Footwear-integrated force sensing resistor sensors: A machine learning approach for categorizing lower limb disorders

Abstract

Lower limb disorders are a substantial contributor to both disability and lower standards of life. The prevalent disorders affecting the lower limbs include osteoarthritis of the knee, hip, and ankle. The present study focuses on the use of footwear that incorporates force-sensing resistor sensors to classify lower limb disorders affecting the knee, hip, and ankle joints. The research collected data from a sample of 117 participants who wore footwear integrated with force-sensing resistor sensors while walking on a predetermined walkway of 9 meters. Extensive preprocessing and feature extraction techniques were applied to form a structured dataset. Several machine learning classifiers were trained and evaluated. According to the findings, the Random Forest model exhibited the highest level of performance on the balanced dataset with an accuracy rate of 96%, while the Decision Tree model achieved an accuracy rate of 91%. The accuracy scores of the Logistic Regression, Gaussian Naive Bayes, and Long Short-Term Memory models were comparatively lower. K-fold cross-validation was also performed to evaluate the models’ performance. The results indicate that the integration of force-sensing resistor sensors into footwear, along with the use of machine learning techniques, can accurately categorize lower limb disorders. This offers valuable information for developing customized interventions and treatment plans.