Mind the Steps: Towards Auditory Feedback in Tele-Rehabilitation Based on Automated Gait Classification

We describe a proof-of-concept for the implementation of a mobile auditory biofeedback system based on automated classification of functional gait disorders. The classification is embedded in a sensor-instrumented insole and is based on ground reaction forces (GRFs). GRF data have been successfully used for the classification of gait patterns into clinically relevant classes and are frequently used in clinical practice to quantitatively describe human motion. A feed-forward neural network that was implemented on the firmware of the insole is used to estimate the GRFs using pressure and accelerator data. Compared to GRF measurements obtained from force plates, the estimated GRFs performed highly accurately. To distinguish between normal physiological gait and gait disorders, we trained and evaluated a support vector machine with labeled data from a publicly accessible database. The automated gait classification was sonified for auditory feedback. The high potential of the implemented auditory feedback for preventive and supportive applications in physical therapy, such as supervised therapy settings and tele-rehabilitation, was highlighted by a semi-structured interview with two experts.