MUMBAI, India, July 30 -- Intellectual Property India has published a patent application (202641087503 A) filed by Dr. R. Gunasekaran; and Dr. R. Kathiroli on July 17, 2026, for Ai-Driven Multimodal Clinical Decision Support System For Early Stroke Risk Prediction And Method Thereof.
Inventors include Dr. Kalimuthu Karuppanan; and Aakash L.
The application for the patent was published on July 24, 2026, under issue no. 30/2026.
Abstract: ABSTRACT: Title: AI-Driven Multimodal Clinical Decision Support System for Early Stroke Risk Prediction and Method thereof The present invention is a generative AI-driven multimodal clinical decision support system (100) and method (200) for early stroke risk prediction in patients with atrial fibrillation (AFib). The system integrates 12-lead ECG signals and structured clinical tabular data through six integrated modules: a data preprocessing pipeline; a Temporal-Channel Attention Fusion (TCAF) encoder employing multi-scale depthwise-separable convolutional tokenization and cross-lead attention to generate a 128-dimensional ECG embedding; a Feature Tokenizer Transformer (FT-Transformer) that encodes structured clinical variables including CHA2DS2-VASc-derived risk strata and AFib-comorbidity interaction terms into a 128-dimensional clinical embedding; a Bidirectional Cross-Modal Attention Fusion (BCAF) module that enables dynamic bidirectional information exchange between modalities to produce a 256-dimensional unified representation; an Explainable AI (XAI) module providing SHAP-based feature attributions, integrated gradient ECG attribution maps, and attention heatmaps; and a Generative Reasoning Module (GRM) powered by a Large Language Model with Retrieval-Augmented Generation that produces clinically interpretable, guideline-referenced natural-language explanations. Main Illustrative:
Disclaimer: Curated by HT Syndication.
