<![CDATA[Sleep disorders are common and can significantly harm human health, with insomnia and sleep apnea being the most prevalent conditions. These disorders are often difficult to detect and treat accurately. Although machine learning (ML) techniques have shown promise in improving diagnostic precision and personalized treatment, most existing studies rely on single source data or conventional ML models, which limit their robustness and generalizability across diverse populations. To address this research gap, this study integrates multi-modal data and ensemble learning techniques to enhance accuracy, interpretability, and real-time applicability in diagnosing insomnia and sleep apnea. A dataset of 400 samples was collected through manual methods and internet of things (IoT) devices from multiple sources. Statistical techniques were applied for data cleaning, followed by principal component analysis (PCA) to reduce dimensionality and improve training efficiency. Four base ML models: decision tree (DT), support vector machine (SVM), naive Bayes (NB), and random forest (RF) were initially trained and evaluated. Subsequently, a boosting-based ensemble model was implemented to further improve performance. The proposed gradient boosting model with RF as the base learner achieved the highest diagnostic accuracy of 96.01%. The results demonstrate that ensemble ML models combined with multi-modal data significantly enhance the accuracy of insomnia and sleep apnea diagnosis.]]>
Copyrights © 2026
