Quantifying the Suicidal Tendency on Social Media: A Survey

TL;DR

This survey reviews recent research on using machine learning and deep learning to quantify suicidal tendencies from social media data, highlighting feature classification, data handling, and future research directions.

Contribution

It provides a comprehensive taxonomy of mental healthcare related to social media analysis and synthesizes over 77 studies from 2013 to 2021 on this topic.

Findings

Classification of heterogeneous social media features for suicide risk detection

Identification of recent advances in ML and DL models for mental health analysis

Highlighting new research directions in social media-based mental health assessment

Abstract

Amid lockdown period more people express their feelings over social media platforms due to closed third-place and academic researchers have witnessed strong associations between the mental healthcare and social media posts. The stress for a brief period may lead to clinical depressions and the long-lasting traits of prevailing depressions can be life threatening with suicidal ideation as the possible outcome. The increasing concern towards the rise in number of suicide cases is because it is one of the leading cause of premature but preventable death. Recent studies have shown that mining social media data has helped in quantifying the suicidal tendency of users at risk. This potential manuscript elucidates the taxonomy of mental healthcare and highlights some recent attempts in examining the potential of quantifying suicidal tendency on social media data. This manuscript presents the classification of heterogeneous features from social media data and handling feature vector representation. Aiming to identify the new research directions and advances in the development of Machine Learning (ML) and Deep Learning (DL) based models, a quantitative synthesis and a qualitative review was carried out with corpus of over 77 potential research articles related to stress, depression and suicide risk from 2013 to 2021.