Presentation Title
Wearables and Deep Learning Classify Fall Risk from Gait in Multiple Sclerosis
Abstract
Falls are a significant problem for persons with multiple sclerosis (PwMS). Clinical fall risk assessments are reactive, which prevents the deployment of preventative interventions and motivates the need for new fall risk assessments in this population. Previous research has identified measures of gait biomechanics that differ between PwMS who have fallen and those who have not, but these biomechanical indices have not yet been leveraged to detect PwMS who have fallen and thus capturing a measure of fall risk. Moreover, they require the use of laboratory-based measurement technologies, which prevents clinical deployment. Here we demonstrate that a bidirectional long short-term (BiLSTM) memory deep neural network was able to identify PwMS who have recently fallen with good performance (AUC of 0.88) based on accelerometer data recorded from two wearable sensors during a one-minute walking task. These results provide substantial improvements over machine learning models trained on spatiotemporal gait parameters (19% improvement in AUC), statistical features from the wearable sensor data (16%), and patient-reported (19%) and neurologist-administered (29%) measures in this sample. The success and simplicity (two wearable sensors, only one-minute of walking) of this approach indicates the promise of inexpensive wearable sensors for capturing fall risk in PwMS.
Primary Faculty Mentor Name
Ryan McGinnis
Secondary Mentor Name
Nick Cheney
Graduate Student Mentors
Lindsey Tulipani, Reed Gurchiek
Faculty/Staff Collaborators
Lindsey J. Tulipani (Graduate Student Mentor), Reed D. Gurchiek (Graduate Student Mentor), Dakota A. Allen, Lukas Adamowicz, Dale Larie, Andrew J. Solomon (Medical Center Collaborator), Nick Cheney (Collaborating Mentor), and Ryan S. McGinnis (PI)
Status
Undergraduate
Student College
College of Engineering and Mathematical Sciences
Program/Major
Biomedical Engineering
Primary Research Category
Health Sciences
Wearables and Deep Learning Classify Fall Risk from Gait in Multiple Sclerosis
Falls are a significant problem for persons with multiple sclerosis (PwMS). Clinical fall risk assessments are reactive, which prevents the deployment of preventative interventions and motivates the need for new fall risk assessments in this population. Previous research has identified measures of gait biomechanics that differ between PwMS who have fallen and those who have not, but these biomechanical indices have not yet been leveraged to detect PwMS who have fallen and thus capturing a measure of fall risk. Moreover, they require the use of laboratory-based measurement technologies, which prevents clinical deployment. Here we demonstrate that a bidirectional long short-term (BiLSTM) memory deep neural network was able to identify PwMS who have recently fallen with good performance (AUC of 0.88) based on accelerometer data recorded from two wearable sensors during a one-minute walking task. These results provide substantial improvements over machine learning models trained on spatiotemporal gait parameters (19% improvement in AUC), statistical features from the wearable sensor data (16%), and patient-reported (19%) and neurologist-administered (29%) measures in this sample. The success and simplicity (two wearable sensors, only one-minute of walking) of this approach indicates the promise of inexpensive wearable sensors for capturing fall risk in PwMS.