Augmenting sleep quality prognostics through internet of things and machine learning: a rigorous comparative analysis for advanced personalized health metrics

Abstract

This study investigates the potential of integrating Internet of Things (IoT) devices with advanced machine learning models to enhance sleep quality assessment. Leveraging a dataset of 826 individuals collected through wearable IoT devices, we perform a comparative analysis across four machine learning algorithms: Logistic Regression, Support Vector Machine (SVM), Artificial Neural Network (ANN), and Random Forest. Our methodology includes comprehensive data preprocessing, feature engineering with a novel Sleep Quality Score (SQS) formulation, and extensive model evaluation through cross-validation. Experimental results reveal that the Random Forest model achieves the highest accuracy at 99%, followed closely by SVM at 98% and ANN at 96%. Key factors impacting sleep quality, such as sleep duration, efficiency, and various sleep stage ratios, are identified and analyzed. Our findings contribute significantly to the field of IoT-based health monitoring, offering insights for developing tailored interventions and improving sleep management. This research lays the groundwork for future advancements in smart health applications and personalized healthcare solutions.