Adaptive Edge-Federated AI Framework for Contactless Menstrual Health Prediction Using Multimodal Physiological...

This study introduces an adaptive edge-federated artificial intelligence (AI) framework that employs contactless biosensing and multimodal physiological intelligence to enable accurate, real-time, and privacy-preserving menstrual health prediction. Traditional menstrual cycle tracking methods, such as manual logs or wearable sensors, often rely on skin contact, are prone to measurement errors, and raise privacy concerns due to centralized data storage. To overcome these limitations, the proposed framework integrates radar-based respiration sensing, photoplethysmography (PPG), and LiDAR-assisted microvascular mapping to non-invasively monitor vital physiological signals including heart rate, vascular activity, and breathing rhythm. The multimodal biosignals are analyzed locally through an adaptive edge-learning engine that continuously updates and personalizes predictions without transferring raw data to central servers. A federated optimization mechanism enables secure, decentralized model training, ensuring that sensitive reproductive health data remains on user devices while still benefiting from global model improvements. By dynamically adjusting to individual physiological variations and irregular cycle patterns, the system enhances prediction accuracy, reduces latency, and preserves user privacy. This intelligent, non-invasive framework offers a next-generation approach to menstrual health management, combining edge intelligence, federated learning, and multimodal biosensing to deliver a discreet, adaptive, and trustworthy reproductive health solution.•Contactless biosensing using radar, PPG, and LiDAR for real-time menstrual health tracking.•Edge-federated learning enables decentralized, privacy-preserving AI model training.•Adaptive multimodal AI intelligence enhances prediction accuracy for irregular cycles.