Anxiety Analyzer: A Deep Learning Approach for Classification of Stress

Abstract

In today’s digital era and the pandemic situation like Covid 19, most of the things are conducted in online mode like lectures, businesses, conferences, etc. So the interaction and communication amongst the peoples is lagging behind. Due to this, the chances of mental health diseases like depression, anxiety, loneliness are increasing. In existing solutions, they are diagnosed by means of abstract methods such as quiz, questionnaire or discussion with the patient. But there is inaccuracy and uncertainty in diagnosing mental health diseases with the help   of traditional abstract methods. Nearly 9 out of 10 Indians are stressed. The 2018 Cigna 360 Well-Being Surveys — Future Assured, conducted by Cigna TTK Health Insurance, shows that Indians are more stressed than people in other developed countries like the United States, and other developed and emerging countries. Moneycontrol reports that 95% of Indian millennials aged 18-34 are stressed, surpassing the global average of 86%. Shockingly, 1 out of 8 Indians struggle immensely with managing stress. Some Indian respondents also revealed that 17% of Indians are hesitant to seek help from medical professionals when   it comes to discussing their problems or stress issues. Also, the consultation cost is one of the major barriers to seeking medical professionals. Due to stress many times patients suffer from severe headaches so our solution was going around this fact. There are many types of headaches from which humans suffer. They are migraine, hypertension, stress etc. The proposed solution to this problem mentioned in this paper is to detect the mental health diseases with accuracy by means of generating a body heat map of the patient with the help of a thermal camera and applying techniques of image processing and deep learning image classification models to measure the magnitude of the detected disease. And finally the model will predict whether the person is in stress or not.