Abstract
Sleep disorders like sleep apnea and insomnia significantly impair athletes' recovery and performance. Sleep apnea, exacerbated in supine positions due to increased airway resistance, and insomnia, evidenced by fragmented sleep and restlessness, highlight the necessity of monitoring sleep postures. This study introduces a novel RFID-embedded smart mattress capable of non-invasive monitoring and detection of these disorders by capturing body postures and movements using passive RFID sensors. A multi-layered mattress design integrates advanced RFID technology with machine learning algorithms-Gaussian process regression (GPR) and linear regression (LR)-to classify postures and detect movement anomalies. Evaluated with data from five participants in supine and prone positions, the system achieved a posture recognition root mean square error (RMSE) of 0.42 and movement detection RMSE of 0.15. Data processing included standardization and Gaussian filtering for enhanced accuracy, with a 5-fold cross-validation framework ensuring robust performance. The results demonstrate the mattress's effectiveness as a cost-efficient, non-intrusive alternative to traditional polysomnography, offering insights for early detection and management of sleep disorders. This approach shows significant potential for sports psychology applications, enabling personalized recovery strategies and performance optimization. Future work will focus on expanding the dataset, integrating additional biometric sensors, and refining algorithms to improve diagnostic accuracy and real-time usability in clinical and home settings.