MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641085580 A) filed by Vallurupalli Nageswara Rao Vignana Jyothi Institute Of Engineering And Technology on July 13, 2026, for System And Method For Real-Time Dehydration Detection Using Wearable Physiological Sensors, Rule- Based Label Generation, And Ensemble Machine Learning Models.
Inventors include Soujanya Ambala; A. Varshitha; B. Mahathi; G. Sireesha; and L. Greeshma.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: ABSTRACT [0023] The present invention discloses a Machine Learning-Based Dehydration Detection System (ML-DDS) that leverages physiological sensor data from wearable devices to enable early, accurate, and automated identification of dehydration states. The system processes key signals including Electrodermal Activity (EDA), Blood Volume Pulse (BVP), and Acceleration Magnitude (ACC_mag) through a robust preprocessing pipeline and employs a comparative ensemble of Decision Tree, Random Forest, and XGBoost classifiers. A novel rule-based labeling mechanism derives the target variable from heart rate ( 100 bpm) and body temperature ( 34.0°C) thresholds, transforming unlabeled sensor streams into supervised training data. Among the evaluated models, XGBoost demonstrates superior performance with approximately 95.39% test accuracy, effectively handling class imbalance via weighted learning and scaling techniques. The invention further integrates a Flask-based web application that accepts real-time sensor inputs and delivers instant predictions with confidence scores, temporarily deployed via Ngrok for accessible testing. By combining domain-specific physiological feature engineering, rule-augmented labeling, and optimized ensemble modeling with an intuitive user interface, this system provides a scalable, non-invasive solution for continuous hydration monitoring in healthcare, sports, and occupational settings, significantly advancing proactive physiological health management.
Disclaimer: Curated by HT Syndication.
