Exploring the online social networks for sheet-type piezoelectric sensors-based healthcare analysis

This study introduces a novel health-monitoring model that integrates physiological sensor data with outputs from online social networks (OSNs) to enhance the accuracy, contextual awareness, and early detection of abnormal health trends. Experimental results demonstrate that fused multimodal features significantly outperform the sensor-only and OSN-only approaches. Correlation analysis revealed strong associations between heart-rate responsiveness and stress-related posts (r = 0.63), respiration irregularity and sleep-related discussions (r = 0.57), and motion frequency and fatigue mention (r = 0.46). The predictive performance improved markedly with multimodal fusion, where a random forest (RF) classifier achieved 92.4% accuracy and an F1-score of 0.91. Validation using cross-validation, temporal analysis and stress testing confirmed a high system stability (0.94), reduced error rates (3.7%) and faster anomaly detection with an average response time of 12.4 min. A comparative evaluation further showed superior detection accuracy (93.5%), contextual accuracy (93.8%), and responsiveness (11.2 min) for the proposed framework. Usability testing reports a high System Usability Scale (SUS) score of 87.5, with perceived usefulness and user satisfaction of 94% and 92%, respectively. Deployment experiments confirmed reliable operation, achieving 99.4% a system uptime of 97.2% data synchronization efficiency. Overall, the results establish the integrated sensor–OSN fusion model as a robust, accurate, and user-friendly solution for real-time context-aware health monitoring systems.