A Hardware-Secured Wearable Device for Direct Hydration and Vital Monitoring Using Multimodal Sensor Fusion

Abstract

Dehydration significantly affects physiological stability, cognitive performance, and overall health. Conventional hydration assess-ment methods are invasive and unsuitable for continuous monitoring. This work proposes a hardware-secured wearable system for real-time hydration and vital monitoring using multimodal sensor fusion and AI-driven analysis. The system integrates bio-impedance analysis (BIA), photoplethysmography (PPG), heart rate, and temperature sensors with an ESP32 microcontroller. A multi-stage inference pipeline performs signal quality assessment, trust estimation, contradiction detection, bio-impedance correc-tion, and temporal hydration prediction using CNN and BiLSTM-based architectures. The framework also incorporates determin-istic risk scoring, uncertainty calibration, and graceful degradation during sensor failure. Experimental evaluation demonstrates reliable hydration trend estimation and stable monitoring under noisy and degraded conditions. The proposed system provides real-time hydration risk assessment and continuous physiological monitoring suitable for healthcare, sports, and wellness applications.