Leveraging wearable sensors and machine learning for posture-based detection of carpal tunnel syndrome

Abstract Carpal tunnel is associated with long-term use of the wrist and hand for various activities such as typing, welding, or poor working postures. Carpal Tunnel Syndrome (CTS) may cause severe pain and discomfort in the hand and wrist, and in some circumstances, surgery becomes inevitable. The objective of this study is to prevent typing postures, which can be ascertained as predisposing subjects to CTS development. The data used in this study is an array of wrist wearable sensors to capture flexion, extension, and bending of fingers while using a keyboard or mouse. Machine learning is employed on the data in order to identify risk factors indicative of a high probability of CTS. The analyzed models are linear regression, Support Vector Machine, Random Forest, Multilayer Perceptron, Convolution Neural Network, and Long Short Term Memory. The conditions for assessing the performance of the data models include RMS error, coefficients of determination, and mean absolute percentage error. In this research, I conducted an exploratory data analysis (EDA) to gain an initial understanding of the dataset. Following the exploratory phase, I applied feature extraction techniques, specifically Principal Component Analysis (PCA). As put forward for the proposed research, the strategies to prevent risky occupations have broad potential at the present time, especially in the case of CTS when preventing repetitive wrist movements.