Design and Development of Low-Cost Wearable Dehydration Risk Monitoring System Based on Trend-Based Analysis and Random Forest Algorithm

Abstract

Dehydration is a common health issue for elderly people and patients in critical condition. It is often recognized only after noticeable symptoms appear. Traditional methods for assessing dehydration include blood tests and manual observation by clinicians. These methods are invasive, time-consuming, and not suitable for ongoing monitoring. This paper presents a wearable, non-invasive system for monitoring dehydration risk using bioimpedance, skin temperature, and heart rate variability (HRV). Wearable sensors continuously measure physiological factors related to body hydration. A microcontroller-based system processes the collected data for real-time monitoring. Instead of just using fixed threshold values, the system tracks trends in physiological signals to identify early changes linked to dehydration. Based on the analysis of sensor data, a Dehydration Risk Index is created, classifying hydration status as Normal, Mild Risk, or High Risk. The proposed system is affordable, portable, and fits well for continuous monitoring in hospitals and home care. This system can assist healthcare professionals in detecting dehydration early, allowing for timely intervention and improving patient safety and clinical outcomes.