A deep learning approach to stress recognition through multimodal physiological signal image transformation

Stress is widely acknowledged as a significant contributor to health issues. Recognizing stress involves assessing an individual's physiological and psychological responses to stressors, which is crucial for human well-being. Physiological signal-based stress assessment offers greater accuracy and objectivity compared to traditional methods. To enhance stress level detection, we propose a novel approach using deep learning models that classify mental stress states (stress, baseline, amusement) based on multimodal physiological signals converted into RGB images through Gramian Summation Angular Field (GASF), Gramian Difference Angular Field (GADF), and Markov Transition Field (MTF) transformations. Experimental findings showcase the effectiveness of the proposed model, achieving an accuracy of 90.96% and an F1-score of 91.67%. The consistently high F1 scores across all categories demonstrate the model's exceptional performance. Experimental results underscore the method's effectiveness in capturing the relationship between multimodal physiological signals and stress, offering a promising tool for mental stress recognition.