| Abstrakt: |
A considerable number of inpatients, including patients who have undergone spinal surgery, live in bed for a certain period of time, which exposes them to various complications caused by muscle weakness. To mitigate this risk, medical staff often educate patients about the importance of exercise and posture changes. However, it is challenging to objectively determine whether patients are performing exercises correctly. In this work, we design and implement a deep learning model that recognizes bedside lower extremity exercises and postures using a single digital biomarker. Data from 20 healthy adults were collected to evaluate the performance and utility of the model in various environments. The model achieved a recognition accuracy of 95.42% for 8 postures and 14 exercises. The findings of this study can contribute to faster recovery, improved exercise effects, and prevention of complications by assisting medical staff in monitoring the rehabilitation exercises and postures of bedside patients. [ABSTRACT FROM AUTHOR] |
