A Wearable System to Monitor Gait Modification

Gait modification has been shown to have positive effects on patients with knee osteoarthritis. However, it is challenging to detect whether the patients achieved sufficient modification during gait. To address this, we present a data-driven approach to differentiate between various gaits performed by 20 healthy controls. We calculated features that captured Ground Reaction Force and Center of Position and applied Random Forest to differentiate walking pattern. We analyzed important features that may relate to gait modification. Our experimental results show that our model achieved good performance with an average F-1 score of 0.74 and an average AUC of 0.90. With further development and testing, we believe that our method can be deployed on wearable platforms to improve the rehabilitation progress of patients with osteoarthritis in clinical settings.