Computational Modeling of 60 Hz Subthalamic Nucleus Deep Brain Stimulation for Gait Disorder in Parkinson's Disease

Computational Modeling of 60 Hz Subthalamic Nucleus Deep Brain Stimulation for Gait Disorder in Parkinson's Disease

Computational Modeling of 60 Hz Subthalamic Nucleus Deep Brain Stimulation for Gait Disorder in Parkinson's Disease

National Institute of Neurological Disorders and Stroke (NINDS), The University of Tennessee, Knoxville

The objective of this study is to further the understanding and application of 60Hz subthalamic deep brain stimulation (STN-DBS) in Parkinson's patients with gait disorder. The investigators will achieve this through 2 study aims: Determine the impact of 60Hz subthalamic deep brain stimulation on gait kinematics using wearable sensors Develop machine learning models to predict optimal subthalamic deep brain stimulation frequency based on wearable sensors

Gait disorder, which manifests as shuffling, reduction in speed, multistep turning, and/or freezing of gait (FOG), can arise later in the Parkinson's disease (PD) course and cause significant disability. Ultimately, patients are at risk for falls and can become socially isolated due to their mobility limitations. These symptoms tend not to respond to high frequency STN-DBS. However, lower frequency stimulation (60-80Hz) of the STN in treating gait disorder and/or freezing of gait has demonstrated benefit. This study potentially can expand knowledge of 60hz DBS while improving its utilization in combination with PD medications-enabling sustainable and possibly predictable therapeutic benefit.

Each DBS electrode contact will be reprogrammed in 60hz and High Frequency Stimulation (180hz) in the Levodopa ON (medicated) and OFF (unmedicated) conditions.

Inertial Sensors will be used to quantify gait metrics (postural sway, gait cycle, circumduction) as participants conduct two 7meter walking trials for each stimulation condition ( 60Hz frequency or High Frequency) across DBS electrode pairs in both the medicated and unmedicated states.

Regression models will be created using non-linear regression analysis based on random forest (RF) classifier on the raw gait sensor data acquired from the medicated and unmedicated states.

Regression models will be created using non-linear regression analysis based on random forest (RF) classifier on all raw sensor data (gait and balance, tremor, and speed of limb movements) acquired in the medicated and unmedicated states.

The difference in tremor (e.g. rest, postural, kinetic) severity will be measured with an Inertial sensor for each DBS electrode stimulation pair (60hz or High Frequency) in both the medicated and unmedicated states.

The difference in the speed of limb movements (e.g. finger taps, hand grasps, wrist rotation, leg lifts, toe taps) will be measured with an Inertial sensor for each DBS electrode stimulation pair (60hz or High Frequency) in both the medicated and unmedicated states.

Inclusion Criteria: Male or female, aged 21-80 Patients diagnosed with Parkinson's disease (PD) PD subjects who have bilateral STN-DBS (greater than 3 months) or in the preoperative stage of being implanted with bilateral STN-DBS Have underlying gait disorder Currently treated with oral levodopa therapy Willingness to comply with all study procedures Exclusion Criteria: Cognitive deficits based on historical record that limit participant compliance with study protocol Vestibular disorder or musculoskeletal problems affecting gait or balance