Summary

The integration of machine learning into technological systems is revolutionizing health care by enhancing patient care and increasing efficiency. This study highlight the feasibility and effectiveness of using wearable sensors and ML algorithms to accurately classify shoulder rehabilitation exercises.

Abstract

Background

Rehabilitation treatment for shoulder disorders is essential for resuming daily activities and preventing chronic conditions. The integration of machine learning (ML) into technological systems is revolutionizing health care by enhancing patient care and increasing efficiency. Specifically, the classification of shoulder rehabilitation exercises holds significant potential for improving the monitoring of exercises and patient adherence to the prescribed physiotherapy.
AIMS
The aim of this study is to train and test ML models to automatically classify shoulder physio- therapy exercises.

Methods

Healthy volunteers and patients with rotator‐cuff (RC) tears performed six shoulder rehabilitation exercises while wearing a wearable system equipped with three magneto‐inertial sensors. Each exercise was repeated six times following guidelines developed by the American Society of Shoulder and Elbow Therapists. Six supervised ML models (k‐Nearest Neighbours, Support Vector Machine, Decision Tree, Random Forest (RF), Logistic Regression and Adaptive Boosting) were trained for the classification. The algorithms' ability to accurately classify exercise activities was evaluated using the nested cross‐validation method, with different combinations of outer and inner folds.

Results

Nineteen healthy individuals and seventeen patients with RC tears participated in the study. The Random Forest (RF) classifier achieved the highest classification performance, with an accuracy of 89.91% and an F1-score of 89.89%.

Discussion

Wearable sensors and ML algorithms accurately classify shoulder rehabilitation exercises. The study demonstrates the effectiveness of the models in learning the patterns and distinguishing features from the collected signal data also in patients with rotator cuff tears. Further research will focus on integrating additional sensors, such as electromyography sensors. By remotely evaluating patients’ performance, researchers and therapists can gain deeper insights into the effectiveness of rehabilitation programs, facilitating personalized treatment for each individual. This approach will enhance the quality of care for patients with shoulder musculoskeletal conditions, leading to better rehabilitation outcomes.

Acknowledgement: This work has been supported by the Italian Ministry of Health in the framework of RICERCA FINALIZZATA 2021 (RF-2021-12372810).