Stress Detection from Pupil Diameter Signal Using Neural Network

In this work, the goal is to diagnose the occurrence of stress in individuals by observing the changes of pupil diameter signal that done by conducting a question and answer experiment on a group of people. The effect of autonomic nervous system function on involuntary behaviors, such as pupil diameter, heart rate, respiration rate, blood pressure, etc. has made it possible for researchers to identify certain events, such as stress in the nervous system of individuals, by observing these behaviors throughout time. Extraction of the pupil diameter signal is cheaper than other physiological signals, which can be acquired noninvasively and can easily combined with other physiological measurements. In addition, extraction of the pupil diameter signal can be done on a wide range of people. In order to classify the signals, we use the deep-belief neural network. the deep-belief neural network consists of restricted Boltzmann machine's layers has a good ability to extract signal characteristics and then classify them. The results and comparison them with other similar works indicate that the pupil diameter signal has a high ability to show the mental states of people in terms of relaxation and stress.