Recognition of Human Mental Stress Using Machine Learning Paradigms

Abstract

Nowadays it is very normal for humans to experience mild or moderate mental stress in a variety of situations. A manageable amount of stress is good for an individual, however, too much of stress affects person’s mental health and is a guarantor for suicidal risks if left unnoticed over a longer period. It has been proven that long term stress correlates with physical health problems. With the increasing number of people undergoing stress, it is crucial to be able to detect it at an early stage and help people realize and resolve it before much damage is done. The traditional methods of assessing stress levels are by interviewing the individual and by observing the facial gestures. In the interview, stress related questions are asked to have a better understanding of individual’s condition. People under stress react by giving different facial expressions i.e. the eyebrows shape differently, their pupils dilate, or the blinking rate might differ. These methods are limited as they may miss stress episodes.

Keywords: Stress, EEG, Speech, Audio-visual, Emotion, Machine Learning