Human Everyday Experience Metric Recognition Based on Classifier Chains Using Heart Rate

This study examines the potential of heart rate data from wearable devices to assess sleep quality, emotional responses, and stress levels, among other sleep-related factors. There is a growing need for a nuanced understanding of emotional and physical health, but conventional assessment methods often lack real-time analysis and objectivity. Utilizing data from smartwatches, this study employs a multifaceted analytical technique that integrates feature extraction, feature selection, and machine learning algorithms to explore the interconnectedness of various health metrics. The examination of the relationships among multiple labels has shown that many features, except for sleep efficiency (S2) and wake after sleep onset (S4), can serve as statistically significant biomarkers. The findings illustrate the efficacy of classifier chains, enhanced by effective feature selection, in uncovering hierarchical relationships among seven indicators, indicating that it achieved a high macro F1-score in binary classification. These outcomes signify a significant advancement in predictive accuracy and contribute to the evolution of health monitoring by providing reliable, data-driven insights to support personalized interventions. This study bridges the divide between clinical assessments and daily monitoring, allowing future research to refine methodologies and develop enhanced health and wellness solutions.