Walking pattern analysis using the gait cycle to classify a healthy and an unhealthy form

Abstract

This study proposed an algorithm to classify normal and abnormal forms of walking patterns by identifying the swing and stance phases, known as the gait cycle. The walking patterns were examined by collecting data using Razor IMU sensors. The Wi-Fi transmitter was utilized to transfer raw data for further analysis, representing the gait cycle using a linear graph. The data were preprocessed and transformed into phase graphs using the polar coordinate equation. To identify areas of density, data were incorporated into K-mean clustering. A k-value of 2 groups represented the major walking phases and classified patterns based on a regional model. The experimental results enable us to define healthy and unhealthy forms. This algorithm conveniently assists in diagnosing basic physical conditions, and the dataset is efficient for monitoring the medical treatment process among recipients.