Detection of Asymmetric Abnormalities in Gait using Depth Data and Dynamic Bayesian Networks

The objective of this study is to develop a framework to identify gait abnormalities, such as a limp in either leg, and determine the degree/severity of such abnormalities. Monitoring and evaluating gait characteristics over time may enable early diagnosis of diseases and help prevent their complications in healthcare facilities. This paper focuses on non-invasive no-marker gait data acquisition using depth sensor. This data are used to distinguish the gait abnormality severity, by using various classifiers such as LDA and SVM. For gait abnormality analysis over time, a Dynamic Bayesian Network is designed and used as an alternative classifier.