote from a hospital. In some embodiments, the method is performed at the subject’s home.

[0035] In another aspect, a computer implemented method for programming a deep brain stimulator to treat bradykinesia in a subject is provided, the computer performing steps comprising: receiving recorded brain electrical signal data from a contralateral cortex region and a subthalamic nucleus region of the brain of the subject; analyzing the recorded brain electrical signal data using a movement classification model that identifies patterns of electrical signals in the recorded brain electrical signal data associated with intended movement of the subject; adjusting one or more programmed stimulation parameters based on the recorded brain electrical signal data according to an algorithm control law; and instructing the deep brain stimulator to apply an electrical stimulation to the subthalamic nucleus region or a globus pallidus internus region of the brain to provide a prokinetic effect on the intended movement of the subject.

[0036] In certain embodiments, the computer implemented method uses a machine learning algorithm to generate the movement classification model. In some embodiments, the machine learning algorithm is a supervised machine learning algorithm. In some embodiments, the movement classification distinguishes between when the subject is in an intended movement state or a stationary state.

[0037] In certain embodiments, the computer implemented method further comprises training a machine learning model to predict whether the subject is in an intended movement state using accelerometry data for detected movement of the subject. In some embodiments, the accelerometry data is from a wearable accelerometer worn by the subject that detects natural undirected movement of the subject. In some embodiments, the accelerometry data is from a wearable accelerometer worn by the subject that detects movement during directed movement tasks selected from finger tapping, opening or closing of a hand, wrist pronation or supination, walking, and keyboard typing.

[0038] In certain embodiments, the brain electrical signal data used by the computer implemented method comprises field potential data.

[0039] In certain embodiments, the computer implemented method further comprises performing principal components (PC) decomposition on power spectral data to identify personalized power band features for distinguishing the intended movement state from the stationary state.

[0040] In certain embodiments, the computer implemented method further comprises: transforming a time-domain of a local field potential (LFP) into a spectrogram using a short-time Fourier transform (STFT); performing principal components analysis (PCA) on z-scored STFT outputs to identify coordinated bands of neural activity; and applying a peak-finding algorithm to PCA component weights to find a frequency range best approximating each principal component. In some embodiments, the top principal components for the contralateral cortex region or the subthalamic nucleus region correspond to personalized frequency ranges for beta frequency neural oscillations and gamma frequency neural oscillations for the brain electrical signal associated with the intended movement of the subject.

[0041] In certain embodiments, the electrical stimulation is selectively increased during the intended movement state and not during the stationary state.

[0042] In certain embodiments, the computer implemented method further comprises using accelerometry data, gyroscope data, multi-view video recordings of the subject, keylogging data from a computer used by the subject, or any combination thereof in combination with the recorded brain electrical signal data to distinguish between the intended movement state and the stationary state.

[0043] In certain embodiments, the computer implemented method further comprises: a) ranking predicted stimulation effectiveness for available settings of a DBS device based on classifier scores for stimulation effectiveness of each setting using a linear classification model; b) selecting stimulation settings predicted to have highest stimulation effectiveness based on the linear classification model; c) receiving recorded brain electrical signal data from the contralateral cortex region and the subthalamic nucleus region of the brain of the subject after applying electrical stimulation with the DBS device to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the settings predicted to have the highest stimulation effectiveness; d) analyzing the recorded brain electrical signal data to evaluate neural response of the subject to the electrical stimulation; e) updating the linear classification model based on the neural response of the subject to the electrical stimulation to generate an updated linear classification model; f) updating the ranking of predicted stimulation effectiveness for the available settings of the DBS device using the updated linear classification model; g) selecting stimulation settings predicted to have the highest stimulation effectiveness based on the updated linear classification model; h) receiving recorded brain electrical signal data from the contralateral cortex region and thesubthalamic nucleus region of the brain of the subject after applying the electrical stimulation with the DBS device to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the settings predicted to have the highest stimulation effectiveness based on the updated linear classification model; and i) repeating e) - h) to adjust the available settings of the DBS device to optimize stimulation effectiveness.

[0044] In certain embodiments, the linear classification model uses linear discriminant analysis (LDA) to adjust amplitude of current and frequency of the electrical stimulation.

[0045] In another aspect, a non-transitory computer-readable medium is provided, the non-transitory computer-readable medium comprising program instructions that, when executed by a processor in a computer, causes the processor to perform a computer implemented method described herein.

[0046] In another aspect, a kit comprising the non-transitory computer-readable medium, described herein, and instructions for determining symptom severity of a subject having a movement disorder.In another aspect, a system for treating bradykinesia in a subject is provided, the system comprising: a stimulation electrode adapted for positioning at a first location in a subthalamic nucleus region or a second location in a globus pallidus internus region of the brain of the subject to deliver electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region; a first neural recording electrode adapted for positioning at a third location in the subthalamic nucleus region of the brain of the subject to record brain electrical signal data; a second neural recording electrode adapted for positioning at a fourth location in a contralateral cortex region of the brain of the subject to record brain electrical signal data; and a processor programmed according to the computer implemented method, described herein, to instruct the stimulation electrode to apply an electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject in a manner effective to provide a prokinetic effect on the intended movement of the subject when the brain electrical signal associated with the intended movement is detected using the first neural recording electrode or the second neural recording electrode.

[0047] In certain embodiments, the system further comprises an accelerometer, a gyroscope, or a combination thereof. In some embodiments, the accelerometer, the gyroscope, or the combination thereof is provided by a wearable device. In some embodiments, the wearable device is a smartwatch.

[0048] In certain embodiments, the system further comprises a video recording device.

[0049] In certain embodiments, the stimulation electrode is a non-brain penetrating surface electrode array or a brain-penetrating electrode array.

[0050] In certain embodiments, the first neural recording electrode or the second neural electrode is a non-brain penetrating surface electrode array or a brain-penetrating electrode array.

[0051] In certain embodiments, the first neural recording electrode or the second neural electrode is an electroencephalogram (EEG) electrode array, a subgaleal or burrhole mounted or cranially mounted neurostimulator electrode, or an electrocorticogram (ECoG) electrode array.

[0052] In certain embodiments, the ECoG electrode array spans regions of the precentral gyrus and the postcentral gyrus region that control hand and arm movement.

[0053] In certain embodiments, the ECoG electrode array spans the central sulcus.

[0054] In certain embodiments, the system further comprises a user interface comprising an input electronically coupled to the processor for instructing the stimulation electrode to apply an electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region to treat the bradykinesia in the subject.

[0055] In certain embodiments, the user interface is password protected and is operable by a health care practitioner.BRIEF DESCRIPTION OF THE DRAWINGS

[0056] FIGS. 1A-1C. Movement-responsive DBS theoretical framework. (FIG. 1A) In the neurologically healthy state (left) the basal ganglia inhibit actions that have not yet been selected through activation beyond a gating threshold (black dashed line). When activation associated with a particular action reaches this threshold, movement is released with the appropriate timing and vigor (subthreshold activation is represented as a light gray line, which becomes dark gray to indicate suprathreshold activation / selection and downstream movement). In the Parkinsonian state (middle), excessive and fluctuating inhibition in the basal ganglia raises the movement threshold. This results in difficulty initiating vigorous movements. In the treated Parkinsonian state (medication / stimulation, right), the movement threshold is lowered through tonic disinhibition without regard for motor intent. This can result in unintended movements (red asterisks) when activation exceeds the reduced, fluctuating threshold. (FIG. 1 B) A “Movement-responsive” framework is proposed where stimulation is tied to motor intent, reaching higher levels during intended movement and reducing to lower levels at rest. The far right plot illustrates how this prevents dyskinesia when movement is not intended while providing targeted disinhibition for vigorous, volitional movement. (FIG. 1C) To enable movement- responsive experiments, neural data were streamed from bilateral STN and cortical electrodes in a participant implanted with the Medtronic Summit RC&S DBS system. External devices for streaming data from the implants were worn over the upper chest; however, adaptive stimulation was performed,embedded on the device.. Video from three cameras and wrist-worn accelerometry (Apple Watches) were simultaneously recorded and time-synced (note that the camera placement in the cartoon image is for illustration only and not indicative of the precise locations and orientations used in the study).

[0057] FIGS. 2A-2C. Optimizing the RC&S movement classifier. (FIG. 2A) Six days of data were collected for model training, testing, and validation. During each session the participant performed a series of self-guided clinical tasks: rest recording (assessing tremor / dyskinesia), rhythmic finger tapping, hand open-close, wrist pronation-supination, and typing (pictured in order from left to right). The participant alternated between hands such that one hand was always resting while the other performed the tasks, except for when typing bimanually. (FIG. 2B) Wrist-worn accelerometry was recorded during behavior, and LFP’s were simultaneously recorded from three brain regions: the STN and the Postcentral and Precentral gyri. A three minute example window is displayed showing contralateral wrist acceleration alongside beta (13-30Hz) and gamma (60-90Hz) power bands recorded from the PreCentral gyrus (PreC). A threshold was applied to the wrist accelerometry (red line) to label binary moving and non-moving states. Movement periods are indicated by orange shading, with movement tasks indicated by symbols corresponding to the cartoons in (FIG. 2A). (FIG. 2C) An automated pipeline for feature selection and parameter optimization was designed for programming the RC+S to optimally classify movement state. Prior to parameter optimization, two sets of candidate power bands were identified that would later be used as stimulation control signals: “Personalized” bands, computed using a modified form of principal components decomposition, and “Canonical” bands, defined by traditional frequency ranges used in existing literature. A progressive narrowing of each candidate pool began with creating separate models for every possible combination of power bands and feeding each through a brief (20 iteration) Bayesian optimization routine (“Broad search”). The top three models underwent an additional 200 iterations in a “Final optimization” stage, after which the single top performer was selected for each hemisphere.

[0058] FIGS. 3A-3D. Personalizing power bands improves algorithm performance. (FIG. 3A) To obtain Personalized power bands, principal component (PC) models were fit to spectrogram data from an independent dataset where the participant was freely behaving in their home for 2 hours. The PC weights (displayed for the top two PC’s of each recording channel) were used with a peak-finding algorithm to identify contiguous frequency ranges that approximated each PC in a way that was compatible with the device-embedded controller. Note that frequencies below 4Hz and above 100Hz were excluded due to known noise issues. (FIG. 3B) The top four personalized power bands were selected for each of the three recording locations, resulting in 12 unique biomarkers per hemisphere. The frequency ranges for each Personalized power band are displayed for comparison with Canonicalpower bands, with the color indicating the PC number (blue: PCO, orange: PC1 , green: PC2, red: PC3; colors arbitrary for the canonical bands). (FIG. 3C) The marginal performance contribution for the top four power bands (ranked by mean) are displayed for each power band pool. (FIG. 3D) The top 200 models from each of the biomarker pools (Canonical and Personalized) were directly compared to determine if the Personalized power bands provided significant performance benefits as compared to the Canonical power bands. This performance difference relates only to the choice of input power bands. Parameter optimization was performed identically for all models from both pools, therefore isolating the additive impact of power band personalization on top of the performance gains from the rest of the machine learning pipeline. Vertical dotted lines indicate the mean of each distribution. The top Personalized power band models outperformed their matched-rank Canonical models by an average of 6% (left hemisphere) and 2% (right). Asterisks indicate statistical significance at p<0.05 (permutation test - difference in means).

[0059] FIGS. 4A-4D. Final offline model performance. (FIG. 4A) An example time window of decoder predictions is displayed with the true acceleration (top row), the predicted acceleration (middle row) and an overlay of the true (blue) and predicted (orange) movement states (bottom row). A notable false positive is indicated with red arrows on the right hemisphere plots, highlighting a common observation: even when the neural classifier incorrectly predicted movement, there was often observable sub-threshold acceleration. (FIG. 4B) Confusion matrices are displayed for decoder predictions from a held-out dataset. (FIG. 4C) Distributions of the accelerometry during predicted movement are compared to predicted stationary periods to further assess the discrimination performance of the neural decoder. Acceleration of the left hand (right hemisphere) notably lacked strong bimodality, making binary movement state predictions challenging. (FIG. 4D) The moment-by- moment stimulation current was simulated according to the predictions of the neural classifier and the stimulation parameters that were used in the later online adaptive experiments. The proportion of time spent at each stimulation amplitude is plotted for each true movement state of the offline test data.

[0060] FIGS. 5A-5D. Performance of neural classifiers is maintained during online adaptive stimulation. (FIG. 5A) For evaluating movement-responsive DBS, three test conditions were used: “Movement responsive” increased stimulation during sensed movement, “Inverted” decreased stimulation during sensed movement, and “Constant” kept stimulation at an intermediate level. (FIG. 5B) Example data from the Movement responsive block of a single session displays the wearable accelerometry (top row) and neural predictions of the accelerometry (second row) together with the corresponding discrete movement states (third row) and the responsive stimulation adjustments (bottom row). (FIG. 5C) Confusion matrices depict the state-specific prediction accuracy from theneural classifier averaged across conditions. (FIG. 5D) The proportion of time at the low, high, and intermediate stimulation amplitudes was computed separately for each movement state and test condition. The goal of the algorithm was to maximize the time spent at high stim when in the moving state, while minimizing the time spent at high stim when in the stationary state (and vice-versa for the Inverted condition).

[0061] FIGS. 6A-6E. Movement responsive stimulation impacts self-perceived therapeutic quality, movement speed, and dyskinesia in an accuracy-dependent manner. (FIG. 6A) Blinded self-scores were reported by the subject at the conclusion of all 12 sessions to quantify the participant’s selfperceived therapeutic quality of each stimulation condition. The order in which the conditions were performed in each session was randomized in balanced fashion and was included as a covariate in all across-condition statistical testing. (FIG. 6B) The impact of classifier accuracy on therapeutic outcomes was assessed by regressing self-scores on the F1 score on a per-session basis. A single self-score was provided for both hands; the reported F1 score represents the average over the two hemispheric classifiers. Only data from the Movement responsive condition are shown. (FIG. 6C) As a control analysis to determine if therapeutic outcomes were simply better with higher average stimulation (e.g., from classifier biases), self-scores were regressed on mean stimulation amplitude across the two hemispheric stimulators on a per-session basis. (FIG. 6D-6E) Quantifiable features of movement from representative motor tasks are shown independently for each hand, with the left bar and datapoints in each pair representing the left hand. (FIG. 6D) Movement speed was assessed using the repetition rate during two repetitive motor tasks: wrist rotations (left) and nose tapping (right). Note that the right hemisphere (left hand) classifier displayed a bias towards predicting stationarity during movement. (E) Dyskinesia was assessed during the bimanual rest period where movement was assumed unintentional. Low-frequency power in the wrist-worn accelerometry data was used to quantify dyskinesia. Note that both hemispheric classifiers were highly accurate at predicting the stationary state. P-values were determined by one-way (FIG. 6A) and two-way (FIGS. 6D-6E) ANCOVA, and Pearson correlation (FIGS. 6B-6C). Asterisks indicate statistical significance at p<0.05.

[0062] FIGS. 7A-7C. Movement adaptive stimulation increases keypress responsiveness during naturalistic typing. Three metrics of typing performance were evaluated across stimulation conditions. (FIG. 7A) Movement responsive stimulation resulted in a significant decrease in mean keypress duration relative to both of the other conditions. (FIG. 7B) Faster individual keypresses translated to an overall increase in typing speed when comparing only the Movement Responsive and Invertedconditions. (FIG. 7C) No significant differences in backspace rate (errors) were observed. KP: keypress; BS: backspace. Asterisks indicate statistical significance at p<0.05 (one-way ANCOVA).

[0063] FIG. 8. Decoder stability. Decoder performance and mean stimulation during the movement- responsive blocks are displayed. Each datapoint represents the average over the two hemispheres. Two manual updates were made to the threshold parameter of the algorithm due to sudden signal scaling: once for the left hemisphere on the fifth recording session 384 days after completion of the training data collection, and once for the right hemisphere on the tenth recording session 429 days after completion of the training data collection. Note that the mean stimulation is the result of both decoder biases and actual time spent in each movement state, and should therefore not be interpreted as a pure depiction of decoder performance or bias (refer to accuracy and F1 score for these two assessments, respectively).

[0064] FIG. 9. Condition-specific decoder performance. Confusion matrices showing decoder performance in each of the three stimulation conditions are displayed. Decoder performance did not vary substantially across the stimulation conditions despite their differential effects on behavior and stimulation artifacts.

[0065] FIG. 10. Effects of movement-responsive stimulation on resting tremor. Tremor was quantified using power in the 4-7 Hz range of the wrist-worn accelerometers during rest behavior. The same results were observed as in the analysis of dyskinesia during the same periods (FIG. 6), with the Movement responsive condition having less tremor than both other conditions. The observed trends likely reflect spectral bleed from the dyskinesia band (1-4 Hz), as the movements appeared to be dyskinesia rather than tremor upon visual inspection of the video, and the on-average lower stimulation during rest in the Movement responsive condition would not be expected to ameliorate tremor.

[0066] FIGS. 11A-11G. (FIG. 11A) Monitoring naturalistic motion during DBS. We collected 610 hours of neural data from 15 patients with idiopathic Parkinson’s disease (PD) while they performed activities of daily living in naturalistic settings. (FIG. 11 B) Cohen’s d-gram averaged across 15 patients. We averaged the Cohen’s d-grams from both hemispheres for patients with bilateral recordings. The resulting group-level, cluster-based permutation test-corrected Cohen’s d-gram indicated a decrease in cortical and subcortical beta and an increase in broadband gamma for mobile versus stationary states. (FIG. 11C) Single subject flattened STN power spectral densities. (FIG. 11 D) Effect of stimulation level on STN / Gpi movement biomarkers. In the STN / GPi, the magnitude of low beta (LMM; p = -0.080; p < 0.0001 ) and high beta (LMM; p = -0.108; p < 0.0001 ) MRD decreased with stimulation. Similarly, the magnitudes of low gamma (LMM; p = -0.040; p < 0.001) and highgamma (LMM; p = -0.040; p < 0.0001 ) MRS were negatively correlated with. An example of this for a single hemisphere is shown in FIG. 1 1 C. In the cortical regions, stimulation amplitude had no significant effect on high beta MRD or gamma MRS. There was, however, a substantial decrease in the magnitude of alpha (S1 : LMM; p = -0.041 ; p < 0.01 ) (M1 : LMM; p = -0.032; p < 0.05) and low beta (Site: S1 ; LMM; p = -0.061 ; p < 0.0001 ) (Site: M1 ; LMM; p = -0.053; p < 0.05) MRD. (FIG. 11 E) Movement state classifier performance. We developed linear models from one or all intracranial regions using five canonical PBs (alpha, low beta, high beta, low gamma, and high gamma). We then optimized the FFT window via five-fold cross-validation and investigated using up to 15 principal components (PCs) as features. The models trained on data from all sites had the highest performance (p < 0.05), followed by those trained only on signals from the S1 (p < 0.01). (FIG. 11 F) Permutation feature importance of classifiers. Permutation feature importance revealed differences in predictive power between the four sites (One-way ANOVA; F= 5.1 , p = 2.7e-8). S1 (Mean change in r statistic ± SEM = -0.08 ± 0.02) and M1 (Mean change in r statistic ± SEM = -0.07 ± 0.02) high beta attained the greatest mean feature importance. We also investigated the use of non-linear models, but they did not produce significant differences in performance compared to the linear models (One-sided Wilcoxon signed rank test; p = 0.0009). (FIG. 11G) Effect of stimulation level on classifier performance. We found our single-site and combined models stable over time, with no significant effect of duration between sessions on model performance. We also evaluated differences in model performance at different stimulation levels. We observed that the STN / GPi classifiers (LMM; p = - 0.047; p < 0.001 ) and regressors (LMM; p = -0.054; p < 0.001 ) were negatively correlated with stimulation amplitude. However, the models trained on S1 and M1 data were not significantly affected by variations in the DBS amplitude.DETAILED DESCRIPTION OF THE INVENTION

[0067] Devices, systems, software, and methods are provided for treating bradykinesia with deep brain stimulation (DBS). In particular, DBS is performed with a neural recording device that records brain electrical signals from neural activity associated with intended movement of the subject and automatically adjusts deep brain stimulator settings and / or delivers electrical stimulation to the brain when pre-specified patterns of neural activity associated with the intended movement are detected. Machine learning computational models are used to detect and classify patterns of neural activity associated with intended movement. The neural signatures of “intended movement” are used to assist with DBS programming to determine therapeutic stimulation parameters that boost movementwhen a patient starts to prepare for movement and during movements but stop when the patient is resting.

[0068] Before the present devices, systems, software, and methods are described, it is to be understood that this invention is not limited to the particular devices, systems, software, and methods described, as such may, of course, vary. It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only, and is not intended to be limiting, since the scope of the present invention will be limited only by the appended claims.

[0069] Where a range of values is provided, it is understood that each intervening value, to the tenth of the unit of the lower limit unless the context clearly dictates otherwise, between the upper and lower limits of that range is also specifically disclosed. Each smaller range between any stated value or intervening value in a stated range and any other stated or intervening value in that stated range is encompassed within the invention. The upper and lower limits of these smaller ranges may independently be included or excluded in the range, and each range where either, neither or both limits are included in the smaller ranges is also encompassed within the invention, subject to any specifically excluded limit in the stated range. Where the stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included in the invention.

[0070] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, some potential and preferred methods and materials are now described. All publications mentioned herein are incorporated herein by reference to disclose and describe the methods and / or materials in connection with which the publications are cited. It is understood that the present disclosure supersedes any disclosure of an incorporated publication to the extent there is a contradiction.

[0071] As will be apparent to those of skill in the art upon reading this disclosure, each of the individual embodiments described and illustrated herein has discrete components and features which may be readily separated from or combined with the features of any of the other several embodiments without departing from the scope or spirit of the present invention. Any recited method can be carried out in the order of events recited or in any other order which is logically possible.

[0072] It must be noted that as used herein and in the appended claims, the singular forms "a", "an", and "the" include plural referents unless the context clearly dictates otherwise. Thus, for example, reference to "an electrode" includes a plurality of such electrodes and reference to "the electrical signal" includes reference to one or more electrical signals, and so forth.

[0073] The publications discussed herein are provided solely for their disclosure prior to the filing date of the present application. Nothing herein is to be construed as an admission that the present invention is not entitled to antedate such publication by virtue of prior invention. Further, the dates of publication provided may be different from the actual publication dates which may need to be independently confirmed.Definitions

[0074] The term "about," particularly in reference to a given quantity, is meant to encompass deviations of plus or minus five percent.

