MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641078571 A) filed by Sr University on June 25, 2026, for A Clinically Safe Llm Framework Integrating Dual-Rag Grounding, Claim- Level Medical Verification, & A Trust Scoring Pipeline With Physician-In-The-Loop Decision Gating.
Inventors include Dr. Kalaivani K; and Dr. Rajchandar K.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: A CLINICALLY SAFE LLM FRAMEWORK INTEGRATING DUAL-RAG GROUNDING, CLAIM-LEVEL MEDICAL VERIFICATION, & A TRUST SCORING PIPELINE WITH PHYSICIAN-IN-THE-LOOP DECISION GATING The present invention relates to a clinically safe large language model framework for clinical decision support. The framework comprises a clinical context builder module, a dual-retrieval augmented generation (Dual-RAG) grounding engine, a large language model generation module, a medical validation pipeline, a trust scoring module, and a physician-in-the-loop decision gating module. The system receives clinician queries and patient-specific electronic health record data, retrieves validated evidence from clinical guideline repositories and biomedical literature sources, and generates evidence-constrained clinical recommendations. A medical validation pipeline decomposes generated responses into atomic clinical claims and independently verifies each claim through evidence matching, contradiction detection, guideline alignment, and patient-specific compatibility assessment. Trust scores are assigned based on evidence quality and clinical consistency. Responses failing predefined safety thresholds are automatically flagged for physician review. The framework provides explainable outputs with claim-level traceability, citations, and confidence information, thereby reducing hallucinated recommendations and improving safety, reliability, accountability, and regulatory compliance in clinical decision support systems.
Disclaimer: Curated by HT Syndication.
