MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641077932 A) filed by Ms. Benasir Begam. F; and Dr. A. Packialatha on June 24, 2026, for Enhancing Prenatal Diagnostic Accuracy Through Deep Learning Models For Fetal Chromosome Analysis.
Inventors include Ms. Benasir Begam. F; and Dr. A. Packialatha.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: The present invention provides a distributed medical data integration federated deep learning-based framework for improving the accuracy of prenatal diagnosis by advanced fetal chromosome analysis with the patient privacy preserved in multiple medical institutions. It uses a comprehensive data preprocessing pipeline with data cleaning, data standardization, KNN-based data imputation, and normalization to maintain high-quality data inputs, as well as a Conditional Latent Diffusion Model (CLDM) to balance and synthesize data. The methods of feature selection are based on CLDM combined with statistical methods to find significant biomarkers, including ANOVA, Chi-square test, and Pearson correlation test. The core federated deep learning model processes a variety of key indicators that can affect fetal health, such as crown-rump length (CRL), nuchal translucency (NT), fetal heart rate (FHR), and nasal bone detection, and hyperparameter optimization is implemented using the Red-Legged Tortoise Optimization Algorithm (RLTOA) to enhance model performance. The framework also embraces the interpretability aspect by using CO-LIME explanations to make the decisions transparent and transparent, and includes model quantization and compression for efficient deployment. The final predictive module is a risk assessment for chromosomal abnormalities, including Down syndrome, which provides accurate, scalable and privacy-preserving decision support for clinicians in prenatal care.
Disclaimer: Curated by HT Syndication.