[0075] The term "movement disorder" refers to any type of neurological disorder that causes either increased movements or reduced or slow movements. Movement disorders include, but are not limited to, Parkinson's disease, parkinsonism, progressive supranuclear palsy, ataxia, cervical dystonia, chorea, dystonia, functional movement disorder, Huntington's disease, multiple system atrophy, myoclonus, tardive dyskinesia, Tourette syndrome, tremor, restless legs syndrome, and Wilson's disease. Symptoms may include, but art not limited to, tremor, involuntary movements, slowness of movement (bradykinesia), rigidity, postural instability, twisting movements, poor balance, irregularity of movements, stumbling, and difficulty with walking. In some cases, a movement disorder is caused by genetic and / or environmental factors, head trauma, infections, inflammation, metabolic disturbances, toxins, adverse reactions to medications, or stressful life events.

[0076] The terms “individual”, “subject”, “recipient”, and “patient” are used interchangeably herein and refer to any mammalian subject for whom treatment or therapy is desired, particularly humans. "Mammal" for purposes of treatment refers to any animal classified as a mammal, including human and non-human mammals such as non-human primates, including chimpanzees and other apes and monkey species; laboratory animals such as mice, rats, rabbits, hamsters, guinea pigs, and chinchillas; domestic animals such as dogs and cats; and farm animals such as sheep, goats, pigs, horses and cows.

[0077] The term “user” as used herein refers to a person that interacts with a device and / or system disclosed herein for performing one or more steps of the presently disclosed methods. The user may be the patient receiving treatment for bradykinesia. The user may be a health care practitioner, such as the patient’s physician.

[0078] The terms "treatment", "treating", "treat" and the like are used herein to generally refer to obtaining a desired pharmacologic and / or physiologic effect. The effect can be prophylactic in termsof completely or partially preventing a disease or symptom(s) thereof and / or may be therapeutic in terms of a partial or complete stabilization or cure for a disease and / or adverse effect attributable to the disease. The term “treatment" encompasses any treatment of a disease in a mammal, particularly a human, and includes: (a) preventing the disease and / or symptom(s) from occurring in a subject who may be predisposed to the disease or symptom but has not yet been diagnosed as having it; (b) inhibiting the disease and / or symptom(s), i.e., arresting their development; or (c) relieving the disease symptom(s), i.e., causing regression of the disease and / or symptom(s). Those in need of treatment include those already inflicted (e.g., those with bradykinesia) as well as those in which prevention is desired those with a genetic predisposition to developing a movement disorder, those with increased susceptibility to developing a movement disorder, those suspected of having a movement disorder, etc.).

[0079] A therapeutic treatment is one in which the subject is inflicted prior to administration and a prophylactic treatment is one in which the subject is not inflicted prior to administration. In some embodiments, the subject has an increased likelihood of becoming inflicted or is suspected of being inflicted prior to treatment. In some embodiments, the subject is suspected of having an increased likelihood of becoming inflicted.

[0080] A "therapeutically effective dose" or “therapeutic dose” is an amount sufficient to effect desired clinical results (i.e., achieve therapeutic efficacy). A therapeutically effective dose can be administered in one or more administrations.

[0081] "Pharmaceutically acceptable excipient or carrier" refers to an excipient that may optionally be included in the compositions of the invention and that causes no significant adverse toxicological effects to the patient.

[0082] "Pharmaceutically acceptable salt" includes, but is not limited to, amino acid salts, salts prepared with inorganic acids, such as chloride, sulfate, phosphate, diphosphate, bromide, and nitrate salts, or salts prepared from the corresponding inorganic acid form of any of the preceding, e.g., hydrochloride, etc., or salts prepared with an organic acid, such as malate, maleate, fumarate, tartrate, succinate, ethylsuccinate, citrate, acetate, lactate, methanesulfonate, benzoate, ascorbate, para-toluenesulfonate, palmoate, salicylate and stearate, as well as estolate, gluceptate and lactobionate salts. Similarly salts containing pharmaceutically acceptable cations include, but are not limited to, sodium, potassium, calcium, aluminum, lithium, and ammonium (including substituted ammonium).

[0083] The term “responsive” as used herein means that the treatment is having the desired effect such as reducing symptom severity (e.g., bradykinesia) caused by a movement disorder orneurological disorder. When the individual does not improve in response to the treatment, it may be desirable to seek a different therapy or treatment regime for the individual.Methods

[0084] The present disclosure provides methods for administering adaptive DBS to a subject who has bradykinesia, wherein therapeutic stimulation is provided to boost movement when a patient starts to prepare for movement and during movement but is discontinued when the patient is resting. In particular, DBS is performed with a neural recording device that records brain electrical signals from neural activity associated with intended movement of the subject and automatically adjusts deep brain stimulator settings and / or delivers electrical stimulation to the brain when pre-specified patterns of neural activity associated with the intended movement are detected. The methods and systems can be used in performing open-loop therapy to provide clinical guidance to clinicians or technicians for adjusting deep brain stimulation programming. Methods and systems are also provided for performing closed-loop therapy with a deep brain stimulator that records brain electrical signals associated with intended movement and automatically adjusts deep brain stimulator settings and / or delivers electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject when pre-specified patterns of neural activity associated with intended movement are detected. Various steps and aspects of the methods will now be described in greater detail below.

[0085] The method includes positioning an electrode in a subthalamic nucleus region and / or a globus pallidus internus region of the brain of a subject to deliver electrical stimulation to the brain (i.e., DBS electrode, stimulation electrode) and an electrode at a second location in the subthalamic nucleus region of the brain of the subject and / or an electrode at a location in a contralateral cortex region of the brain of the subject to measure brain electrical signals from neural activity associated with intended movement of the subject (i.e., neural recording electrodes). In some embodiments, one or more DBS electrodes are positioned at the subthalamic nucleus region and / or the globus pallidus internus region, and one or more neural recording electrodes are positioned at the subthalamic nucleus region and / or contralateral cortex region. In some embodiments, one or more neural recording electrodes are positioned in a precentral gyrus region, a postcentral gyrus region, or both the precentral gyrus region and the postcentral gyrus region of the contralateral cortex. In some embodiments, one or more neural recording electrodes are positioned in a central sulcus region.

[0086] The DBS electrodes and the neural recording electrodes may be non-brain penetrating surface electrodes, extracranial electrodes, for example, subgaleal or skull mounted (in burrhole cap or in case of cranially mounted neurostimulator), subdural electrodes, or brain-penetrating depth electrodes. The electrical stimulation may be applied to the subthalamic nucleus or the globus pallidus internus using the DBS electrode in a manner effective for assisting movement for treating bradykinesia when brain electrical signals from neural activity associated with intended movement are detected from the subthalamic nucleus region and / or contralateral cortex region of the brain using a neural recording electrode.

[0087] In certain embodiments, one or more neural recording electrodes are used to record brain electrical signals from neural activity associated with intended movement in one or more brain regions. A neural recording electrode may be placed, for example, in a subthalamic nucleus region and / or a contralateral cortex region (e.g., a central sulcus region, precentral gyrus region, or postcentral gyrus region) to detect brain electrical signals from neural activity associated with intended movement, or in other regions of the brain suitable for detection. In certain embodiments, the brain electrical signal data comprises field potential data. The site chosen for detection may differ for different subjects and may depend on mapping of the brain of an individual subject to identify the optimal location(s) for positioning an electrode for detecting brain electrical signals from neural activity associated with intended movement, as discussed further below.

[0088] As used herein, the phrases “an electrode” or “the electrode” refer to a single electrode or multiple electrodes such as an electrode array. As used herein, the term “contact” as used in the context of an electrode in contact with a region of the brain refers to a physical association between the electrode and the region. In other words, a neural recording electrode that is in contact with a region of the brain is physically touching the region of the brain. A DBS electrode can conduct electricity to specific targets in the brain. Electrodes used in the methods disclosed herein may be monopolar (cathode or anode) or bipolar (e.g., having an anode and a cathode).

[0089] Positioning a neural recording electrode for recording neural activity at specified region(s) of the brain may be carried out using standard surgical procedures for placement of intra-cranial electrodes. In certain cases, placing the neural recording electrode may involve positioning the electrode on the surface of the specified region(s) of the brain. For example, electrodes may be placed on the surface of the brain at a subthalamic nucleus region, a contralateral cortex region (e.g., a central sulcus region, a precentral gyrus region, or postcentral gyrus region), or a combination thereof. The electrode may contact at least a portion of the surface of the brain at the subthalamic nucleus region or the contralateral cortex region (e.g., a central sulcus region, precentral gyrusregion, or postcentral gyrus region). In some embodiments, the electrode may contact substantially the entire surface area at the subthalamic nucleus region and / or contralateral cortex region. In some embodiments, the electrode may additionally contact area(s) adjacent to the subthalamic nucleus region and / or contralateral cortex region. In some embodiments, the neural recording electrodes may contact any area of the subthalamic nucleus region and / or contralateral cortex region that allows detection of brain electrical signals from neural activity associated with intended movement of the subject. In some embodiments, the electrodes may be placed extracranially, for example in the subgaleal space. In some embodiments, the electrodes may be placed in a subdural space over the contralateral cortex or under the scalp. In some embodiments, the neural recording electrode may be contained within a burr hole cap or on the case of the cranially mounted implantable neural stimulator device. In some embodiments, an electrode array arranged on a planar support substrate may be used for detecting brain electrical signals from neural activity associated with intended movement from one or more of the brain regions specified herein. The surface area of the electrode array may be determined by the desired area of contact between the electrode array and the brain. An electrode for implanting on a brain surface, such as, a surface electrode or a surface electrode array may be obtained from a commercial supplier. A commercially obtained electrode / electrode array may be modified to achieve a desired contact area. In some cases, the non-brain penetrating electrode (also referred to as a surface electrode) that may be used in the methods disclosed herein may be an electrocorticography (ECoG) electrode, a subgaleal electrode, a subdural electrode, or an electroencephalography (EEG) electrode. In certain embodiments, a plurality of electrodes is positioned in an electrode grid for detection of brain electrical signals from neural activity associated with intended movement. In certain embodiments, a plurality of electrodes is positioned at one or more of the brain regions specified herein for detection of brain electrical signals from neural activity associated with intended movement by stereoelectroencephalography (sEEG).

[0090] In certain cases, placing the neural recording electrode at a target area or site (e.g., a subthalamic nucleus region and / or contralateral cortex region of the brain) may involve positioning a brain penetrating electrode (also referred to as depth electrode) in the specified region(s) of the brain. For example, a neural recording electrode may be placed in a subthalamic nucleus region and / or a contralateral cortex region of the brain. In some embodiments, the neural recording electrode may additionally contact area(s) adjacent to a subthalamic nucleus region and / or a contralateral cortex region of the brain. In some embodiments, an electrode array may be used for detecting neural activity from a cortical area, for example, a central sulcus region, a precentral gyrus region, or postcentral gyrus region, or a combination thereof, as specified herein.

[0091] The depth to which a neural recording electrode is inserted into the brain may be determined by the desired level of contact between the electrode array and the brain. A brain-penetrating electrode array may be obtained from a commercial supplier. A commercially obtained electrode array may be modified to achieve a desired depth of insertion into the brain tissue.

[0092] Positioning an electrode in the subthalamic nucleus region or the globus pallidus internus region of the brain for delivering electrical stimulation to the brain may be carried out using standard surgical procedures for placement of electrodes for deep brain stimulation. For example, the electrode may be placed in a subthalamic nucleus region, globus pallidus internus region, or other intracranial region. Medical imaging using, for example, magnetic resonance imaging (MRI) or computerized tomography (CT) may be used to provide guidance for placement of DBS electrodes and verify correct placement of the DBS electrodes in the brain. In addition, a neurostimulator that generates electrical pulses is placed under the skin of the chest, typically below the collarbone or in the abdomen. In some embodiments the neurostimulator is cranially mounted. The surgical procedure may involve placing DBS electrodes within the brain through small holes in the skull. An electrode lead is tunneled under the skin down the neck and under the skin of the chest to connect to a chest implanted neurostimulator.

[0093] Current is supplied by the neurostimulator to the DBS electrodes. Parameters such as pulse width, shape, frequency, amplitude, pattern, and temporal distribution can be adjusted in response to changes in neural activity in the subthalamic nucleus region and / or contralateral cortex region of the brain, or alternatively accelerometry, surface electromyographic data, pulse oximetry, temperature, or heart rate. In some embodiments, a closed loop system is used to adjust DBS settings automatically in response to detection of brain electrical signals from neural activity associated with intended movement in the subthalamic nucleus region and / or contralateral cortex region of the brain. In other embodiments, an open loop system is used in which DBS settings are adjusted by a user or medical practitioner based on the detection of brain electrical signals from neural activity associated with intended movement in the subthalamic nucleus region and / or contralateral cortex region of the brain.

[0094] Electrical stimulation may be applied using a single electrode, electrode pairs, or an electrode array. In some embodiments, the number of electrodes used to deliver electrical stimulation to the brain ranges from 8 to 32, including any number of electrodes in this range such as 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, or 32 electrodes. In some embodiments, the electrical stimulation is applied to more than one site in the subthalamic nucleus or globus pallidus internus. The site to which the electrical stimulation is applied may be alternated or otherwise spatially or temporallypatterned. Electrical stimulation may be applied to the sites simultaneously or sequentially. The site chosen for stimulation may differ for different subjects and will depend on mapping of the subthalamic nucleus region and / or globus pallidus internus region of the brain of an individual subject to identify the optimal location for positioning an electrode for delivery of electrical stimulation to assist movement for treating bradykinesia.

[0095] In some embodiments, an electrode array arranged on a planar support substrate may be used for electrically stimulating the subthalamic nucleus or globus pallidus internus. The surface area of the electrode array may be determined by the desired area of contact between the electrode array and the subthalamic nucleus or globus pallidus internus. In some cases, cylindrical electrode arrays, paddle-style electrode arrays, or plate-style electrode arrays may be used in the methods disclosed herein for deep brain stimulation. Such DBS electrode arrays for implanting in the brain, may be obtained from a commercial supplier. A commercially obtained electrode / electrode array may be modified to achieve a desired contact area.

[0096] The precise number of DBS electrodes or neural recording electrodes contained in an electrode array (e.g., for electrical stimulation or detection of neural activity) may vary. In certain aspects, an electrode array may include two or more electrodes, such as 3 or more, including 4 or more, e.g., about 3 to 6 electrodes, about 6 to 12 electrodes, about 12 to 18 electrodes, about 18 to 24 electrodes, about 24 to 30 electrodes, about 30 to 48 electrodes, about 48 to 72 electrodes, about 72 to 96 electrodes, or about 96 or more electrodes. The electrodes may be arranged into a regular repeating pattern (e.g., a grid, such as a grid with about 1 cm spacing between electrodes), or no pattern. An electrode that conforms to the target site for optimal delivery of electrical stimulation may be used. One such example, is a single multi contact electrode with eight contacts separated by 21 / 2 mm. Each contract would have a span of approximately 2 mm. Another example is an electrode with two 1 cm contacts with a 2 mm intervening gap. Yet further, another example of an electrode that can be used in the present methods is a 2 or 3 branched electrode to cover the target site. Each one of these three-pronged electrodes has four 1 -2 mm contacts with a center to center separation of 2 of 2.5 mm and a span of 1 .5 mm.

[0097] The size of each electrode may also vary depending upon such factors as the number of electrodes in the array, the location of the electrodes, the material, the age of the patient, and other factors. In certain aspects, an electrode array has a size (e.g., a diameter) of about 5 mm or less, such as about 4 mm or less, including 4 mm-0.25 mm, 3 mm-0.25 mm, 2 mm-0.25 mm, 1 mm-0.25 mm, or about 3 mm, about 2 mm, about 1 mm, about 0.5 mm, or about 0.25 mm.

[0098] In certain embodiments, the method further comprises mapping the brain of the subject to optimize positioning of an electrode for applying electrical stimulation. Positioning of a DBS electrode is optimized to maximize clinical responses to electrical stimulation to assist movement for treating bradykinesia. In some embodiments, the subthalamic nucleus region, globus pallidus internus region, or other regions of the brain are mapped to determine optimal positioning of DBS electrodes.

[0099] Assessment of the effectiveness of electrical stimulation for treating bradykinesia may be performed using any standard method. In some embodiments, a Movement Disorder Society- Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) or a Hoehn and Yahr (HnY) scale, a Parkinson’s Disease Composite Scale (PDCS), or a Schwab and England Activities of Daily Living (ADL) Scale may be used to assess the effectiveness of electrical stimulation in treating bradykinesia. In some embodiments, assessing effectiveness of the treatment of bradykinesia comprises monitoring the subject using multi-view video recordings of the subject, keylogging data from a computer used by the subject, or a wearable monitor that can acquire accelerometry, gyroscope and / or surface electromyographic (sEMG) data. For example, a wristwatch style wearable monitor such as the Parkinson’s KinetiGraph®, PKG®, available from PKG Health (San Francisco, CA), can be used to monitor movement continuously to detect various motor symptoms of a movement disorder such as bradykinesia, dyskinesia, tremor, daytime immobility, stiffness, slow movements, gait / walking, daytime somnolence, and sleep fragmentation. An Apple Watch is available from Apple Inc. (Cupertino, CA), which can be used, for example, for monitoring, bradykinesia, dyskinesia, tremors, gait, and arm movement.

[0100] In certain embodiments, the method further comprises mapping the brain of the subject to optimize positioning of a neural recording electrode. Positioning of the neural recording electrode in a subthalamic nucleus region and / or contralateral cortex region is optimized to detect brain activity features, including brain electrical signals from neural activity associated with intended movement that can be assisted by treatment with electrical stimulation. For example, the levels of overall power, or power in specific frequency ranges (e.g., alpha, beta, gamma, delta, and / or theta) may be correlated with intended movement by the subject. In certain embodiments, electrical stimulation is applied to the subthalamic nucleus region or the globus pallidus interna region when the level of theta frequency (such as 4 Hz to 7 Hz) power, alpha frequency (such as 8 Hz to 12 Hz) power, beta frequency (such as 13 Hz to 30 Hz) power, and / or gamma frequency (such as 60 Hz to 90 Hz) power detected with a neural recording electrode in a subthalamic nucleus region and / or contralateral cortex region indicates that the patient is in an intended movement state. Thus, neural recording electrodes may be positioned to optimize detection of brain activity in specific frequency ranges thatcorrelate with intended movement that can be assisted by treatment with electrical stimulation such as theta frequency neural oscillations in a range from 4 Hz to 7 Hz, alpha frequency neural oscillations in a range from 8 Hz to 12 Hz, beta frequency neural oscillations in a range from 13 Hz to 30 Hz, and / or gamma frequency neural oscillations in a range from 60 Hz to 90 Hz. In certain embodiments, personalized power band features for distinguishing an intended movement state from a stationary state are identified for a subject. In certain embodiments, accelerometry data, gyroscope data, multi-view video recordings of the subject, keylogging data from a computer used by the subject, or any combination thereof is used in combination with recorded brain electrical signal data to aid in distinguishing between an intended movement state and a stationary state of a subject.

[0101] Detection of brain activity may be performed by any method known in the art. For example, functional brain imaging of neural activity may be carried out by electrical methods such as electroencephalography (EEG), stereoelectroencephalography (sEEG), electrocorticography (ECoG), magnetoencephalography (MEG), single photon emission computed tomography (SPECT), as well as metabolic and blood flow studies such as functional magnetic resonance imaging (fMRI), and positron emission tomography (PET). In some embodiments, the subthalamic nucleus, contralateral cortex, or other regions are mapped to determine optimal positioning for neural recording electrodes. One or more of these regions may be implanted with neural recording electrodes to measure electrical signals from neural activity, including brain electrical signals from neural activity associated with intended movement that can be assisted by treatment with electrical stimulation.

[0102] As set forth herein, the subject methods involve applying electrical stimulation to a subthalamic nucleus region or a globus pallidus internus region in a manner effective to assist movement of a subject when brain electrical signals from neural activity associated with intended movement are detected. In certain embodiments, the brain electrical signals from neural activity associated with intended movement comprise theta frequency neural oscillations in a range from 4 Hz to 7 Hz, alpha frequency neural oscillations in a range from 8 Hz to 12 Hz, beta frequency neural oscillations in a range from 13 Hz to 30 Hz, and / or gamma frequency neural oscillations in a range from 60 Hz to 90 Hz. In certain embodiments, neural oscillations in a personalized frequency range identified for a subject are used for distinguishing the intended movement state from the stationary state. In certain embodiments, the electrical stimulation is selectively increased when an intended movement state of the subject is detected and discontinued during a stationary state.

[0103] Closed-loop therapy can be performed with a neurostimulator used in combination with a neural recording device that records brain electrical activity, wherein electrical stimulation is deliveredto the subthalamic nucleus or globus pallidus internus of the brain of the subject when a pattern of neural activity associated with intended movement is detected. The parameters for applying the electrical stimulation to the brain may be determined empirically during treatment or may be predefined, such as, from a trial study with a subject. For example, brain electrical signals from neural activity associated with intended movement are recorded from a subthalamic nucleus region and / or a contralateral cortex region (e.g., from the central sulcus region, precentral gyrus region, or postcentral gyrus region) of the brain of the subject. Varying stimulation settings may be applied when certain features are detected, including baseline (stimulation off), optimal therapeutic stimulation, modified and ineffective stimulation, and maximum tolerated stimulation to identify personal neural signatures of “an intended movement” for a patient, which are used to assist with programming of a DBS device to determine optimal therapeutic stimulation parameters for assisting movement for treating bradykinesia. The parameters of the electrical stimulation may include one or more of frequency, pulse width / duration, duty cycle, intensity / amplitude, pulse pattern, program duration, program frequency, and the like.

