Flexible Piezoresistive Pressure Sensor with Direct Electrode Contacts for Knee Movement Classification

Abstract

Gait analysis is one of the important tools for obtaining physical information of arthritis patients. Vital signals collected from individuals give the necessary information to recognize the knee movement for patients with knee pain. Herein, a kind of conductive polymer is prepared and investigated to develop a piezoresistive pressure sensor to monitor pressure variation during the knee movement in real time. Pressure sensor developed using the casting method is characterized by its ease in preparation, surface resistance, bulk resistance and impedance with low frequency dependence. In the experiments, the prepared flexible piezoresistive sensor film is easily inserted between two copper foil electrodes for better conductivity. Signals for four types of knee joint movements viz., bend, raise, half squat, knee extension are collected in real time. The main characteristics of these four movements were detected using selected statistical time domain features. Decision Tree, Navie Bayes and Random forest classifiers were implemented and compared. While a smaller number of training and test samples, decision tree classifier achieved better accuracy, less time for model training and model prediction.