MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641085565 A) filed by P Venkata Krishna on July 13, 2026, for Computer-Implemented System And Method For Robust Diagnosis Of Autoimmune Skin Diseases Using Domain-Adapted Deep Convolutional Neural Network With Multi-Modal Fusion, Uncertainty Estimation And Conditioned Clinical Decision Support.
Inventors include T. Jahnavi; and J. Katyayani.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: A computer-implemented system and method for robust diagnosis of autoimmune skin diseases (Lichen Planus, Dermatomyositis, Lupus, Pemphigus, Vasculitis) from consumer device images is disclosed. The method performs domain-adapted preprocessing to correct smartphone artifacts and normalize for Indian skin phototypes, extracts features via a modified ResNet-50 backbone incorporating CBAM attention modules, optionally fuses multi-modal clinical metadata, generates disease classification with confidence and uncertainty estimates (via Monte Carlo dropout or evidential layers), produces Grad-CAM explainability maps, and delivers conditioned medication and recurrence-prevention suggestions. The system is optimized for on-device inference on resource- constrained mobile devices and includes structured text-to-speech output for accessibility. Technical effects include improved accuracy and robustness on real-world diverse images, enhanced clinical utility through uncertainty-aware and explainable outputs, and practical deployment in Indian primary care and telemedicine settings.
Disclaimer: Curated by HT Syndication.
