Enabling Unobtrusive Gait Analysis for Self-Health Monitoring using a Smartphone

Gait analysis states that the scientific study of body movements are responsible for locomotion in human beings. Many people can move about with abnormal gait patterns for years without knowing the odd symptoms. However, an abnormal walking style can lead to more significant health problems when someone's regular gait pattern is altered due to injury or illness. Therefore, one must self-assess and diagnose the early stages of abnormal gait. We propose an interdisciplinary implementation of the Human Gait Analysis System project (HGAS), a human pose estimation using the PoseNet machine learning model that can run on a smartphone browser. Gait analysis conducted in a laboratory environment is expen-sive and cannot be used to assess the person's gait daily. There-fore, it is desired to focus on developing a new gait analysis system that is affordable for personal use, user-friendly, and mobile. The present work explores machine learning techniques that can be used to determine basic gait parameters using observational gait analysis. The individual can utilize the developed system for a self-gait assessment, help doctors/nurses to understand the patient gait patterns for their treatment, or an individual pose assessment during yoga/exercises.