AI-driven gait classification using portable wearable sensors: advances and case study

Abstract

This chapter explores the latest advancements in AI-driven gait classification. The work mainly focuses on developing and implementing techniques for characterizing and measuring human gait using a wearable portable sensor system. By leveraging deep learning (DL) and machine learning (ML) models, we aim to accurately classify different types of gait activity performed by the user. The chapter also presents a review of recent advancements in gait classification, focusing on machine learning and deep learning techniques. A case study is discussed to illustrate the practical applications and benefits of these technologies. For the case study, we collect the gait data using IMU (Inertial Measurement Unit) sensor attached on right shank of a subject. The data is collected from 20 young subjects, who performed six gait activities: (i) walking uphill, (ii) walking downhill, (iii) walking on ground, (iv) climbing upstairs, (v) climbing downstairs, and (vi) standing, all at a speed comfortable to them. We train and test multiple ML and DL models for gait classification, presenting comprehensive experimental results with performance evaluation. Overall, this chapter explores advancements in gait classification with a focus on wearable sensor systems and machine learning techniques, along with a case study relevant to gait rehabilitation applications.