An Empirical Machine Learning Approach Towards Effective Sleep Disorder Prediction

Sleep is a biological necessity for the healing of the body and the mind. Reduced physical activity and consciousness are its defining characteristics. The duration, timing, and quality of sleep can all be negatively impacted by sleep issues. They may thereby raise the chance of type 2 diabetes, heart disease, strokes, obesity, and cognitive diseases including dementia. Machine learning (ML) offers a more effective, scalable, and impartial way to understand complicated sleep data than manual assessments. This study used a data set of 374 entries gathered from Kaggle to predict sleep issues with an emphasis on sleep apnea and insomnia. Pre-processing steps included feature extraction, feature splitting, and categorical data management. In addition to five classification models (Random Forest, Support Vector Classifier (SVC), Logistic Regression, K-Nearest Neighbour (KNN), and Decision Tree), we also employed three boosting models (AdaBoost, CatBoost, and Gradient Boosting). Random Forest, Gradient Boosting, and CatBoost achieved 95% accuracy in predicting sleep issues, while the other methods achieved 91%. We believe that our suggested framework can be served as a decision support system (DSS) for early sleep problem diagnosis, with the potential to make a substantial contribution to the healthcare industry.