Towards Automated Measurement of Rehab Dose and Response

Abstract

Careful measurement of dose and response is a central tenant of biomedical research, but is largely lacking in current rehabilitation practice after stroke. A major rate limiting step has been the availability of precise, practical tools to measure upper limb movements (dose) and movement quality (response) in rehabilitating stroke patients. In this talk, Dr. Schambra will discuss work in her lab to build these tools. She will discuss the concept of movement primitives as fundamental units of measure, the design of deep learning approaches to count and appraise primitives from inertial sensor data, and new directions using video-based motion capture.