SLEEP DISORDER PATTERN DETECTION USING SMARTPHONE USAGE

Sleep disorders are increasingly prevalent worldwide and significantly impact physical health, cognitive performance, and overall quality of life. Traditional diagnostic techniques such as polysomnography and clinical observation provide reliable results but require specialized equipment, medical supervision, and high operational costs, which limits accessibility for large populations. With the rapid growth of smartphone usage and the availability of digital behavioral data, smartphones have emerged as a promising platform for passive health monitoring and behavioral analysis. This study proposes a Sleep Disorder Pattern Detection System that leverages smartphone activity data to identify potential sleep disorders using machine learning techniques. The system collects behavioral indicators such as screen on/off events, application usage patterns, device charging habits, and session durations to estimate sleep onset time, wake time, sleep duration, and nighttime interruptions. Feature engineering techniques are applied to transform raw activity logs into meaningful behavioral metrics, which are then processed using a Random Forest classification model. The proposed model categorizes users into three sleep pattern groups: Normal Sleep, Insomnia, and Delayed Sleep Phase Syndrome (DSPS). The system architecture follows a three-tier design consisting of a mobile application developed using React Native, a Node.js and Express-based REST API backend for data management and authentication, and a Python Flask machine learning service responsible for feature extraction and prediction. PostgreSQL is used as the database for storing user activity logs, prediction results, and recommendations. The results demonstrate that smartphone behavioral patterns can provide valuable insights into sleep habits and may support early detection of sleep disorders. This approach offers a scalable, noninvasive, and cost-effective solution for large-scale sleep health monitoring.