Computational modeling of 60Hz STN DBS for gait disorder in Parkinson's disease

Feinstein Institute For Medical Research, Manhasset NY

Investigators

Linked publications & trials

Abstract

Project Summary The therapeutic gain of subthalamic deep brain stimulation (STN-DBS) for Parkinson?s disease (PD) has relied on high frequency stimulation (HFS) (e.g., 130-185Hz) to alleviate cardinal motor symptoms such as rigidity, bradykinesia and tremor, along with levodopa motor complications (e.g., motor fluctuations and dyskinesia). As the disease progresses, gait disorder, including freezing of gait, may emerge, increasing the risk of falls while limiting socialization and quality of life. Oftentimes, such symptoms are recalcitrant to dopaminergic medications and STN-DBS. Studies suggest that low frequency stimulation (60-80Hz) may improve gait in those with chronic DBS. Furthermore, we demonstrated that 60Hz STN-DBS provides gait improvement within the first year of DBS implantation. However, there is limited data on the nature of this response, in terms of what aspects of gait respond and its relation to dopaminergic medications. In addition, based on preliminary results, it is possible to use machine-learning models to identify patients who may overall benefit from 60Hz STN-DBS based solely on preoperative assessments. However, the extent to which patients? symptoms may improve under 60Hz STN-DBS needs to be further investigated. This project aims to (1) investigate the impact of 60Hz STN-DBS on gait kinematics using wearable sensors; and (2) develop machine learning models to predict response of STN-DBS patient symptomatology at both high frequency and 60Hz based on sensor profiles of PD motor symptoms at baseline. Parkinson?s patients with gait disorder who will either be undergoing STN-DBS implantation or already have chronic bilateral STN-DBS will be recruited. We will conduct the study in two phases: Phase I ? obtain baseline kinematic profile for each subject in terms of tremor, bradykinesia, and gait using objective sensor measurements in the unmedicated and levodopa-ON states. We will specifically assess gait parameters within three domains: postural sway, gait speed and rhythm (cadence), and circumduction. Phase II ? participants will be evaluated in two conditions: 1) OFF medication/ON stimulation; 2) ON medication/ON stimulation. Sensor-based gait analysis will be done in 60Hz and HFS modes for each DBS electrode contact pair in the different medication states. Following data collection, statistical analysis will be performed and machine learning models will be developed to accomplish aims (1) and (2). Accordingly, this study will help determine the aspects of gait influenced by 60Hz stimulation and how it is affected by levodopa, while potentially identifying certain 60Hz responsive PD phenotypic subtypes. It will also allow for instantaneous recommendations by artificial intelligence on which stimulation frequency is best suitable for a subject based on a priori sensor measurements of motor symptoms. The ranking provided can guide clinical evaluations and programming through a sensor guided paradigm.