Improving Sleep Disorder Diagnosis Through Optimized Machine Learning Approaches

Abstract

Sleep disorders, particularly insomnia and sleep apnea, have become increasingly prevalent and pose significant challenges to physical health, cognitive function, and overall quality of life. Conventional diagnostic techniques such as polysomnography provide accurate assessments but are often expensive, time-consuming, and inaccessible for large-scale screening. This study proposes an optimized machine learning framework for the early diagnosis of sleep disorders using the Sleep Health and Lifestyle dataset. The proposed methodology incorporates data cleaning, blood pressure decomposition, categorical encoding, feature scaling, and stratified train-test partitioning to enhance data quality and predictive performance. Three supervised machine learning algorithms, namely Random Forest, Decision Tree, and Support Vector Machine (SVM), are implemented and comparatively evaluated for multiclass classification of individuals into insomnia, sleep apnea, and no sleep disorder categories. Model performance is assessed using accuracy, precision, recall, F1-score, confusion matrix analysis, and five-fold cross-validation. Experimental results demonstrate that the Random Forest classifier achieves the highest classification accuracy of 94.67%, outperforming the Decision Tree and SVM models while exhibiting superior robustness and generalization capability. The proposed framework provides an efficient and interpretable decision-support solution for early sleep disorder identification and has the potential to assist healthcare professionals in timely diagnosis and personalized treatment planning.