MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641049239 A) filed by Sr University, Warangal, Telangana, on April 17, for 'hybrid neural network framework for real-time biomedical signal classification.'
Inventor(s) include Dr Charulatha G; and Dr Balajee Maram.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to the development of a hybrid neural network framework (HNNF) designed for real-time analysis of biomedical signals including ECG and EEG using CNNs for extracting the spatial features and LSTM for capturing the time dependency of the input data. This framework uses a multi-scale attention model to assign weights to the features of the biomedical signals, making the framework resilient to noise and artifacts commonly associated with wearable devices. Trained using an innovative gradient-based pruning technique, HNNF provides 98.2% accuracy on MIT-BIH and CHB-MIT datasets while reducing the computational cost by 75% as compared to existing frameworks, making it viable for deployment on edge IoT devices."
Disclaimer: Curated by HT Syndication.
