MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202641025332 A) filed by Dharun Chandru, Coimbatore, Tamil Nadu, on March 3, for 'explainable ai framework for leukemia prediction and recommendation system.'
Inventor(s) include Yoga Priya G; Danu Mithra S; Abinaya N; and Inika H.
The application for the patent was published on March 13, under issue no. 11/2026.
According to the abstract released by the Intellectual Property India: "Leukemia is a severe blood-related cancer that demands timely and precise diagnosis to enhance treatment effectiveness and increase patient survival rates. With the rapid growth of healthcare data and continuous progress in artificial intelligence, machine learning and deep learning methods have become powerful approaches for disease prediction and clinical decision support systems. Nevertheless, real-world medical datasets frequently contain bias, imbalanced class distributions, and high-dimensional heterogeneous features, which create difficulties in achieving dependable predictions. In this study, a Leukemia Prediction and Clinical Recommendation System is designed using a large-scale healthcare dataset consisting of 143,194 patient records gathered from 22 countries. The dataset combines demographic, clinical, genetic, and socioeconomic information to represent diverse real-world patient populations. Extensive data preprocessing steps, such as handling missing values, encoding categorical variables, normalization, and feature selection, are implemented to enhance overall data quality. Fifteen clinically relevant features are ultimately chosen to reduce dimensional complexity while maintaining strong predictive performance. Three predictive approaches-Decision Tree, Random Forest, and Deep Neural Network (DNN)-are developed and assessed for detecting leukemia. A comparative evaluation shows that although all three models can identify leukemia status effectively, the DNN demonstrates the highest accuracy because of its strength in capturing complex non-linear patterns within high-dimensional healthcare data. Therefore, the DNN model is chosen for real-time implementation. The trained DNN is incorporated into a Flask-based web application that allows users to enter patient information and receive immediate prediction results along with confidence scores. Based on the predicted outcome, the system generates outcome-oriented clinical recommendations, including medical advice, recommended diagnostic tests, suggestions for consulting specialists, and personalized dietary guidance. The proposed framework serves as an intelligent clinical decision support system that facilitates early risk detection and informed medical decisions, thereby enhancing healthcare analytic and AI-supported diagnosis."
Disclaimer: Curated by HT Syndication.
