Personnel

PI: Sujit Dey

Ph.D. Student: Alex Postlmayr

M.S. Student: Bhanu Garg

B.S.E Students: Shubham Kumar, Gaopo Huang

Overview

Physical therapy is proven effective for those who fully complete their training. However, many patients do not fully complete training due to low efficacy and motivation. To address these issues, this project is focused on building a virtual physical therapy system for mobile applications, coined “PocketPT”. This will allow patients to perform exercises at home, outside of the clinic with or without physical therapist (PT) supervision. We propose a smartphone application that can track, evaluate, and provide feedback to patients performing their prescribed PT exercises.

Mobile Device Virtual Physical Therapist

Traditionally, motion capture technologies rely on external sensors or specialized cameras that can reliably capture the 3D movement of the individual. In comparison, our system uses machine learning to estimate the  3D skeleton movement from the RGB video captured by a conventional smartphone camera. This allows for our application to be used by any individual with access to a smartphone and internet. The application will be built for mobile devices, where patients watch an exercise demonstration, record themselves performing the exercises, and receive in real-time feedback (auditory/visual) on their movements. The PT is able to see the scores and progress of the patients and can use this data to update the exercises as needed."

Broader Impact

The project has demonstrated the feasibility and efficacy of detecting and analyzing human motion captured by a smartphone camera using novel machine vision algorithms. We are building a tool that allows for physical therapists to collect data on their patients and monitor the patient’s adherence and progress at home, which previously was not possible. Furthermore, we are enhancing patient experience with our application by allowing them to receive feedback on the prescribed exercises, between physical therapy sessions.