MUMBAI, India, July 30 -- Intellectual Property India has published a patent application (202641080976 A) filed by Sathya T on July 01, 2026, for Docjarvis - Agentic Medical Assistant.
Inventors include Sathya T; Brundha R; Dakshinya L; Dharshanaa S R; and Meenaumadevi M.
The application for the patent was published on July 24, 2026, under issue no. 30/2026.
Abstract: ABSTRACT The present invention relates to an automated system and method for an intelligent multi-modal - Medical Assistant with the integration of Convolutional Neural Networks (CNNs). MLOps techniques, and retrieval augmented generation (RAG). The system involves the processing of heterogeneous medical datasets such as MRI scans, CT scans. X-rays, and dennatological images for detecting and classifying diseases including brain tumors, kidney tumors, pneumonia, tuberculosis, and other types of skin lesions. For disease detection and classification, a custom CNN is designed to learn hierarchically structured features, along with Grad-CAM based visualizations to provide interpretability through highlighting of clinically-relevant regions of medical images. Further, the framework uses a conversational module which includes a large language model with the capability to retrieve information from real-time sources, allowing contextual and evidencebased response generation. The present invention also involves voice-based interactions and handling of patient data in a secured manner. The backend of the system architecture is designed for scalability and real-time operations; confidence-based validation strategies are also used to detect the uncertain predictions. Evaluation on the publicly available multi- modal datasets indicates that the system achieves an accuracy of 94.3%, a precision of 93.8%, a recall of 94.1%, and an Fl-score of 93.9%.
Disclaimer: Curated by HT Syndication.
