Automatic Stress Recognition System with Deep Learning using Multimodal Psychological Data

There are numerous applications for detecting stress states from physiological inputs. Human stress detection may be used to enhance the human experience as well as monitor and prevent stress-related disorders. Previous research has leveraged existing machine learning techniques for automated stress identification. However, there are also some limitations. Deep learning allows you to discover basic patterns in your body's reactions that are often difficult to notice. This study offers a convolutional network-based approach for detecting human stress utilizing multimodal data sets obtained by smart sensors for physiological and motion, which can aid humans in avoiding health issues caused by stress. Sensor information like triaxial accelerates, heartbeat, body temperature, respiration, electromyographic (EMG) data for physiological states stressed and non-stressed, are included in the WESAD data collection. The use of Long short - term memory method and the DNN method to detect stress in people in real-time is investigated in this study. Our research is based on their physical reactions. The study's findings provide an assessment of both models' ability to predict real-time users' mental expressions based on physiological parameter fluctuation.