MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202641085095 A) filed by Sasi Institute Of Technology And Engineering; G. Prasanth Kumar; P. V. V. S Eswara Rao; Lakshmana Rao Talapakula; and Pcs Nagendra Setty on July 11, 2026, for Artificial Intelligence-Based System And Method For Early Diabetes Prediction Using Hybrid Feature Selection And Attention-Guided Siamese Neural Networks.
Inventors include G. Prasanth Kumar; P. V. V. S Eswara Rao; Lakshmana Rao Talapakula; Pcs Nagendra Setty; and Sasi Institute Of Technology And Engineering.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: The present invention relates to an intelligent healthcare analytics system and method for predicting diabetes risk using a hybrid feature optimization framework and an attention-enhanced Siamese deep learning architecture. The invention comprises a preprocessing module configured to cleanse, normalize, and encode structured clinical data obtained from electronic health records and healthcare repositories. A hybrid feature optimization module employs Information Gain Filtering (IGF), Dynamic Metaheuristic Optimization (DMO), and Refined Feature Optimization (RFO) to identify a compact set of highly discriminative clinical features while reducing redundancy and computational complexity. The optimized feature set is supplied to a Siamese one-dimensional Convolutional Neural Network (1D-CNN) integrated with a Global Spatial-Channel Attention (GSCA) mechanism. The Siamese architecture performs contrastive representation learning on paired patient records to generate discriminative embeddings capable of distinguishing diabetic and non-diabetic subjects. The GSCA module enhances feature representation by adaptively emphasizing clinically significant attributes and suppressing irrelevant information. A classification module utilizes the learned embeddings to generate diabetes risk predictions with improved accuracy and robustness. The proposed invention enables efficient large-scale diabetes screening, clinical decision support, population health monitoring, and early disease intervention. The architecture provides enhanced predictive performance, improved handling of class imbalance, reduced feature dimensionality, and scalable deployment across healthcare environments. The invention is applicable to medical diagnosis systems, intelligent healthcare platforms, and predictive analytics applications.
Disclaimer: Curated by HT Syndication.
