MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641077252 A) filed by Gnaneswari Gnanaguru; and Cmr Institute Of Technology on June 23, 2026, for Smart Headgear-Based Personalized Seizure Prediction System Using Dynamic Preictal Window Learning With Uncertainty-Aware Clinical Decision Support.
Inventor includes Gnaneswari Gnanaguru.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: The present invention relates to a wearable intelligent healthcare system for the early prediction of epileptic seizures and, more particularly, to a smart headgear-based seizure forecasting platform that combines electroencephalogram (EEG) monitoring, multimodal physiological sensing, advanced artificial intelligence, and explainable decision support. The system comprises a wearable EEG headgear equipped with dry or semi-dry electrodes for continuous acquisition of brain activity signals, a signal acquisition and preprocessing module, an embedded processing unit, wireless communication circuitry, and a mobile application interface. The acquired EEG signals, together with optional physiological parameters such as heart rate, electrodermal activity, motion data, sleep information, and other contextual factors, are integrated into a multimodal data framework for comprehensive neurological assessment. The invention employs a dynamic preictal window learning mechanism configured to automatically identify and continuously adapt patient-specific seizure prediction horizons based on historical seizure records, temporal EEG patterns, and evolving physiological conditions. A multimodal artificial intelligence engine utilizing a Temporal Fusion Transformer (TFT) architecture processes the integrated data to capture complex temporal dependencies and generate personalized multi-horizon seizure likelihood predictions. The system further incorporates uncertainty-aware prediction mechanisms to estimate confidence levels associated with each forecast, thereby improving the reliability of clinical decision-making and reducing false alarms. To enhance transparency and clinical acceptance, the invention includes an explainability module based on SHapley Additive exPlanations (SHAP), which identifies influential EEG channels, frequency bands, physiological parameters, and temporal features contributing to the prediction outcome. The generated explanations are presented in a clinician-readable format to facilitate interpretation and validation of the predictive results. When the predicted seizure probability exceeds an adaptive threshold, the system automatically transmits alerts, risk scores, confidence measures, and explanatory information to a connected mobile device, caregiver platform, or healthcare provider interface, enabling timely preventive intervention and improved patient safety. The invention thereby provides a real-time, personalized, interpretable, and uncertainty-aware seizure prediction and early warning system for enhanced epilepsy management and clinical support.
Disclaimer: Curated by HT Syndication.
