MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641084451 A) filed by Vallurupalli Nageswara Rao Vignana Jyothi Institute Of Engineering And Technology on July 09, 2026, for Explainable Multi-Modal Ai System For Pulmonary Disease Detection And Clinical Decision Support.
Inventors include Mrs. Soujanya Ambala; D. Sai Kiran; S. Sai Vineeth; D. Aravind Kumar; and P. Charan Teja.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: ABSTRACT The present invention introduces a novel Multi-Modal Interpretable Lung Infection Diagnosis System (MILIDS) designed for the early and accurate detection of lung infections such as pneumonia and tuberculosis by intelligently integrating diverse medical data sources including chest X-ray and CT images, unstructured clinical notes, and structured laboratory parameters. The proposed system employs specialized neural architectures—Convolutional Neural Networks for visual feature extraction, transformer- based models like BERT for textual understanding, and Multi-Layer Perceptrons for numerical clinical data—fused through an innovative attention-based mechanism that dynamically weighs the contribution of each modality. To address the critical need for clinical trust, the invention incorporates advanced explainable AI techniques including Grad-CAM for highlighting infection regions in images and SHAP for quantifying feature importance across all inputs. Developed as a complete end-to-end framework with a user- friendly interface, the system processes multi-modal inputs in real-time, classifies cases into Normal, Pneumonia, or Tuberculosis categories, and delivers transparent, actionable insights. Trained and validated on comprehensive datasets with robust preprocessing and augmentation strategies, MILIDS achieves superior diagnostic performance with accuracy exceeding 95%, reduced processing time, and enhanced generalization, offering a transformative tool for supporting physicians in emergency, rural, and telemedicine settings while minimizing diagnostic delays and errors.
Disclaimer: Curated by HT Syndication.
