Presentation Title
Classifying Postural Features from Patients with Multiple Sclerosis
Abstract
For patients with Multiple Sclerosis (MS) the risk of falling due to their decreased mobility is a serious concern. One in two MS patients will experience a fall in any three-month period. These falls decrease a patient’s quality of life as well as impair their ability to perform physical therapy and remain active, both of which are effective treatments for persons with MS. Individuals with MS are hypothesized to exhibit different patterns of movement than healthy individuals. Analysis of walking and standing are of particular interest as the postural metrics associated with them are likely sensitive enough to differentiate MS patients from healthy individuals. 40 patients with multiple sclerosis performed in lab movement tests while recording accelerometer data from the medial chest and right anterior thigh sensor locations. These same patients recorded data for an additional 48 hour period at home. Using in lab and in home accelerometer data from these subjects, activity identification and postural metrics were calculated. Training a support vector machine (SVM) using a Gaussian kernel on accelerometer data achieved 99.85% accuracy, 99.79% sensitivity, 99.94% specificity, and 99.96% for identifying standing activity. Additional postural metrics (time domain, frequency domain and JERK measures) were quantified from accelerometer data to help predict fall risk.
Primary Faculty Mentor Name
Dr. Ryan McGinnis
Secondary Mentor Name
Andrew Solomon
Graduate Student Mentors
Reed Gurchiek, Lindsey Tulipani
Status
Undergraduate
Student College
College of Engineering and Mathematical Sciences
Program/Major
Mechanical Engineering
Primary Research Category
Engineering & Physical Sciences
Secondary Research Category
Engineering & Physical Sciences
Tertiary Research Category
Engineering & Physical Sciences
Classifying Postural Features from Patients with Multiple Sclerosis
For patients with Multiple Sclerosis (MS) the risk of falling due to their decreased mobility is a serious concern. One in two MS patients will experience a fall in any three-month period. These falls decrease a patient’s quality of life as well as impair their ability to perform physical therapy and remain active, both of which are effective treatments for persons with MS. Individuals with MS are hypothesized to exhibit different patterns of movement than healthy individuals. Analysis of walking and standing are of particular interest as the postural metrics associated with them are likely sensitive enough to differentiate MS patients from healthy individuals. 40 patients with multiple sclerosis performed in lab movement tests while recording accelerometer data from the medial chest and right anterior thigh sensor locations. These same patients recorded data for an additional 48 hour period at home. Using in lab and in home accelerometer data from these subjects, activity identification and postural metrics were calculated. Training a support vector machine (SVM) using a Gaussian kernel on accelerometer data achieved 99.85% accuracy, 99.79% sensitivity, 99.94% specificity, and 99.96% for identifying standing activity. Additional postural metrics (time domain, frequency domain and JERK measures) were quantified from accelerometer data to help predict fall risk.