[0104] Frequency refers to the pulses produced per second during stimulation and is stated in units of Hertz (Hz, e.g., 60 Hz = 60 pulses per second). The frequencies of electrical stimulation used in the present methods may vary widely depending on numerous factors and may be determined empirically during treatment of the subject or may be pre-defined. In certain embodiments, the method may involve applying electrical stimulation to the brain at a frequency of 2 Hz - 250 Hz, such as, 25 Hz - 200 Hz, 50 Hz - 250 Hz, 50 Hz -185 Hz, 50 Hz -150 Hz, 75 Hz - 200 Hz, 100 Hz - 200 Hz, 100 Hz - 180 Hz, 100 Hz - 160 Hz, 120 Hz - 150 Hz, or 130 Hz - 140 Hz. In some embodiments, the electrical stimulation to the brain is applied at a frequency of about 120 Hz to about 160 Hz, including any pulse frequency within this range such as 120 Hz, 122 Hz, 124 Hz, 126 Hz, 128 Hz, 130 Hz, 132 Hz, 133 Hz, 134 Hz, 135 Hz, 136 Hz, 137 Hz, 138 Hz, 139 Hz, 140 Hz, 142 Hz, 144 Hz, 146 Hz, 148 Hz, 150 Hz, 152 Hz, 154 Hz, 156 Hz, 158 Hz, or 160 Hz. In some embodiments, noninteger pulse frequencies are used (e.g. 135.2 Hz, 135.4 Hz, etc.).

[0105] The electrical stimulation may be applied in pulses such as a uniphasic or a biphasic pulse. The time span of a single pulse is referred to as the pulse width or pulse duration. The pulse width used in the present methods may vary widely depending on numerous factors (e.g., severity of the disease, status of the patient, and the like) and may be determined empirically or may be pre-defined. In certain embodiments, the method may involve applying an electrical stimulation at a pulse width of about 10 p.sec - 500 p.sec, for example, 20 p.sec - 450 |isec, 40 gsec - 450 p.sec, 60 |isec - 450 | sec, 60 p.sec - 220 jisec, 60 p.sec - 120 jisec, or 60 jisec - 90 |isec. In some embodiments,the electrical stimulation to the brain is applied at a pulse width of about 60 nsec to about 210 |isec, including any pulse width within this range such as 60 p.sec, 65 |isec, 70 |isec, 75 |isec, 80 p.sec, 85 |isec, 90 p.sec, 95 p.sec, 100 |isec, 105 |isec, 1 10v, 115 p.sec, 120 p.sec, 125 p.sec, 130 p.sec, 135 | sec, 140 p-sec, 145 p.sec, 150 |isec, 155 |isec, 160 i^sec, 165 nsec, 170 |j.sec, 175 |isec, 180 |isec, 185 p.sec, 190 |j.sec, 195 p.sec, 200 .sec, 205 p.sec, 210 p.sec, 215 |j.sec, or 220 .sec.

[0106] The electrical stimulation may be applied for a stimulation period of 0.1 sec-1 month, with periods of rest (i.e., no electrical stimulation) possible in between. In certain cases, the period of electrical stimulation may be 0.1 sec-1 week, 1 sec-1 day, 10 sec-12 hours, 1 min-6 hours, 10 min- 1 hour, and so forth. In certain cases, the period of electrical stimulation may be 1 sec-1 min, 1 sec- 30 sec, 1 sec-15 sec, 1 sec-10 sec, 1 sec-6 sec, 1 sec-3 sec, 1 sec-2 sec, or 6 sec-10 sec. The period of rest in between each stimulation period may be 60 sec or less, 30 sec or less, 20 sec or less, or 10 sec. In some embodiments, electrical stimulation may be applied for a year or more, 2 years or more, 3 years or more, 5 years or more, or 10 years or more. In some embodiments, electrical stimulation may be continued indefinitely as part of a long-term DBS therapy regimen.

[0107] The electrical stimulation may be applied with an amplitude of current of 0.1 mA-30 mA, such as, 0.1 mA-25 mA, such as, 0.1 mA-20 mA, 0.1 mA-15 mA, 0.1 mA-10 mA, 0.1 mA-2 mA, 0.1 mA-1 mA, 1 mA-20 mA, 1 mA-10 mA, 2 mA-30 mA, 2 mA-15 mA, 2 mA-10 mA, or 1 mA-3 mA. In some embodiments, the amplitude of current is 0.1 mA-3.5 mA, or any amplitude of current in this range such as 0.1 mA, 0.2 mA, 0.3 mA, 0.4 mA, 0.5 mA, 0.6 mA, 0.7 mA, 0.8 mA, 0.9 mA, 1 .0 mA, 1.1 mA, 1 .2 mA, 1 .3 mA, 1 .4 mA, 1 .5 mA, 1 .6 mA, 1 .7 mA, 1 .8 mA. 1 .9 mA, 2.0 mA, 2.1 mA, 2.2 mA, 2.3 mA, 2.4 mA, 2.5 mA, 2.6 mA, 2.7 mA, 2.8 mA, 2.9 mA, 3.0 mA, 3.1 mA, 3.2 mA, 3.3 mA, 3.4 mA, or 3.5 mA.

[0108] The electrical stimulation may be applied with an amplitude of voltage of 0.1 V-15 V, such as, 0.1 V-10 V, 0.1 V-5 V, 1 V-10 V, 1 V-5, V, or 1 V-3.5 V. In some embodiments, the amplitude of voltage is 1 V-3.5 V, or any amplitude of voltage in this range such as 1 V, 1 .1 V, 1 .2 V, 1 .3 V, 1 .4 V, 1 .5 V, 1 .6 V, 1 .7 V, 1 .8 V, 1 .9 V, 2.0 V, 2.1 V, 2.2 V, 2.3 V, 2.4 V, 2.5 V, 2.6 V, 2.7 V, 2.8 V, 2.9 V, 3.0 V, 3.1 V, 3.2 V, 3.3 V, 3.4 V, or 3.5 V.

[0109] The electrical stimulation having the parameters as set forth above may be applied over a program duration of around 1 day or less, such as, 18 hours, 6 hours, 3 hours, 2 hours, 1 hour, 45 minutes, 30 minutes, 20 minutes, 10 minutes, or 5 minutes, or less, e.g., 1 minute - 5 minutes, 2 minutes - 10 minutes, 2 minutes - 20 minutes, 2 minutes - 30 minutes, 5 minutes - 10 minutes, 5 minutes - 30 minutes, or 5 minutes - 15 minutes, 10 minutes - 400 minutes, 25 minutes - 300 minutes, 50 minutes - 200 minutes, or 75 minutes - 150 minutes, which period would include the applicationof pulses and the intervening rest period. The program may be repeated at a desired program frequency to assist movement of the subject. As such, a treatment regimen may include a program for electrical stimulation at a desired program frequency and program duration. In some embodiments, the treatment regimen is controlled by a control unit in communication with a pulse generator connected to the one or more DBS electrodes in a closed-loop treatment regimen.

[0110] As noted above, the treatment may assist movement of the subject and reduce motor symptoms such as bradykinesia, dyskinesia, dysarthria, dystonia, tremor, or gait disturbance. Assessment of effectiveness of the treatment may be performed using any known method for evaluating motor symptoms. In some embodiments, efficacy of the treatment is evaluated using a Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) or a Hoehn and Yahr (HnY) scale, a Parkinson’s Disease Composite Scale (PDCS), or a Schwab and England Activities of Daily Living (ADL) Scale. In some embodiments, the subject is monitored for motor symptoms using a wearable monitor that can acquire accelerometry, gyroscope, and / or surface electromyographic (sEMG) data to evaluate the effectiveness of the treatment. A wrist-watch style wearable monitor such as the Parkinson’s KinetiGraph®, PKG® is available from PKG Health (San Francisco, CA), which can monitor movement continuously and detect various motor symptoms of a movement disorder such as dyskinesia, bradykinesia, tremor, daytime immobility, stiffness, slow movements, gait / walking, daytime somnolence, and sleep fragmentation. An Apple Watch is available from Apple Inc. (Cupertino, CA), which can be used, for example, for monitoring, bradykinesia, dyskinesia, tremors, gait, and arm movement.

[0111] In certain cases, effectiveness of treatment may be assessed by detecting brain electrical activity (e.g., brain electrical signals from neural activity associated with intended movement), which may be within a subthalamic nucleus region and / or contralateral cortex region, or another area. For example, the brain region may be the central sulcus region, precentral gyrus region, or postcentral gyrus region. Detection of brain activity may be performed by functional brain imaging. Functional brain imaging may be carried out by electrical methods such as electroencephalography (EEG), chronic subgaleal recordings, burrhole or cranially mounted neurostimulator electrode recording, electrocorticography (ECoG), magnetoencephalography (MEG), single photon emission computed tomography (SPECT), as well as metabolic and blood flow studies such as functional magnetic resonance imaging (fMRI), and positron emission tomography (PET). In some embodiments, electrical methods for assessing effectiveness of treatment may involve use of a neural recording electrode as described herein or placement of an additional electrode for measuring electrical signals at a secondary region of the brain or in the skull, or extracranially. One or more regions of the brainmay be implanted with an electrode and electrical signals measured for assessment of effectiveness of the treatment. Any suitable electrodes may be used for measurements and may include one or more surface electrodes (non-brain penetrating electrode(s)) or one or more depth electrodes (brain penetrating electrode(s)) as described herein.

[0112] Assessment of effectiveness of treatment may be performed at any suitable time point after commencement of the treatment procedure, for example, during open-loop or closed-loop therapy. Embodiments of the subject methods include assessing effectiveness of electrical stimulation in assisting movement for treating bradykinesia in a subject within seconds, minutes, hours, or days after the initial treatment regimen has been completed. In some instances, assessment may be performed at multiple time points. In some cases, more than one type of assessment may be performed at the different time points. In some embodiments, brain activity including brain electrical signals from neural activity associated with intended movement in a subthalamic nucleus region or a contralateral cortex region (e.g., at the central sulcus, precentral gyrus region, or postcentral gyrus region) may be measured prior to the application of electrical stimulation, and assessing may include comparing the subject’s brain activity after the treatment to that before the treatment and a change in the post-treatment brain activity may indicate successful treatment.

[0113] Upon completion of a treatment regimen, the patient may be assessed for effectiveness of the treatment and the treatment regimen may be repeated, if needed. In certain cases, the treatment regimen may be altered before repeating. For example, one or more of the frequency, pulse width, current amplitude, period of electrical stimulation, program duration, program frequency, and / or placement of DBS or neural recording electrodes may be altered before starting a second treatment regimen.

[0114] Application of the method may include a prior step of selecting a patient for treatment based on need as determined by clinical assessment, which may include assessment of severity of bradykinesia, physical condition, cognitive assessment, anatomical assessment, behavioral assessment and / or neurophysiological assessment. In certain cases, a subject may be further assessed to determine if adaptive deep brain stimulation will completely or partially (e.g., at least 50%) relieve bradykinesia. Such a patient may undergo DBS on a temporary trial basis to determine if DBS decreases the severity of bradykinesia experienced by the patient. Such a patient may also be implanted with neural recording electrodes to identify personalized neural signatures of “intended movement” and “relief of bradykinesia” to assist with deep brain stimulation programming to determine therapeutic stimulation parameters for the patient and / or evaluate whether DBS therapy will be effective for treating bradykinesia in the patient.

[0115] In certain aspects, the methods and systems of the present disclosure may include measurement of brain activity, for example, brain electrical signals from neural activity associated with intended movement in a subthalamic nucleus region and / or contralateral cortex region, wherein the level of theta frequency (such as 4 Hz to 7 Hz) power, alpha frequency (such as 8 Hz to 12 Hz) power, beta frequency (such as 13 Hz to 30 Hz) power, and / or gamma frequency (such as 60 Hz to 90 Hz) power is measured to determine if the subject is in an intended movement state or a stationary state. In certain cases, brain electrical signals from neural activity associated with intended movement may be measured from a plurality of locations in subthalamic nucleus and / or contralateral cortex regions and averaged. In some cases, data driven approaches are used to identify spectral features that are individualized and different from canonical power bands. In some cases, when the subject is in a stationary state, the methods and systems do not apply a further stimulation to the brain. Alternatively, when the subject is in an intended movement state, the methods and systems may apply a further stimulation to the brain. In certain cases, the application of electrical stimulation to the brain may alter other neural features from one more regions of the brain. The alterations may be compared to the state of these features prior to the application of stimulation.

[0116] A closed-loop method allows determination of parameters of electrical stimulation based upon real-time feedback signals from the brain of the subject. Closed-loop methods and systems allow for automation of treatment of the subject including real-time need-based modulation of the treatment regimen. Exemplary closed-loop methods and associated systems for treatment of bradykinesia are further discussed in the Examples section and are depicted in FIGS. 1 B and 1 C. Closed-loop methods and systems for automated delivery of electrical stimulation are further described below.Closed-Loop Method for Automated Delivery of Electrical Stimulation

[0117] In certain embodiments, a control algorithm is used to automate the delivery of electrical stimulation to the brain in response to detection of neural activity associated with intended movement in order to assist movement for treatment of bradykinesia. According to certain embodiments, the method may include measuring brain electrical signals from neural activity associated with intended movement from a subthalamic nucleus region and / or contralateral cortex region (e.g., a central sulcus region, precentral gyrus region, or postcentral gyrus region) of the brain of the subject via a neural recording electrode; applying electrical signal metrics to a control algorithm that is tuned to a clinically relevant target (e.g., a range of signal indicative of intended movement or effective treatment of bradykinesia); automatically delivering electrical stimulation to the subthalamic nucleusregion or the globus pallidus internus region of the brain via a DBS electrode in a manner effective to treat the bradykinesia if the electrical signal metrics indicate that the patient is in an intended movement state and in need of treatment. For example, brain electrical signals from neural activity associated with intended movement may be detected from a subthalamic nucleus region and / or contralateral cortex region (e.g., a central sulcus region, precentral gyrus region, or postcentral gyrus region) with a neural recording electrode, wherein the control algorithm receives the electrical activity data from the neural recording electrode and automates delivery of electrical stimulation via a DBS electrode to the brain when the level of theta frequency (such as 4 Hz to 7 Hz) power, alpha frequency (such as 8 Hz to 12 Hz) power, beta frequency (such as 13 Hz to 30 Hz) power, and / or gamma frequency (such as 60 Hz to 90 Hz) power indicates that the patient is in an intended movement state. In some embodiments, one or more programmed stimulation parameters are modulated according to the algorithm’s control law based on the recorded electrical activity data; and modulated electrical stimulation is delivered to the brain via the DBS electrode in a manner effective to mitigate bradykinesia in the subject.

[0118] As described in the foregoing sections, effectiveness of treatment of bradykinesia may be assessed by detecting brain electrical activity, including brain electrical signals from neural activity associated with intended movement using a neural recording electrode. In an open-loop system, stimulation is delivered in a pre-programmed way or manually by a user but is not automatically controlled by real-time neural feedback from the patient’s brain. The electrical activity may be analyzed by a computing means which may output recommendations based on comparing the electrical activity to a predetermined range. A user may then carry out the recommendations, such as changing a parameter of the electrical stimulation program prior to starting another treatment regimen. In a closed-loop system, by contrast, a computing means can automatically update stimulation parameters based upon analysis of the recorded electrical signal and / or automatically deliver stimulation to the brain according to the electrical stimulation program. In some embodiments, either an open-loop or a closed-loop system may be integrated with a mechanism for user intervention, for example by allowing user-override of open-loop or closed-loop stimulation programs to enact or prevent stimulation that would ordinarily occur, or to manually change parameters of such stimulation.

[0119] In some embodiments, the computing means for directing closed-loop stimulation may be a combination of hardware / software which may be connected wirelessly or by wire to the neural recording electrodes. The computing means may communicate with a control unit (also referred to as a control module) that controls a neurostimulator pulse generator connected to the DBSelectrodes. In certain embodiments, the computing means may be connected to a recorder (e.g., a neurophysiological recorder or neural recording device) that records brain activity, including brain electrical signals from neural activity associated with intended movement measured by the neural recording electrodes. The computing means may include a control algorithm that determines modification of stimulation parameters based on real-time outputs of the neural recording device. The algorithm may operate by simple on / off control of stimulation at set parameters, modifying only the on / off parameter with each evaluation cycle, or may determine sophisticated modification of a range of stimulation parameters with each cycle. In some cases, the algorithm may be based on information related to assisting movement of a subject having bradykinesia, such as, a range of electrical activity (e.g., brain electrical signals from neural activity associated with intended movement) that is indicative of the need to be treated with electrical stimulation. The algorithm may also include additional information such as a brain activity profile of a normal subject (not suffering from bradykinesia). Regardless of the particular control algorithm structure, the computing means may be tuned to a clinically relevant target (e.g., a range of signal indicative of an intended movement state and the need for treatment with electrical stimulation to assist movement, a range of signal indicative of a stationary state and the need to discontinue treatment with electrical stimulation, and / or a range of signal indicative of effective treatment of bradykinesia) that directs modulation of one or more programmed stimulation parameters according to the algorithm’s control law, applying the modulated electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain via the DBS electrode.

[0120] In some cases, the computing means, via a control algorithm, may determine whether the received electrical signals (e.g., brain electrical signals from neural activity associated with intended movement) are within or outside a predetermined range of neural signals indicative of the of the need for treatment with electrical stimulation to assist movement. When the received electrical signals are outside this predetermined range, then the computing means determines that the subject is in a stationary state. The computing means may then communicate with the control unit to direct stimulation shut-off by the neurostimulator pulse generator. When the received electrical signals are within the predetermined range of neural signals indicative of an intended movement state, then the computing means determines that the subject should be treated with deep brain stimulation to assist movement. The control algorithm within the computing means may then determine whether the initial step of applying electrical stimulation to the brain should be repeated and / or whether a parameter of the electrical stimulation should be modified prior to the step of applying electrical stimulation when brain activity indicating intended movement by the subject is detected. The computing means, viathe control unit, may then communicate with the control unit to provide the appropriate instructions to the neurostimulator pulse generator.

[0121] Thus, in certain aspects, the subject methods operate as a closed-loop control system which may automatically adjust one or more parameters in response to electrical activity from a region of the brain of a subject and / or automatically deliver electrical stimulation to the brain according to the electrical stimulation program. In some embodiments, the closed-loop control system automatically delivers electrical stimulation according to set parameters when the received electrical signals are within a predetermined range indicative of an intended movement state that should be treated with electrical stimulation to assist movement of the subject. Exemplary closed-loop methods and associated systems are described in the Examples section of the application and are illustrated in FIGS. 1 B and 10.

[0122] In some aspects, the closed loop system may be used to sense a subject’s need for treatment using the methods disclosed herein. For example, the closed loop system may be programmed to monitor brain activity from one or more subthalamic nucleus and / or contralateral cortex regions of the brain to detect brain activity in a range indicative of intended movement. Upon detection of electrical activity indicative of intended movement, the closed loop system may automatically commence a treatment protocol of applying electrical stimulation to the brain to assist movement of the subject.

[0123] In additional aspects, the closed loop system may be used as a system for monitoring brain activity and correlating the brain activity and treatment with electrical stimulation to actual movement of the subject. Multi-view video recordings of the subject, keylogging data from a computer used by the subject, or a wearable monitor that can acquire accelerometry, gyroscope, and / or surface electromyographic (sEMG) data can be used to detect movements of the subject. Since the closed loop system is configured for recording electrical signals from a subject’s brain, movement of the subject may be monitored in real-time continuously and correlated with the measured electrical signals to provide a biomarker that is related to intended movement of the subject. For example, electrical activity (e.g., brain electrical signals from neural activity associated with intended movement) measured when a subject attempts movement (e.g., directed movement such as a finger tap, opening or closing of a hand, wrist pronation or supination, walking, or keyboard typing; or undirected movement such as any natural movement of the subject during daily life) can be used to develop a biomarker, e.g., a range of electrical activity indicative of intended movement, and so on. As such, closed loop systems are useful for detecting personalized neural biomarkers for an individual.

[0124] It is understood that electrical signals that are indicative of intended movement or relief of bradykinesia for a subject may be recorded from a subject’s brain and may be used in aspects outside of a closed loop system. For example, electrical signals indicative of intended movement or relief of bradykinesia for a subject may be recorded from a subthalamic nucleus region and / or contralateral cortex region (e.g., a central sulcus region, precentral gyrus region, or postcentral gyrus region), or other brain region using electrodes or another device operably coupled to the patient’s brain, which electrodes or device may or may not be part of a closed loop system. The patient may be treated as disclosed herein (e.g., by applying electrical stimulation to the brain), and electrical signals (e.g., brain electrical signals from neural activity associated with intended movement) may be recorded from a subthalamic nucleus region and / or contralateral cortex region, or other region in real time as the treatment is administered or after the treatment is administered. The electric signals recorded after the administration of electrical stimulation is commenced may then be compared to the electric signals recorded prior to the treatment to determine features in the recorded electric signals that change post-treatment. These features provide a feedback signal to indicate whether the treatment is having an effect on the patient’s bradykinesia. These features can also serve as feedback signals to a closed loop system. These features may include the overall power, or power in specific frequency ranges (e.g., alpha, beta, gamma, delta, and / or theta). In some cases, these features may be patient specific or specific to bradykinesia, or both. For example, some of the features may be features found in a plurality of patients having bradykinesia; some of the features may be features in a particular patient which may not be found in a significant number of other patients having bradykinesia. In some embodiments, a combination of patient-specific features and bradykinesia-specific features may be monitored to assess efficacy of treatment.

[0125] In a particular aspect, the closed loop system and methods provided herein may involve a recording of electrical signals from one or more subthalamic nucleus and / or contralateral cortex regions (e.g., a central sulcus region, precentral gyrus region, or postcentral gyrus region, or other region) of a patient’s brain. The patient may then be treated by application of electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain, and electrical signals may be recorded from a subthalamic nucleus region and / or contralateral cortex region of the brain (e.g., a central sulcus region, precentral gyrus region or postcentral gyrus region, or other region) and compared to a pre-treatment recording. Features in the recorded signals that change after the electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region would correspond to biomarkers that indicate whether the treatment is having an effect. The change in recorded signals can also optionally be correlated to the level of assistance with movement andrelief of bradykinesia reported by the patient after the treatment. The change can be used for modulating the treatment in a closed loop system to improve assistance with movement. When the change in the recorded signal correlates with the subject being in a stationary state, those features would indicate to a computing means of a closed loop system that electrical stimulation need not be performed.

[0126] In some embodiments, one or more pattern recognition methods can be used in analyzing recorded brain electrical activity data to automate detection of brain activity features such as brain electrical signals from neural activity associated with intended movement. The models and / or algorithms can be provided in machine readable format and may be used to correlate the levels of overall power, or power in specific frequency ranges (e.g., alpha, beta, gamma, delta, and / or theta) with intended movement that can be assisted by treatment with deep brain electrical stimulation. In some embodiments, the level of theta frequency (such as 4 Hz to 7 Hz) power, alpha frequency (such as 8 Hz to 12 Hz) power, beta frequency (such as 13 Hz to 30 Hz) power, and / or gamma frequency (such as 60 Hz to 90 Hz) power is correlated with intended movement to determine if a patient is treated with electrical stimulation to assist movement. Alternatively or additionally, coherence within certain spectral frequency bands or other features of network connectivity may be correlated with intended movement that can be assisted by treatment with electrical stimulation.

