MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202621055460 A) filed by Pravin Badhe on April 30, 2026, for Ai-Enabled System For Llm-Based Discovery, Structured Data Generation, Multi-Level Predictive Scoring, And Computational Formulation Design.
Inventors include Dr Pravin Badhe; Mrs Ashwini Badhe; Dr. Sanjana A; Mr. Ramdas Bhat; Dr. Keserla Bhavani; and Rekha Umakanth.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: The present invention relates to an AI-enabled computational system for large language model (LLM) based discovery, structured biomedical data generation, multi-level predictive scoring, and computational formulation design. The system is configured to generate domain-specific scientific prompts from inputs including plants, phytochemicals, diseases, biological targets, or mechanisms, and to execute the prompts through one or more large language models to obtain unstructured biomedical text outputs. The unstructured outputs are transformed into structured numerical datasets by deterministic extraction of molecular, mechanistic, toxicity, pharmacological, and formulation-related features. The invention further comprises a chained three-level predictive scoring pipeline including a Compound or Formulation Fitness Score (CFFS), a Preclinical Therapeutic Index (PTI), and a Digital Response Index (SDRI), wherein the output of each level is used as input to the subsequent level to generate a unified computational prediction. A predictive formulation engine utilizes the chained scores to identify candidate formulation components, compute predicted compatibility or synergy, generate formulation ratios, and associate predicted delivery systems. The system integrates public-domain formulation datasets derived from pharmacopoeias and traditional medicine monographs with LLM-derived novel formulation designs to generate computationally predicted formulation candidates. An embodiment of the invention is demonstrated using chloride intracellular channel protein-1 (CLIC1) as a biological target associated with oral squamous cell carcinoma, wherein epicatechin and optimized derivatives are evaluated to generate predicted formulation compositions and delivery architectures. All outputs of the system are computational predictions intended for research, analysis, and formulation design purposes.
Disclaimer: Curated by HT Syndication.
