A wearable sensor system combines motion tracking with machine learning to deliver real-time, objective gait assessment, helping clinicians monitor prosthetic rehabilitation more accurately and efficiently.
Researchers have developed and validated a wearable inertial measurement unit (IMU) sensor system that objectively measures walking quality in people with lower-limb amputations. Published in the journal Sensors, the study introduces a machine learning framework that converts movement data into a continuous Gait Goodness Score (GGS), providing clinicians with an objective measure of prosthetic walking performance.
Traditional gait assessment typically relies on specialised motion analysis laboratories equipped with infrared cameras and force plates. While highly accurate, these facilities are expensive, time-consuming and impractical for routine clinical use, leaving therapists to depend largely on visual observation and simple walking tests.
The wearable system addresses this limitation by using four IMU sensors positioned above and below both knees, including on the prosthetic limb. The sensors continuously record acceleration and angular velocity data, transmitting the information wirelessly to analysis software. Rather than relying on deep learning, the researchers used a feature-based machine learning approach to identify key gait characteristics, including walking speed, step length, gait cycle duration and knee movement.
The model was trained using data from 71 individuals with unilateral lower-limb amputations and validated independently in 120 participants treated at military and Veterans Affairs clinics. Eighty-three participants provided complete sensor datasets for analysis. Results showed the wearable system reliably distinguished different levels of mobility, with higher GGS values associated with better walking performance, faster gait speeds and improved scores on established clinical mobility assessments.
The system also demonstrated high reliability during repeated walking trials and maintained accuracy under moderate sensor noise. Researchers say the portable platform could support routine rehabilitation by providing immediate biomechanical feedback, helping clinicians optimise prosthetic alignment, socket fit and therapy decisions.
Future research will focus on extending the technology beyond clinics into home-based monitoring and integrating it with biofeedback systems that could alert users to gait abnormalities in real time, supporting long-term rehabilitation and reducing the risk of secondary injuries.
