Richard B Sowers (Chair, Director of Research)
Carolyn L Beck (Member)
T Kesavadas (Member)
Manuel Hernandez (Member)
Gait Data and Machine Learning: From Clinical Data to Knowledge
We propose new ideas that address a recurring problem in neurological gait disorders, namely disease prediction and disease progression prediction. Specifically, we seek to identify and classify motor impairments by adapting machine-learning algorithms to data-driven dynamical models of gait. We explore a variety of multimodal kinematic (e.g., spatiotemporal gait metrics, acceleration), kinetic (e.g., ground reaction forces), and electrophysiological (e.g., electromyography (EMG), electrocardiography (ECG)) signals. Our success is measured by our ability to use these signals, collected during walking, to classify disability and predict progression of cognitive and motor changes in persons with neuromuscular disorders. Our work is a multidisciplinary effort that involves novel combinations of sensors, vision, machine learning, biomechanics, and dynamical analyses to better characterize normal and impaired movement.