[0127] In some embodiments, a computer implemented method for programming a deep brain stimulator to treat bradykinesia in a subject is provided, the computer performing steps comprising: receiving recorded brain electrical signal data from a contralateral cortex region and / or a subthalamic nucleus region of the brain of the subject; analyzing the recorded brain electrical signal data using a movement classification model that identifies patterns of electrical signals in the recorded brain electrical signal data associated with intended movement of the subject; adjusting one or more programmed stimulation parameters based on the recorded brain electrical signal data according to an algorithm control law; and instructing the deep brain stimulator to apply an electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain to provide a prokinetic effect on the intended movement of the subject. In certain embodiments, the electrical stimulation is selectively increased during an intended movement state and not during a stationary state.

[0128] Analyzing the recorded brain electrical activity (e.g., brain electrical signals from neural activity associated with intended movement) may comprise the use of an algorithm or classifier to distinguish when a subject is in an intended movement state or a stationary state. In certain embodiments, a machine learning algorithm is used to generate a motor symptom classificationmodel. The machine learning algorithm may comprise a supervised learning algorithm. Examples of supervised learning algorithms may include Average One-Dependence Estimators (AODE), Artificial neural network (e.g., Backpropagation), Bayesian statistics (e.g., I Bayes classifier, Bayesian network, Bayesian knowledge base), Case-based reasoning, Decision trees, Inductive logic programming, Gaussian process regression, Group method of data handling (GMDH), Learning Automata, Learning Vector Quantization, Minimum message length (decision trees, decision graphs, etc.), Lazy learning, Instance-based learning Nearest Neighbor Algorithm, Analogical modeling, Probably approximately correct learning (PAG) learning, Ripple down rules, a knowledge acquisition methodology, Symbolic machine learning algorithms, Subsymbolic machine learning algorithms, Support vector machines (SVM), Random Forests, Ensembles of classifiers, Bootstrap aggregating (bagging), and Boosting. Supervised learning may comprise ordinal classification such as regression analysis and Information fuzzy networks (IFN). Alternatively, supervised learning methods may comprise statistical classification, such as AODE, Linear classifiers (e.g., Fisher's linear discriminant, Logistic regression, Naive Bayes classifier, Perceptron, and Support vector machine), quadratic classifiers, k-nearest neighbor, Boosting, Decision trees (e.g., C4.5, Random forests), Bayesian networks, and Hidden Markov models.

[0129] The machine learning algorithms may also comprise an unsupervised learning algorithm. Examples of unsupervised learning algorithms may include artificial neural network (recurrent or convoluted), Data clustering, Expectation-maximization algorithm, Self-organizing map, Radial basis function network, Vector Quantization, Generative topographic map, Information bottleneck method, and IBSEAD. Unsupervised learning may also comprise association rule learning algorithms such as Apriori algorithm, Eclat algorithm and FP-growth algorithm. Hierarchical clustering, such as Single-linkage clustering and Conceptual clustering, may also be used. Alternatively, unsupervised learning may comprise partitional clustering such as K-means algorithm and Fuzzy clustering.

[0130] In some instances, the machine learning algorithms comprise a reinforcement learning algorithm. Examples of reinforcement learning algorithms include, but are not limited to, temporal difference learning, Q-learning and Learning Automata. Alternatively, the machine learning algorithm may comprise Data Pre-processing. In certain embodiments, the motor symptom classification model is trained by analyzing brain electrical signal data recorded over multiple days.

[0131] In certain embodiments, the machine learning algorithm further determines whether the brain electrical signals from neural activity associated with intended movement are better measured by the first neural recording electrode in the subthalamic nucleus region or the second neural recordingelectrode in the contralateral cortex region for use in the classification to distinguish between the intended movement state and the stationary state.

[0132] In certain embodiments, the control algorithm uses a linear classifier or a binary model for classification to distinguish between the intended movement state and the stationary state. For example, linear discriminant analysis may be used for classification to distinguish between the intended movement state and the stationary state. In some embodiments, a receiver operating characteristic curve (ROC) is used to identify a threshold for classification of the subject as being in an intended movement state.

[0133] In certain embodiments, the computer implemented method further comprises performing principal components (PC) decomposition on power spectral data to identify personalized power band features for distinguishing the intended movement state from the stationary state. In some embodiments, the computer implemented method further comprises: transforming a time-domain of a local field potential (LFP) into a spectrogram using a short-time Fourier transform (STFT); performing principal components analysis (PCA) on z-scored STFT outputs to identify coordinated bands of neural activity; and applying a peak-finding algorithm to PCA component weights to find a frequency range best approximating each principal component. In some embodiments, the top principal components for the contralateral cortex region or the subthalamic nucleus region correspond to personalized frequency ranges for beta frequency neural oscillations and gamma frequency neural oscillations for the brain electrical signal associated with the intended movement of the subject.

[0134] In certain embodiments, the computer implemented method further comprises using accelerometry data, gyroscope data, multi-view video recordings of the subject, keylogging data from a computer used by the subject, or any combination thereof in combination with the recorded brain electrical signal data to distinguish between the intended movement state and the stationary state.

[0135] In certain embodiments, the computer implemented method further comprises training a machine learning model to predict whether the subject is in an intended movement state using accelerometry data for detected movement of the subject. In some embodiments, the accelerometry data is from a wearable accelerometer worn by the subject that detects natural undirected movement of the subject. In some embodiments, the accelerometry data is from a wearable accelerometer worn by the subject that detects movement during directed movement tasks selected from finger tapping, opening or closing of a hand, wrist pronation or supination, walking, and keyboard typing. For example, a wrist-watch style wearable monitor such as the Parkinson’s KinetiGraph®, PKG®, available from PKG Health (San Francisco, GA), or the Apple Watch, available from Apple Inc.(Cupertino, CA), can be used to monitor movement of the subject continuously, including arm movement and gait. Alternatively or additionally, movement of the subject can be monitored using multi-view video recordings of the subject during movement, and such recordings can be used to train the machine learning model to distinguish between the intended movement state and the stationary state.

[0136] In certain embodiments, the computer implemented method further comprises: a) ranking predicted stimulation effectiveness for available settings of a DBS device based on classifier scores for stimulation effectiveness of each setting using a linear classification model; b) selecting stimulation settings predicted to have highest stimulation effectiveness based on the linear classification model; c) receiving recorded brain electrical signal data from the contralateral cortex region and the subthalamic nucleus region of the brain of the subject after applying electrical stimulation with the DBS device to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the settings predicted to have the highest stimulation effectiveness; d) analyzing the recorded brain electrical signal data to evaluate neural response of the subject to the electrical stimulation; e) updating the linear classification model based on the neural response of the subject to the electrical stimulation to generate an updated linear classification model; f) updating the ranking of predicted stimulation effectiveness for the available settings of the DBS device using the updated linear classification model; g) selecting stimulation settings predicted to have the highest stimulation effectiveness based on the updated linear classification model; h) receiving recorded brain electrical signal data from the contralateral cortex region and the subthalamic nucleus region of the brain of the subject after applying the electrical stimulation with the DBS device to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the settings predicted to have the highest stimulation effectiveness based on the updated linear classification model; and i) repeating e) - h) to adjust the available settings of the DBS device to optimize stimulation effectiveness. In some embodiments, the linear classification model uses linear discriminant analysis (LDA) to adjust amplitude of current and frequency of the electrical stimulation.

[0137] In certain embodiments, the computer implemented method further comprises splitting the recorded brain electrical signal data into consecutive time epochs. In some embodiments, each time epoch comprises 0.5 second to 10 minutes of time of the recorded brain electrical signal data, including any amount of time within this range such as 0.5 seconds, 0.75 seconds, 1 second, 2 seconds, 3 seconds, 4 seconds, 5 seconds, 6 seconds, 7 seconds, 8 seconds, 9 seconds, 10 seconds, 15 seconds, 20 seconds, 25 seconds, 30 seconds, 35 seconds, 40 seconds, 45 seconds,50 seconds, 55 seconds, 1 minute, 2 minutes, 3 minutes, 4 minutes, 5 minutes, 6 minutes, 7 minutes, 8 minutes, 9 minutes, or 10 minutes of time. In certain embodiments, the computer implemented method further comprises assigning label of intended movement state or stationary state to each time epoch.

[0138] In some embodiments, the computer implemented method further comprises training the linear model to classify each time epoch based on whether the subject is in an intended movement state or a stationary state by analyzing the recorded brain electrical signal data using a non-linear model. In some embodiments, canonical power bands are used as feature inputs to train the linear classification model to classify each time epoch as to whether the subject is in an intended movement state or a stationary state using linear discriminant analysis. In some embodiments, personalized power bands are used as feature inputs to train the linear classification model to classify each time epoch as to whether the subject is in an intended movement state or a stationary state using linear discriminant analysis. In some embodiments, field potentials are used as feature inputs to train the linear classification model to classify each time epoch as to whether the subject is in an intended movement state or a stationary state using linear discriminant analysis.

[0139] In certain embodiments, the computer implemented method further comprises storing a user profile for the subject comprising information regarding the recorded brain electrical signal data including brain electrical signals from neural activity associated with intended movement. In certain embodiments, the computer implemented method further comprises storing a user profile for the subject comprising information regarding the programmed stimulation parameters used to apply electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject to assist movement of the subject based on the recorded brain electrical signal data.

[0140] The methods described herein can be implemented in digital electronic circuitry, or in computer software, firmware, or hardware. The disclosed and other embodiments can be implemented as one or more computer program products, i.e., one or more modules of computer program instructions encoded on a computer readable medium for execution by, or to control the operation of, a data processing apparatus. The computer readable medium can be a machine- readable storage device, a machine-readable storage substrate, a memory device, a composition of matter effecting a machine-readable propagated signal, or any combination thereof.

[0141] A computer program (also known as a program, software, software application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module,component, subroutine, or other unit suitable for use in a computing environment. A computer program does not necessarily correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub programs, or portions of code). A computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.

[0142] In a further aspect, the system for performing the computer implemented method, as described, may include a computer containing a processor, a storage component (i.e., memory), a display component, and other components typically present in general purpose computers. The storage component stores information accessible by the processor, including instructions that may be executed by the processor and data that may be retrieved, manipulated or stored by the processor.

[0143] The processor and / or memory may be operably connected to a display device, for example, via a wired, such as a Universal Serial Bus (USB) connection, or wireless connection, such as a Bluetooth connection. Any convenient display device, such as a liquid crystal display (LCD), lightemitting diode (LED) display, plasma (PDP) display, quantum dot (QLED) display or cathode ray tube display device may be used. The display component may display information regarding whether an intended movement state or a stationary state is detected for the subject, information about brain activity associated with intended movement, current stimulation parameters, or recommended changes to the stimulation parameters.

[0144] The storage component may be of any type capable of storing information accessible by the processor, such as a hard-drive, memory card, ROM, RAM, DVD, CD-ROM, USB Flash drive, write- capable, and read-only memories. The processor may be a general purpose processor, a graphics processor unit, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein.

[0145] A general purpose processor can be a microprocessor, but in the alternative, the processor can be a controller, microcontroller, or state machine, combinations of the same, or the like. A processor can also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration. Although described herein primarilywith respect to digital technology, a processor can also include primarily analog components. A computing environment can include any type of computer system, including, but not limited to, a computer system based on a microprocessor, a graphics processor unit, a mainframe computer, a digital signal processor, a portable computing device, a personal organizer, a device controller, and a computational engine within an appliance, to name a few.

[0146] The steps of a method, process, or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module, engine, and associated databases can reside in memory resources such as in RAM memory, FRAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of non-transitory computer-readable storage medium, media, or physical computer storage known in the art. An exemplary storage medium can be coupled to the processor such that the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium can be integral to the processor. The processor and the storage medium can reside in an ASIC. The ASIC can reside in a user terminal. In the alternative, the processor and the storage medium can reside as discrete components in a user terminal.

[0147] The instructions may be any set of instructions to be executed directly (such as machine code) or indirectly (such as scripts) by the processor. In that regard, the terms "instructions," "steps" and "programs" may be used interchangeably herein. The instructions may be stored in object code form for direct processing by the processor, or in any other computer language including scripts or collections of independent source code modules that are interpreted on demand or compiled in advance.

[0148] Data may be retrieved, stored or modified by the processor in accordance with the instructions. For instance, although the system is not limited by any particular data structure, the data may be stored in computer registers, in a relational database as a table having a plurality of different fields and records, XML documents, or flat files. The data may also be formatted in any computer-readable format such as, but not limited to, binary values, ASCII or Unicode. Moreover, the data may comprise any information sufficient to identify the relevant information, such as numbers, descriptive text, proprietary codes, pointers, references to data stored in other memories (including other network locations) or information which is used by a function to calculate the relevant data.

[0149] In certain embodiments, the processor and storage component may comprise multiple processors and storage components that may or may not be stored within the same physicalhousing. For example, some of the instructions and data may be stored on removable CD-ROM and others within a read-only computer chip. Some or all of the instructions and data may be stored in a location physically remote from, yet still accessible by, the processor. Similarly, the processor may comprise a collection of processors which may or may not operate in parallel.

[0150] In some embodiments, the method can be performed using a cloud computing system. In these embodiments, the recorded brain electrical signal data from the subthalamic nucleus region and / or contralateral cortex region of the brain of the subject and the programming can be exported to a cloud computer, which runs the program, and returns an output to the user.

[0151] Components of systems for carrying out the presently disclosed methods are further described in the examples below.Systems

[0152] The present disclosure also provides systems which find use, e.g., in practicing the subject methods. The system may be an open-loop or closed-loop system configured for performing the methods provided herein. In some embodiments, the system may include a DBS electrode adapted for positioning at a location in a subthalamic nucleus region or a globus pallidus internus region of the brain of the subject to deliver electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region and a neural recording electrode adapted for positioning at a subthalamic nucleus region and / or a contralateral cortex region (e.g., a central sulcus region, precentral gyrus region, or postcentral gyrus region) of the brain of the subject to record brain electrical signal data, including brain electrical signals from neural activity associated with intended movement before, during, or after electrical stimulation is applied to the brain. In a closed-loop system, the system may also include a computing means and control unit programmed to instruct a DBS electrode to apply electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject in a manner effective to provide a prokinetic effect on the intended movement of the subject when brain electrical signals from neural activity associated with intended movement are detected using a neural recording electrode. The recorded brain electrical signal data, including the brain electrical signals from neural activity associated with intended movement is analyzed using a movement classification model that identifies brain electrical signals from neural activity associated with intended movement in the recorded brain electrical signal data and adjusts one or more programmed stimulation parameters based on the recorded brain electrical signal data according to an algorithm control law; and automatically delivers electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject viathe control unit, neurostimulator pulse generator and DBS electrode in a manner effective to provide a prokinetic effect on the intended movement of the patient if the electrical signal metrics indicate that the patient needs assistance with the movement. In some embodiments, one or more programmed stimulation parameters are modulated according to the algorithm’s control law based on the recorded electrical activity data, and modulated electrical stimulation is delivered to the brain via the control unit, pulse generator and DBS electrode in a manner effective to provide a prokinetic effect on the intended movement. The closed loop system may include an on-body pulse generator that is connected to the implanted DBS electrodes and hence can apply electrical stimulation to the brain automatically upon receiving a communication from the control unit or a cranially mounted neurostimulator that can also sense cortical neural signals through electrodes mounted on the case of the device.

[0153] The processor of the closed-loop system may run programming for assessing the effectiveness of treatment and modulate a parameter of the treatment as needed without user intervention. Thus, the closed-loop system may not necessarily include a user interface for a user to instruct the DBS electrode to apply an electrical stimulation to the brain to assist movement of the subject. However, in some embodiments, a user interface may be included in the closed-loop system which may be used to confirm the recommendation of the closed loop system, or to override it, or to change the recommendation.

[0154] In certain aspects, a control algorithm for the methods and systems of the present disclosure may include steps of comparing an electrical signal from a region of the brain of a subject to a normal or reference electrical signal (e.g., substantially free of bradykinesia), wherein when the electrical signal is significantly different from the normal or reference electrical signal, the control algorithm includes steps of directing a device to apply electrical stimulation to the brain of the subject, followed by measurement of electrical signals from the region of the brain and comparing it to a normal or reference electrical signal, wherein when the measured signal is significantly different from a normal or reference electrical signal, the algorithm includes the step of applying another electrical stimulation to the brain.

[0155] In some embodiments, the control algorithm utilizes a machine learning algorithm to analyze inputted brain electrical activity data to automate detection of brain activity features, including brain electrical signals from neural activity associated with intended movement. The control algorithm then directs a device to apply electrical stimulation to the brain of the subject if the brain activity features indicate the subject is in an intended movement state that should be treated with electrical stimulation to assist movement. For example, a machine learning algorithm may be used to correlate the levelsof overall power, or power in specific frequency ranges (e.g., alpha, delta, beta, gamma, and / or theta) with intended movement that should be assisted with deep brain electrical stimulation. In some embodiments, field potential data are fit to a movement classification model that distinguishes between when the subject is in an intended movement state or a stationary state to determine how to adjust one or more programmed stimulation parameters (e.g., increase stimulation during an intended movement state and decrease or discontinue stimulation during a stationary state). In certain embodiments, the machine learning algorithm further determines whether the brain electrical signals from neural activity associated with intended movement are better measured by the first neural recording electrode in the subthalamic nucleus region or the second neural recording electrode in the contralateral cortex region for use in the classification to distinguish between an intended movement state and a stationary state. In certain embodiments the algorithm provides updated optimal stimulation setting recommendations to the clinician for guiding programing and decision making.

[0156] In certain embodiments, the system further comprises a user interface comprising an input electronically coupled to a processor for instructing a DBS electrode to apply an electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region to assist movement of a subject. In some embodiments, the user interface is password protected and is operable by a health care practitioner.

[0157] In some embodiments, the system further comprises a wearable monitor that can acquire accelerometry, gyroscope and / or surface electromyographic (sEMG) data of the subject to detect movement of the subject. In some embodiments, the wearable monitor is a wrist-worn monitor such as a smartwatch. For example, a wrist-watch style wearable monitor such as the Parkinson’s KinetiGraph®, PKG®, available from PKG Health (San Francisco, GA), can be used to monitor movement continuously to detect various motor symptoms of a movement disorder such as bradykinesia, dyskinesia, tremor, daytime immobility, stiffness, slow movements, gait / walking, daytime somnolence, and sleep fragmentation. An Apple Watch is available from Apple Inc. (Cupertino, CA), which can be used, for example, for monitoring, bradykinesia, dyskinesia, tremors, gait, and arm movement. In addition, the system may further comprise one or more video recording devices capable of providing multi-view video recordings of the subject for monitoring movement.

[0158] Components of systems for carrying out the presently disclosed methods are further described in the examples below.Administration of a Pharmacological Agent

[0159] Embodiments of the methods and systems provided in this disclosure may also include administration of an effective amount of at least one pharmacological agent. By “effective amount” is meant a dosage sufficient to treat a movement disorder or neurological disorder associated with development of bradykinesia in a subject. In some embodiments, the bradykinesia is caused by a movement disorder or a neurological disorder such as Parkinson’s disease, Parkinsonism, Lewy body dementia, progressive supranuclear palsy, corticobasal degeneration, Huntington's disease, dystonia, stroke, or multiple system atrophy. The effective amount will vary somewhat from subject to subject, and may depend upon factors such as the age and physical condition of the subject, type of movement disorder, any concurrent treatment, the form of the agent, the pharmaceutically acceptable carrier used if any, the route and method of delivery, and analogous factors within the knowledge and expertise of those skilled in the art. Appropriate dosages may be determined in accordance with routine pharmacological procedures known to those skilled in the art, as described in greater detail below.

[0160] If a pharmacological approach is employed in the treatment of a movement disorder, the specific nature and dosing schedule of the agent will vary depending on the particular nature of the disorder to be treated. Representative pharmacological agents that may find use in treatment of Parkinson’s disease may include, but are not limited to, L-DOPA (l-3,4-dihydroxyphenylalanine, also known as levodopa), carbidopa (N-amino-cc-methyl-3-hydroxy-L-tyrosine monohydrate), carbidopa- levodopa (Rytary, Sinemet, Duopa), a dopamine agonist, including, without limitation, pramipexole (Mirapex ER), rotigotine, apomorphine (Apokyn), and amantadine (Gocovri); a monoamine oxidase B (MAO-B) inhibitor, including, without limitation, selegiline (Zelapar), rasagiline (Azilect) and safinamide (Xadago); a catechol O-methyltransferase (COMT) inhibitor, including, without limitation, entacapone (Comtan), opicapone (Ongentys), and Tolcapone (Tasmar); an anticholinergic agent, including, without limitation, benztropine (Cogentin) and trihexyphenidyl; an adenosine receptor antagonist, including, without limitation an A2A receptor antagonist such as istradefylline (Nourianz), or an antipsychotic, including, without limitation, nuplazid (Pimavanserin), or any combination thereof.

[0161] In certain aspects, the administration of a pharmacological agent involves using a pharmacological delivery device such as, but not limited to, pumps (implantable or external devices), epidural injectors, syringes or other injection apparatus, catheter and / or reservoir operatively associated with a catheter, etc. For example, in certain embodiments a delivery device employed to deliver at least one pharmacological agent to a subject may be a pump, syringe, catheter or reservoiroperably associated with a connecting device such as a catheter, tubing, or the like. Containers suitable for delivery of at least one pharmacological agent to a pharmacological agent administration device include instruments of containment that may be used to deliver, place, attach, and / or insert the at least one pharmacological agent into the delivery device for administration of the pharmacological agent to a subject and include, but are not limited to, vials, ampules, tubes, capsules, bottles, syringes and bags. Administration of a pharmacological agent may be performed by a user or by a closed loop system.Utility

[0162] The methods and systems of the present disclosure find use in the treatment of a subject for bradykinesia by using adaptive deep brain stimulation. Closed-loop stimulation can be finely targeted and tuned in a personalized manner to provide a prokinetic effect on intended movement of the subject. The subject methods can be used to treat bradykinesia caused by a movement disorder or a neurological disorder such as, but not limited to, Parkinson’s disease, Parkinsonism, Lewy body dementia, progressive supranuclear palsy, corticobasal degeneration, Huntington's disease, dystonia, stroke, or multiple system atrophy.

