Prediction of Loss-of-Balance of the Human Based on Plantar Pressure by Using the SA-RF Algorithm

Abstract

Elderly people are easy to suffer from accidental injuries due to loss-of-balance in daily life. Wearable sensing technology is promising for detecting or predicting loss-of-balance events. This paper proposes a human loss-of-balance prediction method based on a customized wearable plantar pressure sensing system. To realize accurate prediction of loss-of-balance, we integrate the simulated annealing algorithm (SA) and the random forest algorithm (RF) to construct a SA-RF prediction model, where the input of the model is the plantar pressure data of the feet and the output of the model is the label of the human motion state. To validate the effectiveness of the proposed SA-RF model, 15 healthy subjects participated in the experiments. The experimental results show that the classification and recognition accuracy of the SA-RF model are significantly improved compared to the RF model, especially for the recognition of the easily loss-of-balance state. The accuracy of the proposed SA-RF model reaches 90%, which is a 5% improvement compared to the RF model. Therefore, the use of the SA-RF model based on plantar pressure can effectively predict loss-of-balance and thus has the potential to be integrated into fall prevention applications.