MUMBAI, India, Feb. 13 -- Intellectual Property India has published a patent application (202641009163 A) filed by Sr University, Warangal, Telangana, on Jan. 29, for 'a system and method for real-time stress detection using multimodal physiological signals and hybrid deep learning models.'
Inventor(s) include Asalam Patel; and V. Thirupathi.
The application for the patent was published on Feb. 13, under issue no. 07/2026.
According to the abstract released by the Intellectual Property India: "With the rising prevalence of stress-related mental health issues worldwide, effective and accurate stress detection is essential for timely intervention. This research aims to enhance the precision of stress detection using EEG signals by developing innovative feature extraction techniques and implementing advanced hybrid deep learning models. A key challenge in EEG-based stress detection is the presence of noise, which can significantly hinder the accuracy of classification. To address this, advanced feature extraction methods are introduced, designed specifically to isolate and enhance stress-related patterns within noisy EEG data. Furthermore, the proposed study will also investigate the design of hybrid deep learning models that combine the strengths of various machine learning and deep learning algorithms. These models will be optimized not only for higher accuracy in classifying stress levels but also for computational efficiency, making them suitable for real-time applications. In addition, the integration of novel deep learning architectures, such as transformers and attention mechanisms, will be explored to further boost the model's capability in learning complex temporal and spatial dependencies in EEG data. In a multimodal framework, this research also integrates complementary physiological signals, including heart rate and electrooculogram (EOG), alongside EEG data. The combination of these signals creates a comprehensive system capable of detecting stress more robustly in diverse real-world settings, where single-modality approaches may fall short. Finally, the proposed models will be rigorously tested against existing methods to evaluate both their accuracy and computational efficiency. This comparison underscores the practicality of the hybrid deep learning models, particularly for real-time stress detection in environments where rapid and reliable responses are critical. The results will demonstrate that the proposed approaches not only improve classification performance but also achieve the computational speed needed for deployment in real-time stress monitoring systems. Keywords: Stress Detection, EEG Signals, Feature Extraction, Hybrid Deep Learning, Attention Mechanism, Transformer Models, Multimodal Physiological Signals, Real-Time Monitoring."
Disclaimer: Curated by HT Syndication.