[0163] Efficacy of the treatment may be measured in an art accepted manner such as, by using a Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) or a Hoehn and Yahr (HnY) scale, a Parkinson’s Disease Composite Scale (PDCS), or a Schwab and England Activities of Daily Living (ADL) Scale. In some embodiments, assessing effectiveness of the treatment of bradykinesia comprises analyzing multi-view video recordings of the subject or keylogging data from a computer used by the subject, or monitoring the subject using a wearable monitor that can acquire accelerometry, gyroscope, and / or surface electromyographic (sEMG) data. A wrist-watch style wearable monitor such as the Parkinson’s KinetiGraph®, PKG® is available from PKG Health (San Francisco, CA), which can monitor movement continuously and detect various motor symptoms of a movement disorder such as bradykinesia, dyskinesia, tremor, daytime immobility, stiffness, slow movements, gait / walking, daytime somnolence, and sleep fragmentation. An Apple Watch is available from Apple Inc. (Cupertino, CA), which can be used, for example, for monitoring, bradykinesia, dyskinesia, tremors, gait, and arm movement.Examples of Non-Limiting Aspects of the Disclosure

[0164] Aspects, including embodiments, of the present subject matter described above may be beneficial alone or in combination, with one or more other aspects or embodiments. Without limitingthe foregoing description, certain non-limiting aspects of the disclosure numbered 1-77 are provided below. As will be apparent to those of skill in the art upon reading this disclosure, each of the individually numbered aspects may be used or combined with any of the preceding or following individually numbered aspects. This is intended to provide support for all such combinations of aspects and is not limited to combinations of aspects explicitly provided below.1. A method for treating bradykinesia in a subject using deep brain stimulation, the method comprising: positioning a stimulation electrode at a first location in a subthalamic nucleus region or a second location in a globus pallidus internus region of the brain of the subject to deliver electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region; positioning a first neural recording electrode at a third location in the subthalamic nucleus region of the brain of the subject and a second neural recording electrode at a fourth location in a contralateral cortex region of the brain of the subject to record brain electrical signal data associated with an intended movement of the subject; detecting a brain electrical signal associated with the intended movement using the first neural recording electrode or the second neural recording electrode; and applying electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the stimulation electrode in a manner effective to provide a prokinetic effect on the intended movement of the subject when the brain electrical signal associated with the intended movement is detected using the first neural recording electrode or the second neural recording electrode.2. The method of aspect 1 , wherein the contralateral cortex region comprises a precentral gyrus region, a postcentral gyrus region, or both the precentral gyrus region and the postcentral gyrus region.3. The method of aspect 1 , wherein the contralateral cortex region comprises the central sulcus.4. The method of any one of aspects 1-3, wherein the electrical stimulation reduces the bradykinesia and increases vigor of the intended movement compared to in absence of the electrical stimulation.5. The method of any one of aspects 1 -4, wherein the electrical stimulation is applied unilaterally or bilaterally.6. The method of any one of aspects 1 -5, wherein the brain electrical signal comprises alpha frequency, beta frequency, gamma frequency, or theta frequency neural oscillations.7. The method of aspect 6, wherein the theta frequency neural oscillations are in a range from 4 Hz to 7 Hz, the alpha frequency neural oscillations are in a range from 8 Hz to 12 Hz, the beta frequency neural oscillations are in a range from 13 Hz to 30 Hz, and the gamma frequency neural oscillations are in a range from 60 Hz to 90 Hz.8. The method of any one of aspects 1-7, wherein the brain electrical signal data comprises field potential data.9. The method of any one of aspects 1 -8, further comprising using a control algorithm to automate said applying electrical stimulation when the brain electrical signal associated with the intended movement is detected.10. The method of aspect 9, wherein the control algorithm uses a machine learning algorithm for movement classification.11. The method of aspect 10, wherein the machine learning algorithm is a supervised machine learning algorithm.12. The method of aspect 10 or 11 , wherein the movement classification distinguishes between when the subject is in an intended movement state or a stationary state.13. The method of aspect 12, further comprising training a machine learning model to predict whether the subject is in an intended movement state using accelerometry data for detected movement of the subject.14. The method of aspect 13, wherein the accelerometry data is from a wearable accelerometer worn by the subject that detects natural undirected movement of the subject.15. The method of aspect 13, wherein the accelerometry data is from a wearable accelerometer worn by the subject that detects movement during directed movement tasks selected from finger tapping, opening or closing of a hand, wrist pronation or supination, walking, and keyboard typing.16. The method of any one of aspects 12-15, further comprising performing principal components (PC) decomposition on power spectral data to identify personalized power band features for distinguishing the intended movement state from the stationary state.17. The method of aspect 16, further comprising: transforming a time-domain of a local field potential (LFP) into a spectrogram using a short- time fourier transform (STFT); performing principal components analysis (PCA) on z-scored STFT outputs to identify coordinated bands of neural activity; and applying a peak-finding algorithm to PCA component weights to find a frequency range best approximating each principal component.18. The method of aspect 17, wherein top principal components for the contralateral cortex region or the subthalamic nucleus region correspond to personalized frequency ranges for beta frequency neural oscillations and gamma frequency neural oscillations for the brain electrical signal associated with the intended movement of the subject.19. The method of any one of aspects 12-18, wherein the electrical stimulation is selectively increased during the intended movement state and not during the stationary state.20. The method of any one of aspects 12-19, further comprising using accelerometry data, gyroscope data, multi-view video recordings of the subject, keylogging data from a computer used by the subject, or any combination thereof in combination with the recorded brain electrical signal data to distinguish between the intended movement state and the stationary state.21 . The method of aspect 20, wherein a wearable device worn by the subject is used to collect the accelerometry data, the gyroscope data, or a combination thereof.22. The method of aspect 21 , wherein the wearable device is a smartwatch.23. The method of any one of aspects 12-22, wherein the control algorithm further uses linear discriminant analysis (LDA) to adjust stimulation amplitude or frequency of the electrical stimulation.24. The method of any one of aspects 1 -23, wherein the intended movement is a finger tap, opening or closing of a hand, wrist pronation or supination, walking, or keyboard typing.25. The method of any one of aspects 1 -24, wherein the electrical stimulation is optimized for each hand of the subject independently.26. The method of any one of aspects 1 -25, wherein the stimulation electrode is placed on a surface of the subthalamic nucleus region or the globus pallidus internus region.27. The method of any one of aspects 1 -26, wherein the first neural recording electrode is placed on a surface of the subthalamic nucleus region.28. The method of any one of aspects 1 -27, wherein the second neural recording electrode is placed on a surface of the contralateral cortex region.29. The method of any one of aspects 1 -28, wherein the stimulation electrode is a nonbrain penetrating surface electrode array or a brain-penetrating electrode array.30. The method of any one of aspects 1 -29, wherein the first neural recording electrode and / or the second neural electrode is a non-brain penetrating surface electrode array or a brainpenetrating electrode array.31 . The method of any one of aspects 1 -30, wherein the first neural recording electrode and / or the second neural electrode is an electroencephalogram (EEG) electrode array, a subgalealor burrhole mounted or cranially mounted neurostimulator electrode, or an electrocorticogram (ECoG) electrode array.32. The method of aspect 31 , wherein the ECoG electrode array spans regions of the precentral gyrus and the postcentral gyrus region that control hand and arm movement.33. The method of aspect 31 , wherein the ECoG electrode array spans the central sulcus.34. The method of any one of aspects 1 -33, wherein the subject is further administered dopaminergic medication.35. The method of any one of aspects 1 -34, further comprising assessing effectiveness of the treatment in the subject.36. The method of aspect 35, wherein said assessing comprises using behavioral data obtained of the subject.37. The method of aspect 36, wherein the behavior data is accelerometry data, gyroscope data, video-based pose kinematic data for the subject, or keylogging data from a computer used by the subject, or a combination thereof.38. The method of any one of aspects 35-37, wherein said assessing comprises using a Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) or a Hoehn and Yahr (HnY) scale, a Parkinson’s Disease Composite Scale (PDCS), or a Schwab and England Activities of Daily Living (ADL) Scale.39. The method of any one of aspects 1 -38, wherein the method is performed at a site remote from a hospital.40. The method of aspect 39, wherein the method is performed at the subject’s home.41 . The method of any one of aspects 1 -40, wherein the subject has Parkinson’s disease, Parkinsonism, Lewy body dementia, progressive supranuclear palsy, corticobasal degeneration, Huntington's disease, dystonia, stroke, or multiple system atrophy.42. A computer implemented method for programming a deep brain stimulator to treat bradykinesia in a subject, the computer performing steps comprising: receiving recorded brain electrical signal data from a contralateral cortex region and a subthalamic nucleus region of the brain of the subject; analyzing the recorded brain electrical signal data using a movement classification model that identifies patterns of electrical signals in the recorded brain electrical signal data associated with intended movement of the subject; adjusting one or more programmed stimulation parameters based on the recorded brain electrical signal data according to an algorithm control law; and instructing the deep brain stimulator to apply an electrical stimulation to the subthalamic nucleus region or a globus pallidus internus region of the brain to provide a prokinetic effect on the intended movement of the subject.43. The computer implemented method of aspect 42, wherein the contralateral cortex region comprises a precentral gyrus region, a postcentral gyrus region, or both the precentral gyrus region and the postcentral gyrus region.44. The computer implemented method of aspect 42, wherein the contralateral cortex region comprises the central sulcus.45. The computer implemented method of any one of aspects 42-44, wherein the electrical stimulation reduces bradykinesia and increases vigor of the intended movement compared to in absence of the electrical stimulation.46. The computer implemented method of any one of aspects 42-45, wherein the brain electrical signal data comprises field potential data.47. The computer implemented method of any one of aspects 42-46, wherein a machine learning algorithm is used to generate the movement classification model.48. The computer implemented method of aspect 47, wherein the machine learning algorithm is a supervised machine learning algorithm.49. The computer implemented method of aspect 47 or 48, wherein the movement classification model distinguishes between when the subject is in an intended movement state or a stationary state.50. The computer implemented method of aspect 49, further comprising training a machine learning model to predict whether the subject is in an intended movement state using accelerometry data for detected movement of the subject.51 . The computer implemented method of aspect 50, wherein the accelerometry data is from a wearable accelerometer worn by the subject that detects natural undirected movement of the subject.52. The computer implemented method of aspect 50, wherein the accelerometry data is from a wearable accelerometer worn by the subject that detects movement during directed movement tasks selected from finger tapping, opening or closing of a hand, wrist pronation or supination, walking, and keyboard typing.53. The computer implemented method of any one of aspects 49-52, further comprising performing principal components (PC) decomposition on power spectral data to identify personalized power band features for distinguishing the intended movement state from the stationary state.54. The computer implemented method of aspect 53, further comprising: transforming a time-domain of a local field potential (LFP) into a spectrogram using a short- time fourier transform (STFT); performing principal components analysis (PCA) on z-scored STFT outputs to identify coordinated bands of neural activity; and applying a peak-finding algorithm to PCA component weights to find a frequency range best approximating each principal component.55. The computer method of aspect 54, wherein top principal components for the contralateral cortex region or the subthalamic nucleus region correspond to personalized frequency ranges for beta frequency neural oscillations and gamma frequency neural oscillations for the brain electrical signal associated with the intended movement of the subject.56. The computer implemented method of any one of aspects 49-55, wherein the electrical stimulation is selectively increased during the intended movement state and not during the stationary state.57. The computer implemented method of any one of aspects 49-56, further comprising using accelerometry data, gyroscope data, multi-view video recordings of the subject, keylogging data from a computer used by the subject, or any combination thereof in combination with the recorded brain electrical signal data to distinguish between the intended movement state and the stationary state.58. The computer implemented method of any one of aspects 42-57, further comprising: a) ranking predicted stimulation effectiveness for available settings of a DBS device based on classifier scores for stimulation effectiveness of each setting using a linear classification model; b) selecting stimulation settings predicted to have highest stimulation effectiveness based on the linear classification model; c) receiving recorded brain electrical signal data from the contralateral cortex region and the subthalamic nucleus region of the brain of the subject after applying electrical stimulation with the DBS device to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the settings predicted to have the highest stimulation effectiveness; d) analyzing the recorded brain electrical signal data to evaluate neural response of the subject to the electrical stimulation; e) updating the linear classification model based on the neural response of the subject to the electrical stimulation to generate an updated linear classification model; f) updating the ranking of predicted stimulation effectiveness for the available settings of the DBS device using the updated linear classification model;g) selecting stimulation settings predicted to have the highest stimulation effectiveness based on the updated linear classification model; h) receiving recorded brain electrical signal data from the contralateral cortex region and the subthalamic nucleus region of the brain of the subject after applying the electrical stimulation with the DBS device to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the settings predicted to have the highest stimulation effectiveness based on the updated linear classification model; and i) repeating e) - h) to adjust the available settings of the DBS device to optimize stimulation effectiveness.59. The computer implemented method of aspect 58, wherein the linear classification model uses linear discriminant analysis (LDA) to adjust amplitude of current and frequency of the electrical stimulation.60. The computer implemented method of any one of aspects 42-59, wherein the subject has Parkinson’s disease, Parkinsonism, Lewy body dementia, progressive supranuclear palsy, corticobasal degeneration, Huntington's disease, dystonia, stroke, or multiple system atrophy.61 . A non-transitory computer-readable medium comprising program instructions that, when executed by a processor in a computer, causes the processor to perform the method of any one of aspects 42-60.62. A kit comprising the non-transitory computer-readable medium of aspect 61 and instructions for determining symptom severity of a subject having a movement disorder.63. A system for treating bradykinesia in a subject, the system comprising: a stimulation electrode adapted for positioning at a first location in a subthalamic nucleus region or a second location in a globus pallidus internus region of the brain of the subject to deliver electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region; a first neural recording electrode adapted for positioning at a third location in the subthalamic nucleus region of the brain of the subject to record brain electrical signal data; a second neural recording electrode adapted for positioning at a fourth location in a contralateral cortex region of the brain of the subject to record brain electrical signal data; anda processor programmed according to the computer implemented method of any one of aspects 42-60 to instruct the stimulation electrode to apply an electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject in a manner effective to provide a prokinetic effect on the intended movement of the subject when the brain electrical signal associated with the intended movement is detected using the first neural recording electrode or the second neural recording electrode.64. The system of aspect 63, wherein the brain electrical signal data comprises field potential data.65. The system of 63 or 64, further comprising an accelerometer, a gyroscope, or a combination thereof.66. The system of aspect 65, wherein the accelerometer, the gyroscope, or the combination thereof is provided by a wearable device.67. The system of aspect 66, wherein the wearable device is a smartwatch.68. The system of any one of aspects 63-67, further comprising a video recording device.69. The system of any one of aspects 63-68, wherein the contralateral cortex region comprises a precentral gyrus region, a postcentral gyrus region, or both the precentral gyrus region and the postcentral gyrus region.70. The system of any one of aspects 63-68, wherein the contralateral cortex region comprises the central sulcus.71 . The system of any one of aspects 63-70, wherein the stimulation electrode is a nonbrain penetrating surface electrode array or a brain-penetrating electrode array.72. The system of any one of aspects 63-71 , wherein the first neural recording electrode or the second neural electrode is a non-brain penetrating surface electrode array or a brainpenetrating electrode array.73. The system of any one of aspects 63-72, wherein the first neural recording electrode or the second neural electrode is an electroencephalogram (EEG) electrode array, a subgaleal or burrhole mounted or cranially mounted neurostimulator electrode, or an electrocorticogram (ECoG) electrode array.74. The system of aspect 73, wherein the ECoG electrode array spans regions of the precentral gyrus and the postcentral gyrus region that control hand and arm movement.75. The system of aspect 73, wherein the ECoG electrode array spans the central sulcus.76. The system of any one of aspects 63-75, wherein the system further comprises a user interface comprising an input electronically coupled to the processor for instructing the stimulation electrode to apply an electrical stimulation to the subthalamic nucleus region to treat the bradykinesia in the subject.77. The system of aspect 76, wherein the user interface is password protected and is operable by a health care practitioner.

[0165] It will be apparent to one of ordinary skill in the art that various changes and modifications can be made without departing from the spirit or scope of the invention.EXPERIMENTAL

[0166] The following examples are put forth so as to provide those of ordinary skill in the art with a complete disclosure and description of how to make and use the present invention, and are not intended to limit the scope of what the inventors regard as their invention nor are they intended to represent that the experiments below are all or the only experiments performed. Efforts have been made to ensure accuracy with respect to numbers used (e.g. amounts, temperature, etc.) but some experimental errors and deviations should be accounted for. Unless indicated otherwise, parts are parts by weight, molecular weight is weight average molecular weight, temperature is in degrees Centigrade, and pressure is at or near atmospheric.

[0167] All publications and patent applications cited in this specification are herein incorporated by reference as if each individual publication or patent application were specifically and individually indicated to be incorporated by reference.

[0168] The present invention has been described in terms of particular embodiments found or proposed by the present inventor to comprise preferred modes for the practice of the invention. It will be appreciated by those of skill in the art that, in light of the present disclosure, numerous modifications and changes can be made in the particular embodiments exemplified without departing from the intended scope of the invention. For example, due to codon redundancy, changes can be made in the underlying DNA sequence without affecting the protein sequence. Moreover, due to biological functional equivalency considerations, changes can be made in protein structure without affecting the biological action in kind or amount. All such modifications are intended to be included within the scope of the appended claims.Example 1Movement-Responsive Deep Brain Stimulation for Parkinson’s Disease Using a Remotely Optimized Neural DecoderINTRODUCTION

[0169] Deep brain stimulation (DBS) has become a widely adopted treatment for patients with advanced Parkinson’s Disease (PD) and medication refractory symptoms. Conventional DBS (eDBS) delivers constant electrical pulses to the basal ganglia at fixed amplitudes and frequencies. While the exact mechanisms remain unclear [Ashkan et al., 2017; McGregor et al., 2019], eDBS has been shown to alleviate movement slowness (bradykinesia, the cardinal symptom of PD) and tremor, while reducing involuntary movements (dyskinesia) that arise as a side-effect of dopaminergic medication [Deuschi et al., 2006; Follett et al., 2010; Odekerken et al., 2013]. However, residual fluctuations due to medication and activity / arousal levels often persist [Pozzi and Isaias, 2022], as the static nature of eDBS fails to address these dynamic clinical needs. This can create challenging trade-offs between maximizing clinical efficacy and minimizing side-effects when programming eDBS therapy.

[0170] Adaptive DBS (aDBS) seeks to overcome the limitations of eDBS by modulating stimulation in accordance with real-time clinical, physiological, or behavioral states. These states can be tracked through local field potential (LFP) biomarkers recorded from the brain. Beta-band (13-30Hz) LFPactivity in the basal ganglia, which covaries with PD symptom severity and medication state [Kuhn et al., 2006; Neumann et al., 2016; Lofredi et al., 2023], has been used as the control signal in the majority of aDBS algorithms previously tested [Little and Brown, 2020]. The earliest approaches to be developed treated beta as a trigger for on-off stimulation control, targeting sub-second bursts of activity with putative roles in pathological brain circuit dynamics [Little et al., 2013; Little et al., 2016; Tinkhauser et aL, 2017; He et aL, 2023]. These studies using “fast” algorithms have provided collective evidence that aDBS may increase efficacy and reduce side-effects through more selective stimulation. Later approaches have begun considering longer timescales and proportional control algorithms more suitable for counteracting the gradual fluctuations of concurrent medication cycles [Rosa et aL, 2017; Arlotti et al., 2018; Velisar et aL, 2019; Gilron et aL, 2021 a]. Both fast and slow beta algorithms are currently in multi-centre clinical trials [ADAPT-PD: NCT04547712], Notably, however, this same beta signal is reduced during movement [Kuhn et aL, 2004; Eisinger et aL, 2020]. This raises concern that beta-driven aDBS will have the unintended consequence of reducing stimulation during movement [Johnson et aL, 2016; Iturrate et aL, 2019; Little and Brown, 2020; He et aL, 2023], which is precisely when the prokinetic effects of DBS are theoretically necessary. Broader exploration of biomarkers and novel algorithm design will be required to navigate these problematic interactions between symptoms, biomarkers, and movement.

[0171] While most aDBS algorithms attempt to identify latent clinical states that reflect probability of symptoms, an alternative approach would be to target movement itself. Bradykinesia in PD characteristically involves difficulty initiating voluntary movement with progressive reduction in vigor during repetitive action [Bologna et aL, 2020]. We therefore sought to develop a movement- responsive aDBS algorithm that selectively increases stimulation during movement for treating bradykinesia, while reducing stimulation during rest to mitigate dyskinesia (FIGS. 1 A-1 B). The gating theory of basal ganglia function in PD pathophysiology [Chevalier and Deniau, 1990; Ivry and Spencer, 2004] provides a theoretical mechanism of action supporting this approach. Under healthy circumstances, the basal ganglia inhibit movement until an action has been selected and released [Klaus et aL, 2019]. In PD, excessive inhibition prevents release, leading to slow, difficult-to-initiate movement. By decoding motor intent to guide targeted stimulation, movement-responsive aDBS would selectively disinhibit movement when it is intended. This approach has been tested in a nonhuman primate model of PD using cortical gamma to indicate movement, and showed therapeutic benefits comparable to conventional stimulation while reducing overall current delivery [Darbin et aL, 2022], Similarly motivated movement- and postural-state dependent aDBS has also been exploredfor essential tremor with either cortical [Herron et al., 2017; Ferleger et aL, 2020; Opri et al., 2020] or thalamic [He et al., 2021 ] algorithm inputs.

