Introduction: Sleep health is important for overall well-being, as disruptions or inadequate sleep can increase the risk of sleep disorders. Traditional detection methods relying on biological signals face practical challenges in data collection and processing. This study explores how data from personal devices and routine health assessments can be leveraged with Machine Learning (ML) algorithms to detect sleep disorders. Methodology: Four ML models – Extreme Gradient Boosting (XGB), Logistic Regression (LR), Decision Tree (DT), and Random Forest (RF) – were tested on a dataset from 374 participants. The dataset includes participant characteristics, sleep parameters, lifestyle factors, and physiological measurements. Data subsets were based on correlation and importance scores, and models were evaluated on accuracy and F1-scores before and after tuning. Results: XGB consistently outperformed other models, achieving an accuracy of 96.49% and F1-score of 97.83% before tuning, and improved to 98.25% and 98.90% after tuning. Feature importance analysis revealed BMI as the most critical feature across models, with LR, DT, and RF ranking it consistently the highest. Discussion: XGB demonstrated high efficacy in detecting sleep disorders using leveraging non-invasive and readily available data. Differences in feature importance across models indicate multiple factors contribute to effective prediction. While self-reported sleep disorder data may introduce biases, integrating objective measures could further validate the approach. Conclusion: ML models, particularly XGB, show strong potential in detecting sleep disorders using accessible data sources. This approach offers a practical, non-invasive alternative for early diagnosis and intervention in sleep health management.