"TRENDS AND GAPS IN SENTIMENT-BASED MENTAL HEALTH MONITORING ON SOCIAL MEDIA: A COMPARATIVE STUDY"

Abstract

This paper provides a comparison of many current articles in this area of mental health surveillance in social media. The goal is to review existing methodologies structured in the way that helps to see the trends in research and identify the existing gaps. All the papers are tested regarding five main dimensions: Technique Advancement and Modality Richness, Dataset Quality, Reported Accuracy, and Innovation Level. The analysis shows that the papers that scored better in their entirety usually utilize more cutting-edge methods like transformer models and use multimodal data sources to infer better mental health. By contrast, studies with lower scoring are usually based on single-modality inputs and standard machine learning algorithms, with fewer innovations in terms of design or construction of the data that they model. The current comparison highlights the increased significance of modality combination and high-quality dataset in improving the precision of mental health analysis and depth on social platforms. It also gives a base to steer the future studies in more wholesome, consistent, and scaled solutions.

KeyWords

Mental Health Analysis, Sentiment Analysis, Social Media Sentiment Analysis, Comparative Study, Multimodal Data, Deep Learning and Machine Learning, Depression detection