Deep learning-based stress detection using multimodal biosignals

Abstract

Abstract Stress-related health problems drive an urgent need for sensitive, real-time monitoring devices capable of capturing psychophysiological stress dynamics in natural settings. This study presents a stress detection system that uses deep learning, combining synchronized physiological signals and survey data from university students who experienced controlled stress through psychological sessions and examinations. We analyzed two model families with differing complexities: a deep neural network (DNN)-based model for static physiological patterns and a long short-term memory (LSTM)-based model designed to capture temporal dependencies in biosignals. The DNN (86.42% accuracy) was outperformed by the LSTM-based model (94.14% accuracy). The statistical results showed a relationship between stress and gender (females: $$\mu $$ μ = 2.2, males: $$\mu $$ μ = 2.7, p < 0.05) and smoking (smokers: $$\mu $$ μ = 3.7, nonsmokers: $$\mu $$ μ = 2.2, p < 0.05). Academic elements also impacted stress perception. The stress scores were in accordance with self-reported measurements, which validated the model.