Artificial Intelligence in Sleep Medicine: Predicting Sleep Quality Using Lifestyle and Demographic Indicators
Artur Yamanov
09/02/2026
Sleep disorders are increasingly common and are associated with impaired bodily functioning and quality of life. Early identification of individuals at risk is therefore essential but challenging as it depends on complex interactions between demographic and lifestyle factors. This study investigates whether machine learning (ML) methods can predict sleep quality from demographic indicators. Using a publicly available Sleep Health and Lifestyle dataset, we analysed demographic variables (age) and lifestyle factors (sleep duration, self-reported sleep quality, physical activity level) alongside labels indicating the presence of no disorder, insomnia or sleep apnea. We implemented and compared several regression-based machine learning models. Linear Regression, Decision Tree Regressor, Random Forest Regressor, K-Neighbors Regressor and an MLP Regressor neural network to estimate a continuous sleep quality score. To enhance accuracy, we further designed a weighted system inspired by the Multiplicative Weight Update (MWU) framework, allowing models with lower prediction error to contribute more strongly to the final output. Model performance was evaluated using Mean Squared Error (MSE), Mean Absolute Error (MAE).The study demonstrates the feasibility of using combined lifestyle and demographic metrics with machine learning techniques to estimate sleep quality, with the decision tree regressor showing great success. Overall, models showed success in predicting possible sleep quality according to the weighted systems. This work provides evidence that AI can be used as a tool for more personalised screening for insomnia and sleep apnea in everyday clinical situations.