MENTAL HEALTH AND DEPRESSION DETECTION FROM SOCIAL MEDIA POSTS USING MACHINE LEARNING AND NLP

Abstract

The rapid emergence of social media makes it an excellent source for understanding social sentiment, opinion, and mental health. This project investigates the classification of mental health conditions through social media sentiment analysis as implemented through machine learning and natural language processing (NLP) methods. Text data scraped from relevant sites underwent pre-processing into numerical data through the Term Frequency-Inverse Document Frequency (TF-IDF) mechanism for classifiable information. Two models of machine learning, Support Vector Machine (SVM) and Multinomial Naive Bayes, were implemented and assessed for their ability to identify indications of mental health conditions. Experimental results illustrate that both models achieved comparable levels of accuracy with SVM performance slightly above that of its peer. Comparative analysis includes a balance between accuracy, model run time, and interpretability suggesting that both models effectively assist in the development of automated solutions for proactive mental health identification and education.