MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202641097941 A) filed by Sri Eshwar College Of Engineering on August 12, 2026, for Vision Rural Ai - Multi-Modal Ai Diabetic Retinopathy Screening.
Inventors include Dr. S. Sumathi; Devaranjan J; K S Abishek; and Harini S.
The application for the patent was published on August 21, 2026, under issue no. 34/2026.
Abstract: The VisionRural AI diagnostic ecosystem addresses the critical shortage of ophthalmologists in rural India through a proprietary multi-modal AI architecture that combines systemic metabolic biomarkers with high-resolution retinal imaging to achieve 99.9% diagnostic accuracy. Its primary innovation is a Stage 1 predictive algorithm powered by an XGBoost Regressor, which analyses biomarkers such as HbA1c, serum creatinine, and blood pressure to predict vascular risk factors before they manifest as visible retinal abnormalities. The resulting systemic risk score is validated by an advanced Vision Engine built on a Swin Transformer with Slicing Aided Hyper Inference (SAHI), enabling accurate detection of early retinal micro-lesions that are typically lost during conventional image downsampling. A novel Gated Reciprocal Logic framework cross-references systemic and retinal findings, significantly reducing false positives in normoglycaemic patients while improving sensitivity for detecting Stage 4 proliferative diabetic retinopathy. Optimised using 16-bit quantisation, the architecture operates efficiently on offline edge-computing devices, making it well suited for deployment in resource-constrained rural healthcare environments. To ensure clinical transparency, the system integrates an Explainable AI (XAI) framework using Grad-CAM visualisations for retinal lesion localisation and SHAP values to explain the contribution of each clinical biomarker. By combining predictive analytics, advanced computer vision, and interpretable AI, VisionRural AI provides rural healthcare professionals with a reliable, scalable, and accessible screening solution to help prevent diabetes-related blindness.
Disclaimer: Curated by HT Syndication.
