Foot Posture Gait Values Analysis Using Xgboost Algorithm for Cerebellar Ataxia

Abstract

Patients with Cerebellar Ataxia (CA) may benefit from the use of predicting systematic flow of motions with analysis of gait (AoG) prediction. The importance of this research is to identify the real model for evaluate the poor gait pattern in the human movements. The foot positions are captured and used for this research. The machine learning model will predict the neurology disease using the foot gait values. The accelerometer was attached to the patients to capture the walking positions. 12 different walking styles were carried out with the walking speed of 0.6 to 1.7 m/s. The values and motion movements are carried for gait analysis to find the disease prediction. Two AoG prediction models based on step-based abnormal gait patterns and conventional AoG generated from the signals sensing features were built using XGBoost to assess whether the use of the unsteady gait characteristics can more effectively identify AoG. Compare with the existing model, Ensemble method is providing better accuracy for foot gait value analysis.