MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641076117 A) filed by Dayananda Sagar College Of Engineering on June 19, 2026, for Ai-Driven Early Cancer Detection Using Microbial Metabolic Fingerprints.
Inventors include Ishwari Suyog Bhawalkar; Aprameya Bharadwaj; C V Balaji; Anjana U; K Parjanya Ram; Prof. Poornima D; and Prof. Leelavathi R.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: The present invention discloses an AI-Driven Early Cancer Detection System Using Microbial Metabolic Fingerprints, a novel multi-cancer diagnostic platform that detects tumor-induced alterations in microbial metabolite production profiles from non-invasive biological samples (saliva, stool) and applies a graph-attention neural network to generate simultaneous early-stage cancer risk scores for six distinct malignancy types. The inventive core sets the present system apart from all prior art: cancer detection is not achieved by detection of tumour derived biomarkers (e.g., circulating tumour DNA, CA-125, CEA) or enumeration of microbial species, but by decoding the unique metabolic fingerprint, the collective pattern of microbially synthesised small molecules that occurs when the tumour microenvironment reprograms surrounding and distal microbial communities to altered biosynthetic phenotypes.The system comprises: (i) a Multi-Metabolite Biosensor Chip (MFP-CHIP) for detection of polyamines, hydrogen sulphide species, nitrosamines, short-chain fatty acids, secondary bile acids, and indole derivatives; (ii) a Metabolic Fingerprint Extraction Engine encoding 256-dimensional patient fingerprints from temporal metabolite profiles; (iii) a Graph-Attention Multi-Cancer AI Model with six simultaneous cancer-specific output heads; and (iv) a Microbial Metabolic Fingerprint Database (MFP-DB) of 15,000+ annotated clinical profiles. The system predicts risk for early stage (Stage I–II) cancer in colorectal, pancreatic, oral/head-neck, gastric, hepatic and gut-axis breast cancers with AUC values of 0.85–0.96. This combination has not been disclosed in any prior patent: tumor-induced microbial metabolic fingerprinting as the primary input signal, a graph-attention network architecture to encode metabolite pathway relationships, multi-cancer simultaneous prediction from a single non-invasive sample, and a federated metabolic reference database. The invention provides a truly new diagnostic modality.
Disclaimer: Curated by HT Syndication.