[0172] In the present study, we report the first fully embedded movement-responsive aDBS in a human subject with PD. We first remotely optimized neural decoders using a custom machine learning pipeline seeking to predict movement from cortical and subthalamic LFP’s. Algorithm tuning was performed in a data-driven, automated fashion using data collected independently by the patient in their own home without any direct researcher involvement. We then assessed offline performance to evaluate the pipeline, in particular the benefit associated with including cortical signals and using a custom power band personalization method. Finally, we used the optimized algorithms to drive movement-responsive aDBS and compared it to conventional DBS and a control aDBS condition, demonstrating comparable or superior efficacy across all categories of motor evaluation, including movement speed, dyskinesia, and subjective patient reports of therapeutic quality.RESULTSEmbedded aDBS functionality and algorithm tuning overview

[0173] One subject with PD who had previously received bilateral DBS implants with embedded adaptive capabilities (Medtronic RC+S, FIG. 1 C) was recruited for the study. The hardware in each hemisphere consisted of a depth lead for stimulating and sensing in the subthalamic nucleus (STN), a sensorimotor electrocorticography (ECoG) strip for sensing over the pre-central (PreC, motor) and post-central (PostC, somatosensory) cortices, and a neurostimulator for driving adaptive stimulation using LFP’s from the three sensed brain areas. The onboard algorithm guiding stimulation used linear discriminant analysis with additional layered operations for controlling the state transition dynamics (Supplementary Table 1 ) [Sellers et aL, 2021], LFP’s were first transformed into power bands with programmable frequency limits and smoothing properties, then up to four power bands were linearly combined and passed through a series of threshold-based logic operations to define the algorithm state. The present study aimed to optimize the algorithm parameters for decoding movement of the contralateral arm in order to deliver targeted stimulation increases when movement was desired (FIGS. 1 A-1 B).

[0174] A remote data collection platform [Strandquist et aL, 2023] was deployed in the participant’s home office to collect model training data and evaluate both decoding and therapeutic performance. In addition to neural data, we recorded accelerometer and gyroscope data from bilateral Apple Watches (thresholded to label moving or stationary), multi-view video recordings, and keyloggingfrom the participant’s home computer (FIG. 1C). Five days of data were collected for model training and validation, while a sixth day was dedicated for model testing. Each recording day lasted approximately 25 minutes, during which the participant performed a self-guided series of clinically standardized movement tasks using both hands (Unified Parkinson's Disease Rating Scale Part III: rest tremor, finger tapping, hand movements - open-close, pronation-supination) as well as a typing task (FIGS. 2A,2B).

[0175] Given the unique architecture of the neurostimulators’ algorithms, a custom ML pipeline was developed for optimization (FIG. 20). In brief, we first fit coefficient weights for predicting continuousvalued watch accelerometry from neural inputs. Using Bayesian optimization, we then tuned the parameters that define the algorithm dynamics and translate the accelerometry predictions into binary movement state predictions. Finally, we applied a feature selection routine to search for the optimal set of neural inputs to the algorithm.Personalizing neural inputs improves algorithm performance

[0176] Since power band inputs may be defined over any frequency range, a method of constraining the search space was required. We used an approach that first identified candidate pools of potentially information-rich power bands, then performed an exhaustive search over those candidates to select optimal algorithm inputs. We compared two methods of identifying candidates. The first method utilized “canonical” power bands, representing commonly studied components of cortical and subcortical LFP’s related to movement [Buzsaki and Draguhn, 2004]: theta (4-7Hz), alpha (8-12Hz), Beta (13-30Hz), and Gamma (60-90Hz). The second method identified “personalized” power bands, determined using a signal decomposition approach aimed at capturing the primary neural signals within each hemisphere for any specific participant.

[0177] The personalized method consisted of first transforming the time-domain LFP into a spectral representation using short-time fourier transform (STFT), then performing principal components analysis (PCA) on the z-scored STFT outputs to identify the most strongly coordinated bands of neural activity. A peak-finding algorithm was finally applied to the PCA component weights to find the frequency range best approximating each PC (FIG. 3A). This process was run on an independent, multi-hour dataset. The resulting bands often overlapped canonically described power bands (FIG. 3B). The top two PC’s in all cortical regions reflected personalized versions of beta and gamma, capturing 11 -17% and 7-9% of the overall spectral variance, respectively. The top two PC’s in the STN were found in low-beta and gamma ranges, with gamma (11 -31%) accounting for a largerproportion of the spectral variance than beta (5-7%). The third and fourth PC band edges were more variable across brain areas, typically representing a combined theta-alpha band or low-gamma.

[0178] We performed an exhaustive search over all combinations of power bands from each candidate pool to find optimal combinations of inputs. For each of the two candidate pools, combining all three brain regions, four power bands in each region, and at most four total inputs yields 682 models. These were narrowed down to three candidates per pool through Step 1 ( “Broad search”, FIG. 20) by selecting the models with the highest F1 scores after independently optimizing each model over 20 iterations of Bayesian optimization. The models that advanced from Step 1 were further optimized in Step 2 (“Focused optimization”, FIG. 2C) with an additional 200 iterations, and the single top performing model was then selected for final analysis and adaptive stimulation experiments.

[0179] During optimization, we investigated whether inclusion of cortical recordings and power band personalization added value to the algorithm. We first computed the marginal performance contribution of each individual power band as a measure of feature importance using classifier accuracies from Step 1 . The top power bands were uniformly located in cortical locations, indicating that ECoG signals were the most valuable for decoding movement (FIG. 3C). We then directly compared the two power band identification methods to assess the personalization procedure. Accuracy for the top personalized model was 6% greater than the top canonical model for the left hemisphere, and 3% greater for the right. This effect was consistent across the top 200 models from each pool, displaying significantly greater mean accuracy for the personalized bands (p<0.001 , permutation test; FIG. 3D). Personalized power bands therefore offered significant performance improvements and were selected for subsequent analysis and adaptive therapy.Evaluation of offline movement classification

[0180] We evaluated performance of the final neural classifiers using a held-out dataset to determine whether they would be feasible for guiding adaptive stimulation (FIG. 4A). Prediction accuracy for the left and right hemispheres was 83% and 76%, respectively (binomial test HA: accuracy=0.5, p<0.001 ). Both True Positive Rate (TPR, 88% and 93%) and True Negative Rate (TNR, 81% and 68%) were significantly above chance (binomial test Ho: TPR, TNR=0.5, p<0.001 ; FIG. 4B), yet there remained modest prediction bias towards the moving state (8% and 18%). Subthreshold acceleration of the contralateral arm occasionally occurred during unimanual arm movements and was predicted by the classifier as movement (FIG. 4A, arrow), contributing to this bias. Accelerometry distributions segregated according to predicted movement state (FIG. 4C), illustrating the classifier’s ability todiscern true movements. Finally, the moment-by-moment stimulation amplitudes were simulated (ramp rates:1 .6 mA / sec; state targets: 1 .6-2.2 mA) to predict neurostimulator activity in later adaptive experiments (FIG. 4D). As intended, stimulation was consistently high in each hemisphere during contralateral arm movements (left: 86% of the time, right: 90%), but not when stationary (left: 17%, right: 31 %).Evaluation of movement classification during adaptive stimulation

[0181] We compared two aDBS algorithms with constant DBS for evaluating the therapeutic impact of movement-responsive stimulation (FIG. 5A). The “Movement Responsive” condition consisted of increasing stimulation during predicted movement, while the “Inverted” condition had the inverse association, decreasing stimulation during predicted movement. The Inverted condition, predicted to have the opposite impact of the Movement Responsive condition, acted as a control for the possibility that stimulation intermittency was the driving factor of benefit. Both aDBS conditions varied stimulation between low (1.6 mA) and high (2.2 mA) levels, titrated prior to experimentation as tolerable for the participant. The third “Constant” condition provided conventional therapy at an intermediate level (1 .9 mA) regardless of classifier predictions. Data were collected for each of the three conditions in every session using a randomized and patient-blinded ordering. The same movement tasks that were completed during algorithm training were again performed in each block of the aDBS experiments, except for the hand open-close, which was replaced with nose-tapping to represent ballistic arm movements involving proximal musculature. 12 sessions of data were collected over a 154 day period, with the last day performed 435 days after the original training data were collected.

[0182] The movement classifier continued to perform well during aDBS despite stimulation induced signal artifacts (FIG. 5B). Performance was largely stable for many months at a time; however, changes in neural signal magnitude were observed on two occasions more than a year after training data collection (once in each hemisphere, days 384 and 429). These were addressed through manual adjustment of algorithm thresholds without modifying other parameters, which restored performance to previous levels (FIG. 8). Prediction accuracy for the left and right hemispheres across all testing sessions was 82% and 76%, respectively (binomial test HA: accuracy=0.5, p<0.001 ), and did not differ substantially across stimulation conditions (FIG. 9). The classifiers displayed mild prediction bias. An 8% bias towards predicting movement was observed for the left hemisphere, while a 14% bias towards no movement was observed for the right (FIG. 5C). Overall, intended stimulation contingencies were achieved for each condition (FIG. 5D).Self-perceived therapeutic effects

[0183] We first obtained subjective assessments of therapeutic quality by asking the participant to blindly score each of the blocks on perceived quality of movement (FIG. 6A). One-way ANCOVA with performance order as a covariate revealed a significant effect of condition type (p=3.1 e-4). Pairwise T-tests showed significantly reduced scores for the Inverted condition as compared to both the Movement Responsive (p=5.7e-4) and Constant (p=1 .9e-4) conditions. Adjusting stimulation with respect to movement state therefore influenced subjective therapeutic quality, though primarily driven by reduction in quality reported during the Inverted condition.

[0184] Furthermore, we sought to determine if self-perceived therapeutic quality depended upon algorithm performance or other confounding factors. A positive correlation between self-scores and classification performance is predicted under the foundational hypothesis that increasing stimulation during movement provides therapeutic benefit. We indeed observed a significant correlation between classifier performance and self-scores across sessions (FIG. 6B; Pearson r=0.52, one-sided p=0.043). Additionally, we investigated whether simple increases in average stimulation were driving perceived differences in therapeutic quality, specifically within the two aDBS conditions. Counter to this explanation, a significant negative correlation was observed between mean stimulation amplitude and self-score (FIG. 6C; Pearson r=-0.44, two-sided p=0.032). Collectively, these results show that self-perceived motor improvements resulted from accurately targeting movement states and not indiscriminately increasing stimulation (e.g. from classification biases).Movement vigor and stimulation side-effects

[0185] We next investigated movement speeds, as we predicted that movement-locked increases in stimulation would enable more vigorous action, and stimulation decreases would slow movement. We quantified movement speed as the repetition rate for wrist pronation-supination and nose tapping (FIG. 7E). Two-way ANCOVA’s were performed with performance order as a covariate, revealing significant main effects of condition type (p=1.3e-8), and hand (p=9.2e-3) for wrist rotations, and condition type for nose-tapping (p=1 ,3e-3). A significant interaction between hand and condition type was observed for both wrist rotations (p=1.1e-9) and nose-tapping (p=0.019). Pairwise T-tests showed that, for both movement types, speeds were slower in the Inverted condition compared to other stimulation conditions for the right hand only (p<0.01). This suggests that movement speed was modulated by stimulation condition for the right dominant hand alone. Importantly, classification of the moving state specifically for the left non-dominant hand was lower than any other statepredictions (FIG. 50, Right Hemisphere). This resulted in stimulation being lower on average during left hand movements as compared to the right. However, stimulation was still consistently low during rest for both.

[0186] We further predicted that dyskinesia, which is a common side-effect of dopaminergic medication and DBS [Maciel et al., 2020], would be reduced during rest periods of the Movement Responsive condition when stimulation amplitudes were low. We quantified dyskinesia using low- frequency power (1 -4 Hz) of the watch accelerometry (Rodnguez-Molinero et al., 2019; FIG. 8A). Two-way ANCOVA with performance order as a covariate revealed significant main effects of condition type (p=1.0e-6), and hand (p=0.018), but no significant interaction (p=0.61). Data for the two hands were therefore grouped for pairwise T-tests that showed significantly reduced dyskinesia in the Movement Responsive condition compared to other stimulation conditions (pvs_constant = 8.2e-6, pVsjn erted = 1.1 e-6). We also evaluated tremor, quantified using power in the 4-7 Hz range from the watch accelerometry [Dai et al., 2015], which was previously identified as containing tremor markers for this participant at low stimulation settings (data not shown). We found no evidence that Movement Responsive aDBS worsened tremor, as power in the 4-7 Hz band was also reduced (FIG. 10). In summary, Movement Responsive stimulation provided a bilateral reduction in resting state dyskinesia compared to both other conditions, and a unilateral increase in movement vigor impacting the dominant hand when compared to Inverted stimulation.Naturalistic typing performance

[0187] We finally investigated movement quality in a naturalistic bimanual setting by analyzing typing performance, predicting that movement-locked stimulation increases would enable faster action without compromising dexterity. Three metrics were analyzed (FIG. 7): mean keypress duration (measuring quickness of individual finger movements), overall typing speed, and backspace rate (an approximation of error rate). For keypress duration, one-way ANCOVA with performance order as a covariate revealed a significant effect of condition type (p=4.6e-7). Pairwise T-tests showed significantly shorter keypress durations for the Movement Responsive condition as compared to both of the other conditions (p vs_cconstant— 8.2e-6, Pvs inverted— 1 .1 e-6). For typing speed, a one-way ANCOVA again revealed a significant effect of condition type (p=0.031 ). A significant pairwise difference was observed between the Movement Responsive and Inverted conditions (p=8.5e-3), with Movement Responsive typing being faster by 0.33 keypresses / s on average. No significant differences were observed for backspace rate (ANCOVA, p=0.092). Movement-responsive stimulation thereforeimpacted both quickness of individual finger movements and overall typing speed as predicted without significantly compromising dexterity.DISCUSSION

[0188] The dynamic nature of PD suggests that adaptive stimulation may be a more optimal treatment option than constant stimulation. The varied timescales and underlying sources of these dynamics have given rise to many proposed aDBS paradigms that fit broadly into three categories: algorithms that respond rapidly to specific phasic or transient brain signals to manipulate low-level circuit behavior [Little et al., 2013; Little et al., 2016; Tinkhauser et al., 2017; He et al., 2023; Holt et al., 2016; Grado et al., 2018; Popovych and Tass, 2019], algorithms that use neural biomarkers to track more gradual changes in PD clinical states [Rosa et al., 2017; Arlotti et aL, 2018; Swann et aL, 2018; Gilron et aL, 2021 a; Oehrn et al., 2024], and algorithms that tune stimulation to the current behavioral setting (e.g., movement [Darbin et aL, 2021] and sleep [Gilron et aL, 2021 b; Smyth et aL, 2023]). The present study fits into the third category, attempting to match the prokinetic strength of DBS to moment-by-moment motor needs. However, this feedforward approach can only work if the effects of DBS materialize quickly enough to impact movement. Critically, we have shown that quantifiable alterations in motor performance were discernible on a rapid timescale (FIG. 6), with movement epochs lasting only 20 seconds. This is notable given that STN-DBS effects during simple on-off testing typically operate on the scale of seconds to hours [Hristova et aL, 2000; Koeglsperger et aL, 2019]. These results suggest that movement-responsive stimulation can strengthen therapeutic efficacy while minimizing side-effects through rapid behavior-dependent switching.

[0189] Importantly, effective use of movement-responsive aDBS was found to depend on accuracy of the classifier (FIG. 6B). Two key techniques were used to improve this accuracy. First, we employed a power band personalization method that provided benefit beyond using canonical power bands (FIG. 3D). Second, we included ECoG signals, which were found to be the most valuable for decoding during our feature selection routine (FIG. 3C). This adds to a growing body of support for the inclusion of ECoG sensors in aDBS systems (Gilron et aL, 2021 a; Merk et aL, 2022; Oehrn et aL, 2024). Stability of these signals is also critical to maintain performance. ECoG is considered a stable signal for brain-computer interfaces (BCI) [Degenhart et al., 2016; Volkova et aL, 2019], and has been shown robust for decoding over multiple years [Blakely et aL, 2009; Bleichner et aL, 2019; Pels et aL, 2019]. In the present study, scaling of the ECoG signals was observed on two occasions (once per hemisphere) more than one year after initial model training. This was addressed by manually altering algorithm thresholds without changing other parameters (FIG. 8). While the sourceof signal disturbance is uncertain, stability before and after manual adjustments suggests that typical inflammatory responses were not the cause [Degenhart et al., 2016]. Minor electrode position changes could be the cause, as this study involved a more mobile and unconstrained deployment compared to previous investigations of ECoG-based BCI’s. Nonetheless, the isolated effect of signal scaling ought to be readily accounted for with auto-scaling, and the overall performance of both decoders was encouraging for long-term use. Future work should analyze this stability more thoroughly and explore adaptive learning or re-scaling protocols.

[0190] Several other avenues for continued research and engineering improvements are motivated by the foundational results of this report. We have established that movement-responsive stimulation can have a positive impact on movement (FIGS. 6-7), yet the extent of that clinical impact was here limited by the experimental design requirement to counterbalance the two aDBS conditions (to match average stimulation amplitudes). Any improvement from the primary condition was limited by how much reduction in efficacy could be tolerated in the Inverted control condition while preserving patient comfort. Future studies should create an unconstrained Movement Responsive stimulation contingency, beginning with more extreme stimulation levels. Continuously graded stimulation proportional to the intended movement vigor should also be explored [Meidahl et al., 2017], since movement is not truly binary in most contexts. Additionally, coordinating action of the two stimulators may offer opportunity for improvement, since unilateral DBS is known to impact symptoms bilaterally [Shemisa et al., 2011]. Finally, this approach, validated in Parkinson’s disease, could provide benefit in other conditions that are impacted by reductions in movement. Specifically, this technique could benefit patients with cerebrovascular injuries (stroke), where there is a reduction in motor initiation and vigor. Selectively reducing the basal ganglia motor threshold using movement responsive adaptive DBS during decoded voluntary action attempts, has the potential to augment suboptimal movement acutely, acting as an assistive BCI. Further, this approach could also support the selective facilitation of network learning and plasticity if used as a rehabilitative BCI tool though motor imagery combined with neurofeedback. This could be used to preferentially reinforce high gain movements by directly linking cortical motor initiation signals to basal ganglia stimulation, gating and motor thresholds, and therefore potentially be more effective than current peripheral rehabilitative BCI, dependent on Virtual Reality, robotics or direct muscle stimulation.

[0191] This study introduces a technical framework for aDBS optimization with three key pillars that should inform future aDBS research and technology development. First, to build algorithms that work well at home, data must be collected naturalistically, as neural coding changes across laboratory and naturalistic settings [Wu et al., 2006; Jackson et al., 2007; Wang et al., 2016]. Notably, ourtraining pipeline is compatible with simpler hardware implementations, including only wrist-worn accelerometers, to facilitate naturalistic deployment. Second, to perform data-driven optimization of a device-embedded algorithm, one must be able to simulate its operations. This was achieved using the rcssim software package, a high-fidelity in-silico device simulator developed in support of this study. Third, a tailored ML pipeline is required for optimizing aDBS performance. While the elementary operations of the RC&S device are simple, their interactions become highly complex, making conventional optimization approaches impossible. The future of aDBS therapies will rely upon such frameworks to optimize programming in a tractable way.CONCLUSION

[0192] We demonstrate for the first time the clinical efficacy of a fully embedded movement- responsive aDBS algorithm. The algorithm, which was remotely optimized in the participant’s own home, improved self-reported scores of therapeutic efficacy as well as movement speed, dyskinesia, and typing performance. These findings provide early support for aDBS algorithms that modulate stimulation in accordance with movement state for the treatment of PD and address challenges related to automated parameterization supporting scalable aDBS.METHODSDBS implant and patient information

[0193] This study was reviewed by the University of California, San Francisco Institutional Review Board and registered on clinicaltrials.gov (IDE G180097; NCT03582891 ). One participant with PD was recruited from a parent trial testing slow aDBS and enrolled in the study after providing written informed consent. The participant had been previously implanted with one pair of electrode leads in each hemisphere, consisting of (A) quadripolar depth electrodes (Medtronic model 3389) in the STN, and (B) a quadripolar subdural ECoG array (Medtronic model 0913025) spanning the arm / hand regions of the precentral and postcentral gyri (motor and somatosensory areas). Each pair of electrode leads was connected to an investigational sensing-enabled INS (Summit RC&S, Medtronic) in the ipsilateral subclavicular space. The INS provided therapeutic stimulation to the STN through the depth electrodes, recorded LFP signals from the depth and ECoG electrodes, and recorded accelerometry data through an accelerometer in the INS case. All stimulation changes were self-initiated by the participant, and participant safety was ensured by allowing them to revert to their standard, pre-programmed DBS settings at any time using a handheld patient programmer device.Neural data sensing configuration and signal processing

[0194] All neural recordings were performed during monopolar stimulation. Bipolar recordings in the STN were collected using a “sandwich” configuration (electrodes on either side of the stimulating contact) for common-mode rejection of the stimulation artifact. Two channels of neural data were recorded along the ECoG array, using the first and second electrodes for one channel (postcentral gyrus) and the third and fourth electrodes for the other (precentral gyrus). Neural time-domain data were recorded at a sampling rate of 500Hz. Device-embedded signal processing consisted of short- time Fourier transforms (STFT) with an FFT size of 256 pt and an interval of 50 ms.Behavioral data streaming, signal processing, and synchronization

[0195] Detailed descriptions of the behavioral data collection platform can be found in Strandquist et aL, 2023. Briefly, arm movement data were collected using Apple Watches (Apple, Inc) worn on each wrist and uploaded to a remotely accessible third-party data repository (Rune Labs, Inc.). These data were sampled at 50 Hz, and power was calculated using STFT with an FFT size of 64 pt and interval of 100 ms. Power was calculated in the 0-5 Hz range for movement decoding, 1-4 Hz for quantifying dyskinesia [Rodriguez-Molinero et aL, 2019], and 4-7 Hz for quantifying tremor [Dai et aL, 2015]. Video data for participant observation and kinematic pose estimation were recorded on three cameras arranged at varied angles in the participant’s home office. A custom keylogging application was built to provide statistics regarding typing performance without compromising privacy by removing the letter identity of each keypress. This application was deployed on the participant’s home office computer with patient knowledge and consent. All data was securely transferred to remotely accessible databases over a virtual private network (VPN). To overcome drift and time discrepancies across system clocks and align the data streams, a synchronization event was performed at the start of all recordings. The participant tapped each side of their chest over the implanted INS while standing in view of the cameras. This movement created identifiable spikes in the INS and Apple Watch accelerometry signals and could be discerned on the video, allowing realignment of all data to a common time.Data collection protocols

[0196] Three different datasets were collected remotely for this study. For identifying personalized power band candidates, a single two hour session of free behavior was collected during monopolar stimulation at 1.6 mA. The participant was instructed to go about their regular daily routine whileoccasionally taking breaks to perform hand open-close movements and rest their hands on their lap. Since this dataset was only used to identify the most prominent components of the neural signals through PCA, there were not firm constraints on the participant’s actions during data collection other than ensuring that there were periods of both movement and non-movement. Separate datasets were used for identifying the personalized power bands and evaluating their use in decoding movement to ensure that we avoided any risk of overfitting and represented generalizable performance.

