MUMBAI, India, July 30 -- Intellectual Property India has published a patent application (202641088803 A) filed by Santha Seelan Chandran; and T Krishna Murthy on July 21, 2026, for A System And Method For Hierarchical Multimodal Sentiment Analysis Using Graph Neural Network-Based Feature Fusion.

Inventors include T Krishna Murthy; Dr. R. Shekhar; and Alliance University.

The application for the patent was published on July 24, 2026, under issue no. 30/2026.

Abstract: A graph neural network based three-level feature fusion system and method for clinical multimodal sentiment analysis is disclosed. The invention provides an intelligent healthcare analytics framework for deriving emotional and sentiment states from a heterogeneous multimodal data, including textual, audio and visual inputs. Textual and audio data are preprocessed by the system according to the modality, such as tokenization, removal of stop words and noise reduction. A hierarchical feature extraction architecture is used which extracts textual features using an enhanced Clinical Bidirectional Encoder Representations from Transformers (Clinical- BERT) model, audio features by Mel-Frequency Cepstral Coefficients (MFCCs) processed by a Pyramid Pooling (PP) Convolutional Neural Network (CNN), and visual spatiotemporal features by Tri-Stream Inflated Three-Dimensional Convolutional Neural Network (CNN) with depth-separable convolutions. The extracted low-, mid- and high-level features are represented as nodes in a multimodal graph and learned using a Graph Neural Network (GNN) to capture inter-modal relationships and context. The fused representation is then passed through a fully connected layer followed by a SoftMax classifier, which are used to make sentiment predictions. The invention enables more accurate and robust recognition of multimodal sentiment, more interpretability, and more clinically useful support, by effectively capturing complex emotional interactions across multiple data modalities. The system can be used in telemedicine systems, clinical decision support systems, healthcare feedback analysis, mental health assessment, and patient monitoring.

Disclaimer: Curated by HT Syndication.