PREDICTIVE MODELING OF SLEEP DISORDERS VIA SUPERVISED LEARNING TECHNIQUES

Abstract

CT Sleep disorders such as insomnia, sleep apnea, narcolepsy, and REM behavior disorder significantly affect human health, leading to reduced quality of life, cardiovascular complications, stress, and impaired cognitive performance. Traditional diagnostic methods like polysomnography (PSG), although highly accurate, are expensive, time-consuming, and require specialized clinical infrastructure, making early diagnosis difficult for large populations. This project presents a machine learning-based intelligent framework for the automated classification and prediction of sleep disorders using health and lifestyle attributes. The system utilizes the Sleep Health and Lifestyle dataset containing demographic, physiological, and behavioral parameters such as age, gender, sleep duration, sleep quality, physical activity level, stress level, BMI, heart rate, blood pressure, and daily step count. Data preprocessing techniques including missing value handling, categorical encoding, normalization, feature selection, and imbalance management were applied to improve model robustness. Multiple machine learning algorithms including Logistic Regression, Ridge Classifier, Support Vector Machine (SVM), Random Forest, KNearest Neighbors (KNN), and Artificial Neural Networks (ANN) were implemented and evaluated using accuracy, precision, recall, and F1-score metrics. Experimental results indicate that the proposed models achieved strong predictive performance, with Random Forest and SVM demonstrating superior classification capability, achieving approximately 94.74% accuracy in binary sleep disorder prediction. Feature importance analysis identified heart rate, sleep quality, sleep duration, stress level, and physical activity as significant predictors. The proposed system offers a cost-effective, scalable, and efficient solution for early sleep disorder screening and can be extended for wearable health monitoring and realtime healthcare applications.