AI-Empowered Automatic Evaluation of Sitting-Pivot-Transfer Performance Based on 3D Body Motion Tracking and Ambient Object Recognition Using a Time-Of-Flight Sensor

Abstract

This research developed AI technology to automatically assess wheelchair transfer quality using body motion tracking and object recognition via low-cost time-of-flight sensors. A study involving 55 full-time wheelchair users performing sitting-pivot-transfers achieved 93.76% accuracy (95.11% precision and 95.18% recall) across 13 assessed transfer components. The work demonstrates potential to automate the process of transfer assessment and documentation of results, eliminating the need for extensive training and reducing the time taken to assess a patient.