Mental stress assessment using SVM with physiological sensor data

Continuous exposure to stress threatens the physical and mental health of workers and reduces their quality of life. On the business side, stress has negative consequences, such as reduced productivity. Therefore, it needs to assess and manage stress. In this study, we assess stress using physiological signals collected from wearable devices. Stress was investigated by performing two different tasks, stress task and rest task, in a laboratory environment. As a result of the experiment, the SVM with the heart rate and HRV collected from ECG as input features was able to classify stress with an accuracy of 95%, and the SVM using the heart rate collected from PPG as an input feature was able to classify stress with an accuracy of 75%. It was revealed that ECG responded more sensitively to assess stress than PPG and various HRV features are required for stress assessment.