Abstract
In this research proposal, the disparity in stress severity is modeled using a deep learning framework to determine mental stress. A wireless network sensor platform is used to monitor various physiological signals, such as heart rate variation, skin conductance, and breathing pattern irregularities that are activated by providing a challenging atmosphere inside a laboratory. A set of protocols is designed using a range of cognitive experiments that engage participants in a series of mental activities with various levels of challenges. The participant feels stress that varies in severity when undergoing these challenges. To relax the mind and body from stress, a deep breathing technique is used that is performed before and after each cognitive activity. Apart from the traditional physiological signals, cerebral features are also extracted from the neural signals. To identify the stressed activities and their severity, a convolutional neural network (CNN) framework is employed for training and validating the input datasets. It is found that the neural signals significantly improve the efficiency of the proposed classification model in computing mental stress. The study also supports the idea that the deep learning framework results in an improved estimate to determine mental stress.