MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641058172 A) filed by Dr. Arulprakash M; C Manikant A Reddy; and P Yogendra Sal on May 07, 2026, for Accelerated Computing For Healthcare Diagnostics Using Machine And Deep Learning.
Inventors include Dr. Arulprakash M; C Manikant A Reddy; and P Yogendra Sal.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: Abstract The rapid escalation of global healthcare data volume, combined with the growing complexity of disease patterns and the persistent shortage of specialist clinical expertise in many healthcare settings, presents a critical challenge to modem medical diagnostics. Traditional diagnostic systems rely on manual interpretation of discrete patient data points without the computational 1 0 capacity to identify complex, multi-variable patterns across large patient populations. This project presents the Healthcare Diagnostics Platform, a comprehensive AI-powered clinical l-------aecision support system tfiat integrates macfiine learning ana-deep learning tecfinologies to deliver accurate, real-time disease prediction and medical image analysis at scale. The platform collects patient health records from electronic health record systems and loT-enabled medical -Q) C) Ill D.. Q) -1-- N E... 0 -LL. N .".'"..'". CIO Lt) .0.. .. -::1' (0 N 0 -N CIO 1.1') ..... CIO -(0 (0 N 0 N I - Ill ~) ::!: " I "0'" '" 15 monitoring devices, enriches data with contextual imaging inputs from X-ray, MRI, and CT scanners, and subjects all data to a rigorous preprocessing pipeline comprising data cleaning, normalization, feature engineering, and class balancing. The image analysis core employs a transfer"leaming-enhanced Convolutional Neural Network built on a ResNet-50 backbone, achieving a classification accuracy of 94.1% on medical images for tumor detection and 20 anomaly identification across three imaging modalities. The disease prediction engine uses a tuned Random Forest ensemble trained on structured EHR data, delivering 91-93% accuracy across breast cancer, diabetes, and heart disease prediction tasks. A risk stratification module classifies each patient into low, medium, or high risk tiers with real-time alert generation for critical cases. An interactive web-based dashboard built with React and D3.js presents real- 25 time predictive insights, patient health trend visualizations, comparative model performance metrics, and Grad-CAM explainability heatmaps for imaging predictions through an intuitive, responsive interface. An AI-powered chatbot provides clinicians with conversational, patientspecific treatment recommendations based on current prediction outputs.
Disclaimer: Curated by HT Syndication.
