Distinguish Between Obese and Normal Body Types Through Gait Analysis Using Classification Models

For gait analysis, an IMU sensor was mounted on the knee and gait related data was collected. Various gait parameters such as gait time, stance swing ratio, heel strike, and toe off can be extracted from the dataset. To explore the relationship between gait parameters and individual gait characteristics, we analyzed the gait patterns of normal and obese people were analyzed based on BMI (Body Mass Index). To apply it to a classification model of machine learning, different gait cycles between subjects were normalized. Gait data was collected from eight subjects in their 20s. Using this dataset, we applied a logistic regression model, and obtained the classification accuracy of 92%. We also investigated the correlation between BMI and gait parameters and found that, the correlation between BMI and cadence was -0.66.