Pattern matching techniques to automatically detect range of movement tests from wearable sensors

Wearable sensor technology has steadily grown in availability within a wide variety of well-established consumer and medical devices. Wearable sensors have been used in many healthcare applications to monitor patients at home and throughout their rehabilitation. Data collected from wearable sensors allow monitoring of patient recovery during rehabilitation and assist clinicians in diagnosing. Activities of Daily Living (ADL) is considered as an assessment criterion for various disease conditions. Wearable devices enable the collection of information associated with different range of movement (ROM) tests that measure ADL. In an ambulatory monitoring setting, the volume of data collected by wearable sensors can become complex and challenging to process. Extraction of ROM tests can be laboursome, and often fraught with misclassification of movement. Hence it is difficult to analyse and make conclusions/predictions from movement datasets using manual assessment techniques. This paper examines whether ROM tests can be automatically detected and extracted from wearable sensor data using Artificial Intelligence (AI) techniques.This research examines and discusses clinical trial data collected from patients suffering from Axial SpondyloArthritis (AxSpA). AxSpA is a disease that affects spinal cord mobility. In this trial, Inertial Measurement Unit (IMU) sensors are attached to the lower back and neck of the patient, and data corresponding to clinical trial movements are recorded. An AI system is trained and tested using these datasets, and the prediction accuracy of the system is examined. The system will be capable of detecting ROM tests within long-term datasets once the AI system used in this analysis is sufficiently trained by an adequate amount of data for efficient pattern recognition.