iKnow how you walk — A smartphone based personalized gait diagnosing system

Humans, due to aging and hormonal changes are prone to pains in their limbs. As a result of which, the fundamental activity of humans i.e., movement pattern of limbs also known as gait is affected. By exerting unequal weight on both limbs in-order to avoid pain in one leg, humans slowly develop an abnormal gait pattern consisting of limping and sideways bend in the posture. This often goes unnoticed for a long time. We propose a system using smartphones that can sense and detect abnormal walking patterns. The sensing of limb movement is performed by an embedded accelerometer in a smartphone and detection of abnormal walk pattern is performed by classifying different features such as stride length, walk velocity etc. By incorporating Naive Bayes and Decision trees classifiers, we obtained close to 89% accuracies in classifying different levels of abnormalaties. Further validation was done by implementing a decision tree based gait variation detector smartphone application across five users which resulted in 90% accuracy.