Inertial joint angle estimation – How accurate is accurate enough?

Abstract

This paper investigates the accuracy requirements for inertial measurement unit (IMU)-based joint angle estimation in rehabilitation and physical therapy. Current clinical practice relies on visual inspection during gait and isolated functional tests using goniometers. While visual estimation has high variability, isolated assessments achieve SEM values between 1.5° and 6°. For home use, automated systems must achieve both low random error and minimal bias to provide objective long-term monitoring. Clinical relevance is determined not solely by measurement precision, but also by minimal clinically important changes. Results indicate that IMU-based systems do not need to match goniometer accuracy to be clinically useful.