[0197] A second dataset was collected for optimizing the movement classifier and performing feature selection from the identified power band candidates. This dataset included six days of data where the participant performed a prescribed set of motor tasks at three different levels of monopolar stimulation (1.2 mA, 2.4 mA, 2.6 mA). Each day consisted of approximately 25 minutes of total recording time. The participant was given an instruction sheet with the list of motor tasks, which included standardized movement tasks from the Unified Parkinson’s Disease Rating Scale Part III (UPDRS; sections 3.17-3.18 rest tremor, 3.4 finger tapping, 3.5 hand movements - open-close, 3.6 hand movements - pronation-supination) as well as blocks of text for typing that were each approximately 100 words long (FIGS. 2A, 2B). The participant was a proficient typer prior to beginning the study.

[0198] For evaluating the movement responsive aDBS algorithm, 12 days of data were collected while the participant performed a similar set of tasks in three different stimulation conditions (Movement-responsive, Inverted, Constant). Each day again consisted of approximately 25 minutes of total recording time. The open-close hand movements (UPDRS section 3.5) from the initial six day dataset were replaced with finger-to-nose tapping, and no movements of the lower body were performed. All movements were instructed to last 20 seconds, and the rest phase 10 seconds.

[0199] While optimal monopolar stimulation amplitudes were titrated to allow noise-free STN recordings with the sandwich sensing configuration, the participant’s standard therapeutic settings were conventional stimulation (eDBS) in a bipolar arrangement. In the monopolar setting, stimulation amplitudes were titrated separately for each of the three collected datasets before the first session. The high (2.2 mA) and low (1.6 mA) stimulation amplitudes for adaptive recordings were selected to be balanced around the constant stimulation amplitude (1.9 mA). These targets were also conservatively defined to ensure that the Inverted condition remained tolerable for the participant, since the inverted stimulation contingency was perceived as worse therapeutically. The main Movement Responsive test condition was therefore theoretically limited in how positive of an effect it could have, since its reciprocal needed to remain both counterbalanced and tolerable.Personalized power band identification and assessment

[0200] Personalized power bands were identified by performing Principal Component Analysis (PCA) on the power spectral data. A custom peak-finding algorithm was then used on each PC weight vector to approximate it with a single contiguous frequency range that was compatible with RC&S signal processing capabilities (FIG. 3A). The resulting power bands therefore (A) aligned with the dimensions of the neural data that captured the most variance, and (B) produced features that were approximately linearly uncorrelated to facilitate downstream linear combination for classification models. The steps are briefly outlined below:

[0201] Compute a [n samples x L / 2] spectrogram of the time-domain neural data using the rcssim signal processing with the following parameters:Fs_td = 500 Hz (time-domain sampling rate)L = 256 (FFT size) interval = 200 ms

[0202] Perform PCA on the z-scored spectrogram data and retain the top 4 PCs. Take the absolute values of each PC weight vector, which will be size [L / 2 x 1] and represent the weight that a given PC places on each frequency bin in the previously performed FFT.

[0203] Run the peak-finding algorithm on the PC weight vectors to identify a contiguous range of frequency bins that best approximates each PC (restrict searchable range to 4-100Hz to avoid known noise issues):

[0204] Excluding frequency bins that have been included in power bands that were identified prior, find the frequency bin with the largest weight for the current PC (i.e., the peak of the weight vector)

[0205] Extend the current power band away from the peak in both directions until one of the following conditions are met:The frequency bin has been included in a previously identified power band (prevents power band overlap).The weight on the frequency bin drops below 50% of the peak weight (excludes frequencies that are not well-represented in the PC).A trough has been encountered (prevents multi-peak power bands).The frequency bin is more than 25Hz from the peak (limits power bands to 50Hz full range). Repeat for a total of 4 PC’s, independently for each anatomical recording area and brain hemisphere.

[0206] Feature importance was assessed for each power band using its marginal performance contribution, which was computed as the performance increase (or decrease, as a result of overfitting) when a given power band was included as an additional feature in a model. Since models using every possible combination of four or less power bands were independently optimized in Step 1 of the feature selection process, the marginal performance contribution for a given power band was represented as a distribution of 196 values, one for each combination that included that power band.Movement classifier and optimization pipeline

[0207] The Summit RC&S embedded aDBS has four primary components to its onboard operations. (1 ) Neural signals are first processed into power band outputs given a set of sensing parameters that include the frequency edges dictating each band. (2) Up to four power bands are streamed, averaged across a number of subsequent samples dictated by the update rate, and linearly combined via feature weights into a continuous valued linear discriminant output (LD). (3) LD outputs are then passed through a series of logic operations, which include comparison to a threshold value and requisite hold times above (onset duration) or below (termination duration) that value to trigger a state change. (4) Upon state change, stimulation is ramped towards the target amplitude associated with the new state. Note that this is a non-exhaustive description of the device capabilities that is focused on the most important features and parameters for the present study.

[0208] Because of the unique operations of the RC&S, a custom optimization pipeline was required for training a movement classifier that leverages its full functionality. In brief, each loop of GPBO simulated the embedded algorithm for a given set of parameters and neural inputs, evaluated the algorithm performance against labeled movement data, and modeled the relationship between parameters and performance to support an intelligent and efficient search-based optimization. LD weights were the only parameters not modeled and optimized through the GPBO search method, but instead were fit by linearly regressing the continuous-valued wearable accelerometry on the neural signals within each GPBO loop. By fitting these weights in closed form rather than as additional parameters in the GPBO gaussian process model, the GPBO routine was greatly simplified. Notably, the thresholds were not fit in closed form (e.g., through linear discriminant analysis), since it was assumed that interactions between thresholds and parameters such as onset duration would require that they be jointly optimized. Using the regression technique, LD outputs therefore represented an estimate of the continuous-valued accelerometry, which was then processed through the remaining algorithmic steps to produce binary state outputs given the set ofparameters defined in the current GPBO loop. These state outputs were then compared against the true movement state determined from wearable accelerometry to calculate an F1 -score. The F1- score was used as the objective function of the GPBO search, which was selected as the performance metric to promote unbiased and accurate predictions. Prediction bias reported elsewhere in the results was calculated as the difference between the frequency of labels predicted by the classifier and the frequency of labels in the actual data. Optimization was performed using five-fold cross-validation, using a separate day of data for each fold. All optimization code was written in Python, using the rcssim package for simulating RC&S operations and sklearn and skoptfor model fitting and optimization.

[0209] As a minor stage in this process, a state change blanking variable dictating a refractory period to allow stabilization of artifacts associated with changes in stimulation was also optimized. This simple step consisted of modeling the artifact, which was different at cortical and subthalamic recording locations, and automatically assigning the appropriate duration of blanking given the neural inputs present in the model (i.e., whether inputs were from the cortex or subthalamic nucleus). In summary, the following sensing parameters were optimized for each hemisphere:

[0210] Four power bands (though only using three was found to be optimal in the right hemisphere), each consisting of:Frequency lower limitFrequency upper limit

[0211] And the following algorithmic parameters were optimized for each hemisphere:LD weights (four total, one for each power band) Update rate Threshold Onset duration Termination durationLag (not a device-programmable parameter, but allowed for time delay between neural signals and movement) State change blanking

[0212] In total, 16 sensing parameters and 20 algorithmic parameters were therefore optimized.Statistical testing

[0213] All statistical testing and data visualization was performed using the Python programming language and common statistical packages. Basic statistics were performed using scipy, and ANCOVA models were performed using statsmodels. Following ANCOVA, T-tests were performed as planned contrasts only in the event of statistically significant ANCOVA effects, thus no adjustments were made for multiple comparisons. The T-tests were covariate-adjusted for all statistically significant covariates identified through ANCOVA.REFERENCES:

[0214] 1. Kuhn, Andrea A., et al. "Reduction in subthalamic 8-35 Hz oscillatory activity correlates with clinical improvement in Parkinson's disease." European Journal of Neuroscience 23.7 (2006): 1956-1960.

[0215] 2. Herron, Jeffrey A., et al. "Chronic electrocorticography for sensing movement intention and closed-loop deep brain stimulation with wearable sensors in an essential tremor patient." Journal of neurosurgery 27.3 (2016): 580-587.

[0216] 3. Ferleger, B. I., et al. "Fully implanted adaptive deep brain stimulation in freely moving essential tremor patients." Journal of Neural Engineering 17.5 (2020): 056026.

[0217] 4. Opri, Enrico, et al. "Chronic embedded cortico-thalamic closed-loop deep brain stimulation for the treatment of essential tremor." Science translational medicine 12.572 (2020): eaay7680.

[0218] 5. Popovych, Oleksandr V., and Peter A. Tass. "Adaptive delivery of continuous and delayed feedback deep brain stimulation-a computational study." Scientific reports 9.1 (2019): 10585.

[0219] 6. Holt, Abbey B., et al. "Phasic burst stimulation: a closed-loop approach to tuning deep brain stimulation parameters for Parkinson’s disease." PLoS computational biology 12.7 (2016): e1005011.

[0220] 7. Little, Simon, et al. "Adaptive deep brain stimulation in advanced Parkinson disease."Annals of neurology 74.3 (2013): 449-457.

[0221] 8. Little, Simon, et al. "Bilateral adaptive deep brain stimulation is effective in Parkinson's disease." Journal of Neurology, Neurosurgery & Psychiatry 87.7 (2016): 717-721 .

[0222] 9. Swann, Nicole C., et al. "Adaptive deep brain stimulation for Parkinson’s disease using motor cortex sensing." Journal of neural engineering 15.4 (2018): 046006.

[0223] 10. Gilron, Ro’ee, et al. [a] "Long-term wireless streaming of neural recordings for circuit discovery and adaptive stimulation in individuals with Parkinson’s disease." Nature biotechnology 39.9 (2021 ): 1078-1085.

[0224] 11 . Gilron, Ro’ee, et al. [b] "Sleep-aware adaptive deep brain stimulation control: chronic use at home with dual independent linear discriminate detectors." Frontiers in Neuroscience 15 (2021 ): 732499.

[0225] 12. Arlotti, Mattia, et al. "Eight-hours adaptive deep brain stimulation in patients withParkinson disease." Neurology 90.11 (2018): e971-e976.

[0226] 13. Rosa, Manuela, et al. "Adaptive deep brain stimulation controls levodopa-induced side effects in Parkinsonian patients." Movement Disorders 32.4 (2017): 628.

[0227] 14. Jackson, Andrew, Jaideep Mavoori, and Eberhard E. Fetz. "Correlations between the same motor cortex cells and arm muscles during a trained task, free behavior, and natural sleep in the macaque monkey." Journal of neurophysiology 97.1 (2007): 360-374.

[0228] 15. Wang, Nancy XR, et al. "Unsupervised decoding of long-term, naturalistic human neural recordings with automated video and audio annotations." Frontiers in human neuroscience 10 (2016): 165.

[0229] 16. Merk, Timon, et al. "Electrocorticography is superior to subthalamic local field potentials for movement decoding in Parkinson’s disease." Elife 1 1 (2022): e75126.

[0230] 17. Ivry, Richard B., and Rebecca MC Spencer. "The neural representation of time." Current opinion in neurobiology 14.2 (2004): 225-232.

[0231] 18. Tinkhauser, Gerd, et al. "The modulatory effect of adaptive deep brain stimulation on beta bursts in Parkinson’s disease." Brain 140.4 (2017): 1053-1067.

[0232] 19. Strandquist, Gabrielle, et al. "Bringing the Clinic Home: An At-Home Multi-Modal DataCollection Ecosystem to Support Adaptive Deep Brain Stimulation." JoVE (Journal of Visualized Experiments) 197 (2023): e65305.

[0233] 20. Shemisa, Kamal, et al. "Unilateral deep brain stimulation surgery in Parkinson’s disease improves ipsilateral symptoms regardless of laterality." Parkinsonism & related disorders 17.10 (201 1): 745-748.

[0234] 21 . Buzsaki, Gyorgy, and Andreas Draguhn. "Neuronal oscillations in cortical networks." science 304.5679 (2004): 1926-1929.

[0235] 22. Maciel, Ricardo, et al. "Functional Dyskinesias following Subthalamic Nucleus Deep BrainStimulation in Parkinson's Disease: A Report of Three Cases." Movement Disorders Clinical Practice 8.1 (2021 ): 114-1 17.

[0236] 23. Smyth, Clay, et al. "Adaptive Deep Brain Stimulation for sleep stage targeting inParkinson’s disease." Brain Stimulation 16.5 (2023): 1292-1296.

[0237] 24. He, Shenghong, et al. “Beta-triggered adaptive deep brain stimulation during reaching movement in parkinson's disease.” Brain, 2023, doi.org / 10.1093 / brain / awad233.

[0238] 25. Darbin, Olivier, et al. "Subthalamic nucleus deep brain stimulation driven by primary motor cortex y2 activity in parkinsonian monkeys." Scientific Reports 12.1 (2022): 6493.

[0239] 26. Klaus, Andreas, Joaquim Alves da Silva, and Rui M. Costa. "What, if, and when to move: basal ganglia circuits and self-paced action initiation." Annual review of neuroscience 42 (2019): 459-483.

[0240] 27. Rodriguez-Molinero, Alejandro, et al. "Estimating dyskinesia severity in Parkinson’s disease by using a waist-worn sensor: Concurrent validity study." Scientific reports 9.1 (2019): 13434.

[0241] 28. Dai, Houde, Pengyue Zhang, and Tim C. Lueth. "Quantitative assessment of parkinsonian tremor based on an inertial measurement unit." Sensors 15.10 (2015): 25055-25071.

[0242] 29. Lofredi, Roxanne, et al. "Subthalamic beta bursts correlate with dopamine-dependent motor symptoms in 106 Parkinson’s patients." npj Parkinson's Disease 9.1 (2023): 2.

[0243] 30. Neumann, Wolf-Julian, et al. "Subthalamic synchronized oscillatory activity correlates with motor impairment in patients with Parkinson's disease." Movement Disorders 31.11 (2016): 1748-1751.

[0244] 31 . Grado, Logan L., Matthew D. Johnson, and Theoden I. Netoff. "Bayesian adaptive dual control of deep brain stimulation in a computational model of Parkinson’s disease." PLoS computational biology 14.12 (2018) : e1006606.

[0245] 32. Adaptive DBS Algorithm for Personalized Therapy in Parkinson’s Disease (ADAPT-PD),NCT04547712. U.S. National Library of Medicine, 2023. clinicaltrials.gov / study / NCT04547712

[0246] 33. Koeglsperger, Thomas, et al. "Deep brain stimulation programming for movement disorders: current concepts and evidence-based strategies." Frontiers in neurology 10 (2019): 410.

[0247] 34. Hristova, Anna, et al. "Effect and time course of deep brain stimulation of the globus pallidus and subthalamus on motor features of Parkinson's disease." Clinical neuropharmacology 23.4 (2000): 208-211.

[0248] 35. Meidahl, Anders Christian, et al. "Adaptive deep brain stimulation for movement disorders: the long road to clinical therapy." Movement disorders 32.6 (2017): 810-819.

[0249] 36. Degenhart, Alan D., et al. "Histological evaluation of a chronically-implanted electrocorticographic electrode grid in a non-human primate." Journal of neural engineering 13.4 (2016): 046019.

[0250] 37. Volkova, Ksenia, et al. "Decoding movement from electrocorticographic activity: a review." Frontiers in neuroinformatics 13 (2019): 74.

[0251] 38. Blakely, Tim, et al. "Robust, long-term control of an electrocorticographic brain-computer interface with fixed parameters." Neurosurgical focus 27.1 (2009): E13.

[0252] 39. Pels, Elmar GM, et al. "Stability of a chronic implanted brain-computer interface in latestage amyotrophic lateral sclerosis." Clinical Neurophysiology 130.10 (2019): 1798-1803.

[0253] 40. Sellers, Kristin K., et al. "Analysis-rcs-data: open-source toolbox for the ingestion, timealignment, and visualization of sense and stimulation data from the medtronic summit RC+ S system." Frontiers in Human Neuroscience 15 (2021): 714256.

[0254] 41. Wu, Michael C-K., Stephen V. David, and Jack L. Gallant. "Complete functional characterization of sensory neurons by system identification." Annu. Rev. Neurosci. 29 (2006): 477- 505.

[0255] 42. Odekerken, Vincent J J, et al. "Subthalamic nucleus versus globus pallidus bilateral deep brain stimulation for advanced Parkinson's disease (NSTAPS study): a randomised controlled trial." The Lancet Neurology 12.1 (2013): 37-44.

[0256] 43. Follett, Kenneth A., et al. "Pallidal versus subthalamic deep-brain stimulation forParkinson's disease." New England Journal of Medicine 362.22 (2010): 2077-2091 .

[0257] 44. Deuschi, Gunther, et al. "A randomized trial of deep-brain stimulation for Parkinson's disease." New England Journal of Medicine 355.9 (2006): 896-908.

[0258] 45. McGregor, Matthew M., and Alexandra B. Nelson. "Circuit mechanisms of Parkinson’s disease." Neuron 101 .6 (2019): 1042-1056.

[0259] 46. Ashkan, Keyoumars, et al. "Insights into the mechanisms of deep brain stimulation."Nature Reviews Neurology 13.9 (2017): 548-554.

[0260] 47. Johnson, Luke A., et al. "Closed-loop deep brain stimulation effects on parkinsonian motor symptoms in a non-human primate-is beta enough?." Brain stimulation 9.6 (2016): 892-896.

[0261] 48. Kuhn, Andrea A., et al. "Event-related beta desynchronization in human subthalamic nucleus correlates with motor performance." Brain 127.4 (2004): 735-746.

[0262] 49. Eisinger, Robert S., et al. "Parkinsonian beta dynamics during rest and movement in the dorsal pallidum and subthalamic nucleus." Journal of Neuroscience 40.14 (2020): 2859-2867.

[0263] 50. Velisar, A., et al. "Dual threshold neural closed loop deep brain stimulation in Parkinson disease patients." Brain stimulation 12.4 (2019): 868-876.

[0264] 51. Bologna, Matteo, et al. "Evolving concepts on bradykinesia." Brain 143.3 (2020): 727-

[0265] 52. Pozzi, Nicolo G., and loannis U. Isaias. "Adaptive deep brain stimulation: RetuningParkinson's disease." Handbook of Clinical Neurology 184 (2022): 273-284.

[0266] 53. Chevalier, Gilles, and Jean Michel Deniau. "Disinhibition as a basic process in the expression of striatal functions." Trends in neurosciences 13.7 (1990): 277-280.

[0267] 54. Little, Simon, and Peter Brown. "Debugging adaptive deep brain stimulation forParkinson's disease." Movement Disorders 35.4 (2020): 555-561 .

[0268] 55. He, Shenghong, et al. "Closed-Loop Deep Brain Stimulation for Essential Tremor Based on Thalamic Local Field Potentials." Movement Disorders 36.4 (2021): 863-873.

[0269] 56. Oehrn, C. R., Cernera, S., Hammer, L. H., Shcherbakova, M., Yao, J., Hahn, A., Starr, P.A. (2023). Personalized chronic adaptive deep brain stimulation outperforms conventional stimulation in Parkinson’s disease. MedRxiv: The Preprint Server for Health Sciences. doi.org / 10.1101 / 2023.08.03.23293450Example 2Recording Cortical and Subcortical Signals in Patients with PD during Unsupervised Daily Activities within Naturalistic Settings during DBS to Identify Patient-Specific Biomarkers of MovementIntroduction

[0270] Deep brain stimulation (DBS) is an effective treatment for Parkinson’s Disease (PD), and previous studies have explored its effects on cortical and subcortical biomarkers of symptom severity1-4. For example, open-loop DBS has been shown to reduce the excessive subthalamic beta synchronization associated with bradykinesia5-9. However, its impact on the neural correlates of movement remains to be determined. While many previous studies utilized intraoperative recordings with externalized leads, our study involved recording cortical and subcortical signals in patients with PD as they engaged in unsupervised daily activities within naturalistic settings during DBS10-12. Concurrently, we measured forearm speeds using wrist-worn accelerometers, facilitating the examination of group- and patient-level biomarkers of movement.

[0271] We also evaluated movement-predictive machine learning (ML) models for future use in closed-loop DBS. Based on the gating theory of basal ganglia function, stimulating the subthalamic nucleus (STN) or globus pallidus internus (GPi) is hypothesized to lower the barrier to movement execution13. Thus, by augmenting stimulation during movement, we can potentially target hypokinetic symptoms at their peak severity. In this study, we developed classifiers to differentiate betweenmobile and stationary states and regressors to forecast absolute forearm speed during naturalistic motion. We then analyzed the feature importance of each spectral biomarker and explored the impact of varying DBS current amplitudes on these biomarkers and model performance.Methods

[0272] We collected 610 hours of neural data from 15 patients with idiopathic Parkinson’s disease (PD) while they performed activities of daily living in naturalistic settings (FIG. 11 A). Depth leads were surgically implanted into either the subthalamic nucleus (STN) or the globus pallidus internus (GPi), and electrocorticography (ECoG) strips were positioned along the pre- and post-central gyri. These electrodes were connected to a Summit RC+S implantable pulse generator for stimulation, as well as the collection of neural recordings1’2 14. Participants also wore an Apple Watch on the wrist contralateral to the hemisphere where the RC+S device was implanted. Accelerometry signals in the x, y, and z directions were streamed from the watches to nearby iPhones and integrated to derive forearm speed measurements.

