Intellectual Property India Publishes Patent Application for An Attention-Integrated Hybrid Bilstm–cnn Deep Learning Architecture For Accurate Cardiovascular Disease Classification Using Multimodal Cardiac Feature Representations

MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202641098300 A) filed by Sr University, Warangal on August 13, 2026, for An Attention-Integrated Hybrid Bilstm–cnn Deep Learning Architecture For Accurate Cardiovascular Disease Classification Using Multimodal Cardiac Feature Representations.

Inventors include Mrs. Mary Teresa; and Dr. Anurodh Kumar.

The application for the patent was published on August 21, 2026, under issue no. 34/2026.

Abstract: The present invention discloses an Attention-Integrated Hybrid BiLSTM–CNN Deep Learning Architecture for accurate cardiovascular disease (CVD) classification using multimodal cardiac feature representations. The proposed architecture introduces a unified deep learning framework that combines hierarchical convolutional neural networks (CNNs) for spatial feature extraction, bidirectional long short-term memory (BiLSTM) networks for temporal sequence modeling, and an adaptive multi-level attention mechanism for intelligent feature prioritization. The convolutional subsystem learns discriminative morphological characteristics from cardiac signal representations, while the BiLSTM subsystem captures long-range temporal dependencies associated with cardiac rhythms and disease progression. An adaptive attention module dynamically assigns importance to diagnostically relevant feature regions, suppressing redundant or less informative representations before feature integration. A hierarchical feature integration network combines the refined spatial and temporal representations to generate a comprehensive cardiac feature embedding for disease classification. A cardiovascular decision module subsequently predicts disease categories with calibrated confidence scores, while an explainability unit produces attention-based visualizations highlighting the most influential features contributing to the prediction. The proposed architecture improves classification accuracy, robustness, interpretability, and computational efficiency compared with conventional CNN, recurrent neural network, and hybrid deep learning models, making it suitable for intelligent clinical decision support and real-time cardiovascular disease diagnosis.

Disclaimer: Curated by HT Syndication.