Does Social Media Feed Tell about Your Mental State? A Deep Randomised Neural Network Approach

Depression is one of the most severe issues in the present-day technology-driven world. Digital transformation and its offerings have penetrated our life. Social media platforms such as Facebook, Twitter, Instagram etc have enabled the masses to express their opinions/experiences and information instantly to large audience. In the expression of their opinions/experiences and information, Social media users also intensely express their deep feelings/reflections/emotions in the form of text/video/images and emoji etc. These shared emotions/feelings and experiences and used words have a deep connection with their mental health and depression etc and can be used to get insights and early detection of their mood, mental health and depression. Previously detection of depression and mental health has a standard procedure involving a fully structured interview method(SID) or semi-structured interview method (SSDI) to collect a large amount of data to analyse depression. Present-day, social media feed and other associated metadata show the manifestation of depressive disorder symptoms of the user. The proposed research work experiments on social media posts and analyses the mental state of social media users. The model proposes a hybrid Frequency-Inverse Document and a deep learning model Deep Random Neural Networks(DRNNs) and compared with other state of art models such as random forest(RF), support vector machine(SVM) and other deep learning approaches. The proposed model reports an accuracy of approximately 91 cent and takes significantly less computational time as compared to other classification approaches. The proposed experiment can be useful for the individuals, parents, and psychologists to get intelligent assistance about the mental state/depression of persons through their social media posts and content for future course of action.