[0273] Cortical and subcortical power spectra were computed via Fast Fourier Transformation (FFT) from the recorded local field potentials (LFPs). The data from the wearable accelerometers were utilized to differentiate between movement states and calculate Cohen’s d effect sizes for the canonical alpha (8 to 12 Hz), low beta (12 to 20 Hz), high beta (20 to 30 Hz), low gamma (30 to 60 Hz), and high gamma (60 to 100 Hz) power bands (PBs). We also developed classifiers and regressors that utilized power spectral features to distinguish between mobile and stationary states or predict absolute forearm speed. Permutation feature importance was also applied, and principal component analysis (PCA) was conducted to explore the utility of models utilizing personalized features.ResultsSpectral correlates of naturalistic movement

[0274] We used the combined forearm speed measurements of all patients to delineate the stationary (S) and mobile (M) regimes of naturalistic movement. We then computed the Cohen’s d effect sizes using the power spectral values from each movement state. To investigate variations in effect sizes, we shifted the speed threshold between the 5thand 951hpercentiles, producing Cohen’s “d-grams” for each hemisphere. We averaged the Cohen’s d-grams from both hemispheres for patients with bilateral recordings. The resulting group-level, cluster-based permutation test-correctedCohen’s d-gram (FIG. 11 B) indicated a decrease in cortical and subcortical beta and an increase in broadband gamma for mobile vs stationary states.Effects of stimulation current amplitude on movement biomarkers

[0275] We investigated the effects of stimulation amplitude on movement-related alpha / beta desynchronization (MRD) and gamma synchronization (MRS) at the four intracranial sites. In the STN / GPi, the magnitude of low beta (LMM; p = -0.080; p < 0.0001 ) and high beta (LMM; p = -0.108; p < 0.0001 ) MRD decreased with stimulation. Similarly, the magnitudes of low gamma (LMM; p = - 0.040; p < 0.001) and high gamma (LMM; p = -0.040; p < 0.0001) MRS were negatively correlated with stimulation (FIG. 11 D). An example of this for a single hemisphere is shown in FIG. 11 C. In the cortical regions, stimulation amplitude had no significant effect on high beta MRD or gamma MRS. There was, however, a substantial decrease in the magnitude of alpha (S1 : LMM; p = -0.041 ; p < 0.01 ) (M1 : LMM; p = -0.032; p < 0.05) and low beta (Site: S1 ; LMM; p = -0.061 ; p < 0.0001 ) (Site: M1 ; LMM; p = -0.053; p < 0.05) MRD.Evaluation of machine learning models for the prediction of movement speed

[0276] We developed linear models from one or all intracranial regions using five canonical PBs (alpha, low beta, high beta, low gamma, and high gamma). We then optimized the FFT window via five-fold cross-validation and investigated using up to 15 principal components (PCs) as features. The models trained on data from all sites had the highest performance (p < 0.05), followed by those trained only on signals from the S1 (p < 0.01 ) (FIG. 11 E). Permutation feature importance revealed differences in predictive power between the four sites (One-way ANOVA; F= 5.1 , p = 2.7e-8) (FIG. 11 F). S1 (Mean change in r statistic ± SEM = -0.08 ± 0.02) and M 1 (Mean change in r statistic ± SEM = -0.07 ± 0.02) high beta attained the greatest mean feature importance. We also investigated the use of non-linear models, but they did not produce significant differences in performance compared to the linear models (One-sided Wilcoxon signed rank test; p = 0.0009).Effects of stimulation current amplitude on ML model performance

[0277] We found our single-site and combined models stable over time, with no significant effect of duration between sessions on model performance. We also evaluated differences in model performance at different stimulation levels. We observed that the STN / GPi classifiers (LMM; p = - 0.047; p < 0.001 ) and regressors (LMM; p = -0.054; p < 0.001 ) were negatively correlated withstimulation amplitude FIG. 11G). However, the models trained on S1 and M1 data were not significantly affected by variations in the DBS amplitude.Discussion

[0278] We showed that machine learning models with high performance could be developed to predict naturalistic movement during DBS. Models trained on canonical PB features from all three sites outperformed single-site models12 15. Also, cortical high beta emerged as the most influential feature of our models, and S1 gamma was a superior predictor of movement compared to M1 gamma3 16. These findings revealed that different sites along the cortico-basal ganglia motor network may generate signals with complementary information that can be leveraged to enhance movement prediction.

[0279] We also observed that patients could achieve the same movement speed at higher stimulation levels with reduced subcortical beta MRD and gamma MRS. Based on the gating theory of basal ganglia function, stimulating the STN or GPi is hypothesized to lower the barrier to movement execution. Our work suggests this might be associated with decreased subcortical beta MRD and gamma MRS13’17’18. We also found that models trained on subcortical signals showed reduced performance with increased stimulation. In contrast, changes in DBS amplitudes did not significantly affect the main cortical biomarkers of movement - high beta MRD and gamma MRS - and their respective models. These findings provide insights into the potential mechanism by which DBS ameliorates hypokinetic symptoms in PD and highlight the utility of machine learning models for predicting naturalistic movement during DBS, towards development of algorithms that specifically boost or reinforce, through plastic changes, specific types of movement in PD and other disorders affected by abnormal movement.References

[0280] 1. Gilron, R. eta / . Long-term wireless streaming of neural recordings for circuit discovery and adaptive stimulation in individuals with Parkinson’s disease. Nat Biotechnol 39, 1078-1085 (2021).

[0281] 2. Swann, N. C. et al. Adaptive deep brain stimulation for Parkinson’s disease using motor cortex sensing. J Neural Eng 15, (2018).

[0282] 3. Tinkhauser, G. etal. The modulatory effect of adaptive deep brain stimulation on beta bursts in Parkinson’s disease. Brain 140, 1053-1067 (2017).

[0283] 4. Little, S. et al. Adaptive deep brain stimulation in advanced Parkinson disease. AnnNeurol 74, 449-457 (2013).

[0284] 5. Little, S., Pogosyan, A., Kuhn, A. A. & Brown, P. Beta band stability over time correlates with Parkinsonian rigidity and bradykinesia. Exp Neurol 236, 383-388 (2012).

[0285] 6. Kuhn, A. A., Kupsch, A., Schneider, G. H. & Brown, P. Reduction in subthalamic 8-35 Hz oscillatory activity correlates with clinical improvement in Parkinson’s disease. European Journal of Neuroscience 23, 1956-1960 (2006).

[0286] 7. Neumann, W. J. et al. Subthalamic synchronized oscillatory activity correlates with motor impairment in patients with Parkinson’s disease. Movement Disorders 31 , 1748-1751 (2016).

[0287] 8. Lofredi, R. et al. Beta bursts during continuous movements accompany the velocity decrement in Parkinson’s disease patients. Neurobiol Dis 127, 462-471 (2019).

[0288] 9. Kuhn, A. A. et al. Pathological synchronisation in the subthalamic nucleus of patients with Parkinson’s disease relates to both bradykinesia and rigidity. Exp Neurol 215, 380-387 (2009).

[0289] 10. Herz, D. M. et al. Mechanisms Underlying Decision-Making as Revealed by Deep-Brain Stimulation in Patients with Parkinson’s Disease. Current Biology 28, 1 169-1 178.e6 (2018).

[0290] 11 . Herz, D. M. et al. Dynamic control of decision and movement speed in the human basal ganglia. Nat Commun 13, (2022).

[0291] 12. Merk, T. et al. Electrocorticography is superior to subthalamic local field potentials for movement decoding in Parkinson’s disease. Elife 11 , (2022).

[0292] 13. Darbin, O. et al. Subthalamic nucleus deep brain stimulation driven by primary motor cortex y2 activity in parkinsonian monkeys. Sci Rep 12, (2022).

[0293] 14. Olaru, M. etal. Motor network gamma oscillations in chronic home recordings predict dyskinesia in Parkinson’s disease. Brain (2024) doi:10.1093 / brain / awae004.

[0294] 15. Merk, T. et al. Machine learning based brain signal decoding for intelligent adaptive deep brain stimulation. Exp Neurol 351 , (2022).

[0295] 16. Torrecillos, F., He, S., Kuhn, A. A. & Tan, H. Average power and burst analysis revealed complementary information on drug-related changes of motor performance in Parkinson’s disease. NPJ Parkinsons Dis 9, (2023).

[0296] 17. He, S. et al. Beta-triggered adaptive deep brain stimulation during reaching movement in Parkinson’s disease. Brain 146, 5015-5030 (2023).

[0297] 18. Kondylis, E. D. etal. Movement-related dynamics of cortical oscillations in Parkinson’s disease and essential tremor. Brain 139, 2211-2223 (2016).

Claims

What is claimed is:

1. A method for treating bradykinesia in a subject using deep brain stimulation, the method comprising: positioning a stimulation electrode at a first location in a subthalamic nucleus region or a second location in a globus pallidus internus region of the brain of the subject to deliver electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region; positioning a first neural recording electrode at a third location in the subthalamic nucleus region of the brain of the subject and a second neural recording electrode at a fourth location in a contralateral cortex region of the brain of the subject to record brain electrical signal data associated with an intended movement of the subject; detecting a brain electrical signal associated with the intended movement using the first neural recording electrode or the second neural recording electrode; and applying electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the stimulation electrode in a manner effective to provide a prokinetic effect on the intended movement of the subject when the brain electrical signal associated with the intended movement is detected using the first neural recording electrode or the second neural recording electrode.

2. The method of claim 1 , wherein the contralateral cortex region comprises a precentral gyrus region, a postcentral gyrus region, or both the precentral gyrus region and the postcentral gyrus region.

3. The method of claim 1 , wherein the contralateral cortex region comprises the central sulcus.

4. The method of any one of claims 1 -3, wherein the electrical stimulation reduces the bradykinesia and increases vigor of the intended movement compared to in absence of the electrical stimulation.

5. The method of any one of claims 1 -4, wherein the electrical stimulation is applied unilaterally or bilaterally.

6. The method of any one of claims 1-5, wherein the brain electrical signal comprises alpha frequency, beta frequency, gamma frequency, or theta frequency neural oscillations.

7. The method of claim 6, wherein the theta frequency neural oscillations are in a range from 4 Hz to 7 Hz, the alpha frequency neural oscillations are in a range from 8 Hz to 12 Hz, the beta frequency neural oscillations are in a range from 13 Hz to 30 Hz, and the gamma frequency neural oscillations are in a range from 60 Hz to 90 Hz.

8. The method of any one of claims 1 -7, wherein the brain electrical signal data comprises field potential data.

9. The method of any one of claims 1 -8, further comprising using a control algorithm to automate said applying electrical stimulation when the brain electrical signal associated with the intended movement is detected.

10. The method of claim 9, wherein the control algorithm uses a machine learning algorithm for movement classification.

11. The method of claim 10, wherein the machine learning algorithm is a supervised machine learning algorithm.

12. The method of claim 10 or 11 , wherein the movement classification distinguishes between when the subject is in an intended movement state or a stationary state.

13. The method of claim 12, further comprising training a machine learning model to predict whether the subject is in an intended movement state using accelerometry data for detected movement of the subject.

14. The method of claim 13, wherein the accelerometry data is from a wearable accelerometer worn by the subject that detects natural undirected movement of the subject.

15. The method of claim 13, wherein the accelerometry data is from a wearable accelerometer worn by the subject that detects movement during directed movement tasks selectedfrom finger tapping, opening or closing of a hand, wrist pronation or supination, walking, and keyboard typing.

16. The method of any one of claims 12-15, further comprising performing principal components (PC) decomposition on power spectral data to identify personalized power band features for distinguishing the intended movement state from the stationary state.

17. The method of claim 16, further comprising: transforming a time-domain of a local field potential (LFP) into a spectrogram using a short- time fourier transform (STFT); performing principal components analysis (PCA) on z-scored STFT outputs to identify coordinated bands of neural activity; and applying a peak-finding algorithm to PCA component weights to find a frequency range best approximating each principal component.

18. The method of claim 17, wherein top principal components for the contralateral cortex region or the subthalamic nucleus region correspond to personalized frequency ranges for beta frequency neural oscillations and gamma frequency neural oscillations for the brain electrical signal associated with the intended movement of the subject.

19. The method of any one of claims 12-18, wherein the electrical stimulation is selectively increased during the intended movement state and not during the stationary state.

20. The method of any one of claims 12-19, further comprising using accelerometry data, gyroscope data, multi-view video recordings of the subject, keylogging data from a computer used by the subject, or any combination thereof in combination with the recorded brain electrical signal data to distinguish between the intended movement state and the stationary state.21 . The method of claim 20, wherein a wearable device worn by the subject is used to collect the accelerometry data, the gyroscope data, or a combination thereof.

22. The method of claim 21 , wherein the wearable device is a smartwatch.

23. The method of any one of claims 12-22, wherein the control algorithm further uses linear discriminant analysis (LDA) to adjust stimulation amplitude or frequency of the electrical stimulation.

24. The method of any one of claims 1-23, wherein the intended movement is a finger tap, opening or closing of a hand, wrist pronation or supination, walking, or keyboard typing.

25. The method of any one of claims 1-24, wherein the electrical stimulation is optimized for each hand of the subject independently.

26. The method of any one of claims 1 -25, wherein the stimulation electrode is placed on a surface of the subthalamic nucleus region or the globus pallidus internus region.

27. The method of any one of claims 1 -26, wherein the first neural recording electrode is placed on a surface of the subthalamic nucleus region.

28. The method of any one of claims 1 -27, wherein the second neural recording electrode is placed on a surface of the contralateral cortex region.

29. The method of any one of claims 1 -28, wherein the stimulation electrode is a nonbrain penetrating surface electrode array or a brain-penetrating electrode array.

30. The method of any one of claims 1-29, wherein the first neural recording electrode and / or the second neural electrode is a non-brain penetrating surface electrode array or a brainpenetrating electrode array.

31. The method of any one of claims 1-30, wherein the first neural recording electrode and / or the second neural electrode is an electroencephalogram (EEG) electrode array, a subgaleal or burrhole mounted or cranially mounted neurostimulator electrode, or an electrocorticogram (ECoG) electrode array.

32. The method of claim 31 , wherein the ECoG electrode array spans regions of the precentral gyrus and the postcentral gyrus region that control hand and arm movement.

33. The method of claim 31 , wherein the ECoG electrode array spans the central sulcus.

34. The method of any one of claims 1-33, wherein the subject is further administered dopaminergic medication.

35. The method of any one of claims 1 -34, further comprising assessing effectiveness of the treatment in the subject.

36. The method of claim 35, wherein said assessing comprises using behavioral data obtained of the subject.

37. The method of claim 36, wherein the behavior data is accelerometry data, gyroscope data, video-based pose kinematic data for the subject, or keylogging data from a computer used by the subject, or a combination thereof.

38. The method of any one of claims 35-37, wherein said assessing comprises using a Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) or a Hoehn and Yahr (HnY) scale, a Parkinson’s Disease Composite Scale (PDCS), or a Schwab and England Activities of Daily Living (ADL) Scale.

39. The method of any one of claims 1 -38, wherein the method is performed at a site remote from a hospital.

40. The method of claim 39, wherein the method is performed at the subject’s home.41 . The method of any one of claims 1 -40, wherein the subject has Parkinson’s disease, Parkinsonism, Lewy body dementia, progressive supranuclear palsy, corticobasal degeneration, Huntington's disease, dystonia, stroke, or multiple system atrophy.

42. A computer implemented method for programming a deep brain stimulator to treat bradykinesia in a subject, the computer performing steps comprising:receiving recorded brain electrical signal data from a contralateral cortex region and a subthalamic nucleus region of the brain of the subject; analyzing the recorded brain electrical signal data using a movement classification model that identifies patterns of electrical signals in the recorded brain electrical signal data associated with intended movement of the subject; adjusting one or more programmed stimulation parameters based on the recorded brain electrical signal data according to an algorithm control law; and instructing the deep brain stimulator to apply an electrical stimulation to the subthalamic nucleus region or a globus pallidus internus region of the brain to provide a prokinetic effect on the intended movement of the subject.

43. The computer implemented method of claim 42, wherein the contralateral cortex region comprises a precentral gyrus region, a postcentral gyrus region, or both the precentral gyrus region and the postcentral gyrus region.

44. The computer implemented method of claim 42, wherein the contralateral cortex region comprises the central sulcus.

45. The computer implemented method of any one of claims 42-44, wherein the electrical stimulation reduces bradykinesia and increases vigor of the intended movement compared to in absence of the electrical stimulation.

46. The computer implemented method of any one of claims 42-45, wherein the brain electrical signal data comprises field potential data.

47. The computer implemented method of any one of claims 42-46, wherein a machine learning algorithm is used to generate the movement classification model.

48. The computer implemented method of claim 47, wherein the machine learning algorithm is a supervised machine learning algorithm.

49. The computer implemented method of claim 47 or 48, wherein the movement classification model distinguishes between when the subject is in an intended movement state or a stationary state.

50. The computer implemented method of claim 49, further comprising training a machine learning model to predict whether the subject is in an intended movement state using accelerometry data for detected movement of the subject.51 . The computer implemented method of claim 50, wherein the accelerometry data is from a wearable accelerometer worn by the subject that detects natural undirected movement of the subject.

52. The computer implemented method of claim 50, wherein the accelerometry data is from a wearable accelerometer worn by the subject that detects movement during directed movement tasks selected from finger tapping, opening or closing of a hand, wrist pronation or supination, walking, and keyboard typing.

53. The computer implemented method of any one of claims 49-52, further comprising performing principal components (PC) decomposition on power spectral data to identify personalized power band features for distinguishing the intended movement state from the stationary state.

54. The computer implemented method of claim 53, further comprising: transforming a time-domain of a local field potential (LFP) into a spectrogram using a short- time fourier transform (STFT); performing principal components analysis (PCA) on z-scored STFT outputs to identify coordinated bands of neural activity; and applying a peak-finding algorithm to PCA component weights to find a frequency range best approximating each principal component.

55. The computer method of claim 54, wherein top principal components for the contralateral cortex region or the subthalamic nucleus region correspond to personalized frequency ranges for beta frequency neural oscillations and gamma frequency neural oscillations for the brain electrical signal associated with the intended movement of the subject.

56. The computer implemented method of any one of claims 49-55, wherein the electrical stimulation is selectively increased during the intended movement state and not during the stationary state.

57. The computer implemented method of any one of claims 49-56, further comprising using accelerometry data, gyroscope data, multi-view video recordings of the subject, keylogging data from a computer used by the subject, or any combination thereof in combination with the recorded brain electrical signal data to distinguish between the intended movement state and the stationary state.

58. The computer implemented method of any one of claims 42-57, further comprising: a) ranking predicted stimulation effectiveness for available settings of a DBS device based on classifier scores for stimulation effectiveness of each setting using a linear classification model; b) selecting stimulation settings predicted to have highest stimulation effectiveness based on the linear classification model; c) receiving recorded brain electrical signal data from the contralateral cortex region and the subthalamic nucleus region of the brain of the subject after applying electrical stimulation with the DBS device to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject using the settings predicted to have the highest stimulation effectiveness; d) analyzing the recorded brain electrical signal data to evaluate neural response of the subject to the electrical stimulation; e) updating the linear classification model based on the neural response of the subject to the electrical stimulation to generate an updated linear classification model; f) updating the ranking of predicted stimulation effectiveness for the available settings of the DBS device using the updated linear classification model; g) selecting stimulation settings predicted to have the highest stimulation effectiveness based on the updated linear classification model; h) receiving recorded brain electrical signal data from the contralateral cortex region and the subthalamic nucleus region of the brain of the subject after applying the electrical stimulation with the DBS device to the subthalamic nucleus region or the globus pallidus internus regionof the brain of the subject using the settings predicted to have the highest stimulation effectiveness based on the updated linear classification model; and i) repeating e) - h) to adjust the available settings of the DBS device to optimize stimulation effectiveness.

59. The computer implemented method of claim 58, wherein the linear classification model uses linear discriminant analysis (LDA) to adjust amplitude of current and frequency of the electrical stimulation.

60. The computer implemented method of any one of claims 42-59, wherein the subject has Parkinson’s disease, Parkinsonism, Lewy body dementia, progressive supranuclear palsy, corticobasal degeneration, Huntington's disease, dystonia, stroke, or multiple system atrophy.61 . A non-transitory computer-readable medium comprising program instructions that, when executed by a processor in a computer, causes the processor to perform the method of any one of claims 42-60.

62. A kit comprising the non-transitory computer-readable medium of claim 61 and instructions for determining symptom severity of a subject having a movement disorder.

63. A system for treating bradykinesia in a subject, the system comprising: a stimulation electrode adapted for positioning at a first location in a subthalamic nucleus region or a second location in a globus pallidus internus region of the brain of the subject to deliver electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region; a first neural recording electrode adapted for positioning at a third location in the subthalamic nucleus region of the brain of the subject to record brain electrical signal data; a second neural recording electrode adapted for positioning at a fourth location in a contralateral cortex region of the brain of the subject to record brain electrical signal data; and a processor programmed according to the computer implemented method of any one of claims 42-60 to instruct the stimulation electrode to apply an electrical stimulation to the subthalamic nucleus region or the globus pallidus internus region of the brain of the subject in a manner effective to provide a prokinetic effect on the intended movement of the subject when the brain electrical signalassociated with the intended movement is detected using the first neural recording electrode or the second neural recording electrode.

64. The system of claim 63, wherein the brain electrical signal data comprises field potential data.

65. The system of 63 or 64, further comprising an accelerometer, a gyroscope, or a combination thereof.

66. The system of claim 65, wherein the accelerometer, the gyroscope, or the combination thereof is provided by a wearable device.

67. The system of claim 66, wherein the wearable device is a smartwatch.

68. The system of any one of claims 63-67, further comprising a video recording device.

69. The system of any one of claims 63-68, wherein the contralateral cortex region comprises a precentral gyrus region, a postcentral gyrus region, or both the precentral gyrus region and the postcentral gyrus region.

70. The system of any one of claims 63-68, wherein the contralateral cortex region comprises the central sulcus.71 . The system of any one of claims 63-70, wherein the stimulation electrode is a nonbrain penetrating surface electrode array or a brain-penetrating electrode array.

72. The system of any one of claims 63-71 , wherein the first neural recording electrode or the second neural electrode is a non-brain penetrating surface electrode array or a brainpenetrating electrode array.

73. The system of any one of claims 63-72, wherein the first neural recording electrode or the second neural electrode is an electroencephalogram (EEG) electrode array, a subgaleal orburrhole mounted or cranially mounted neurostimulator electrode, or an electrocorticogram (ECoG) electrode array.

74. The system of claim 73, wherein the ECoG electrode array spans regions of the precentral gyrus and the postcentral gyrus region that control hand and arm movement.

75. The system of claim 73, wherein the ECoG electrode array spans the central sulcus.

76. The system of any one of claims 63-75, wherein the system further comprises a user interface comprising an input electronically coupled to the processor for instructing the stimulation electrode to apply an electrical stimulation to the subthalamic nucleus region to treat the bradykinesia in the subject.

77. The system of claim 76, wherein the user interface is password protected and is operable by a health care practitioner.

Images