Gait signal classification using an in-house built goniometer and naïve Bayes classifier

Abstract

This work aims at designing and implementing a knee and an ankle goniometer, both based on potentiometry, and applying the naive Bayes classifier on the signals obtained from the goniometers to differentiate between male and female gait signals, and to also differentiate between healthy and restricted knee gait signals. Gait signals and other parameters were collected from 60 subjects using the goniometers and WEKA was used to classify this data. The designed goniometers were 97.8% accurate and the naive Bayes classifier was highly accurate in categorising the signals with an accuracy of at least 86.7%.