A Smart Crutch Tip for Monitoring the Activities of Daily Living Based on a Novel Neural-Network Intelligent Classifier

Abstract

The determination of the objectives of gait rehabilitation therapies is usually based on partial data provided by clinical tests carried out in specific scenarios and the subjective perception of both the therapist and the patient. However, recent studies have shown that individualization is mandatory to maximize the effect of the therapy on the patient. This requires monitoring the Activities of Daily Living of the patient using objective indicators and measurements, which can be achieved using instrumented devices or wearable sensors. In this work, a smart crutch tip is proposed, which integrates a novel neural-network based intelligent Activities of Daily Living classifier. Based on the sensors integrated on the tip, the classifier is able to detect four typical activities (walking, standing still, going up stairs and going down stairs). In order to design the classifier, data from a group of 13 volunteers is used and different network architectures are evaluated in order to consider the most computationally efficient design, obtaining a success rate of 95%.