Suicide and Self-harm Prediction Based on Social Media Data Using Machine Learning Algorithms

Aug 4, 2026

Upload by Finn

Online social networking (SN) data forms a context- and time-rich stream that can be used to forecast suicidal ideation and self-harm. Despite the promise of digital media, predictive modeling of acute suicidal ideation remains insufficient. This study applies machine learning to a previously published Instagram dataset of youths, using language- and activity-based predictors. Models are evaluated with out-of-sample, cross-validated testing, with explanation methods to assess predictor relevance and subject-level patterns, and ensemble learning to reduce modeling complications. Future work will validate digital biomarkers on larger, more diverse populations.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Social media as a predictive context

Risks and motivation for caution

Method

Data preparation

Datasets

Suicide and self-harm prediction based on social media data using machine learning algorithms Abdulrazak Yahya Saleh a, 1,*, Fadzlyn Nasrini Binti Mostapa a,2 a Faculty of Cognitive Sciences and Human Development, Universiti Malaysia Sarawak, Malaysia [1](1 abdulrazakalhababi@gmail.com)[ abdulrazakalhababi@gmail.com](1 abdulrazakalhababi@gmail.com); [2](2 69662@siswa.unimas.my)[ 69662@siswa.unimas.my](2 69662@siswa.unimas.my) * Corresponding Author ARTICLE INFO ABSTRACT | Article history
Received April, 04, 2023 Revised April, 11, 2023 Accepted May, 01, 2023
Keywords
Social networking Machine learning Algorithms Suicide
Self-harm |
Online social networking (SN) data is a context and time rich data stream that has showed potential for predicting suicidal ideation and behaviour. Despite the obvious benefits of this digital media, predictive modelling of acute suicidal ideation (SI) remains underdeveloped at now. In combined with robust machine learning algorithms, social networking data may provide a potential path ahead. Researchers applied a machine learning models to a previously published Instagram dataset of youths. Using predictors that reflect language use and activity inside this social networking, researchers compared the performance of the out-of-sample, cross-validated model to that of earlier efforts and used a model explanation to further investigate relative predictor relevance and subject-level phenomenology. The application of ensemble learning approaches to SN data for the prediction of acute SI may reduce the complications and modelling issues associated with acute SI at these time scales. Future research is required on bigger, more diversified populations to refine digital biomarkers and assess their external validity with more rigor.
This is an open access article under the CC–BY-SA license.
| | --- | --- | 1. Introduction The term "social media" refers to an online platform that enables social interaction, such as Facebook, Twitter, YouTube, and Instagram [1] . [2] assert that social media is fundamentally defined by three critical concepts which are cognition, communication, and cooperation. These concepts endow social media with a variety of various forms of sociality, including information, facts, and knowledge, activities, relationships, communities, and partnerships [2] . The numerous forms of sociality that social media supports allowing the platform to serve a variety of critical roles for its users, notably communication, relationship creation and maintenance, and thus a platform for providing knowledge [3] . These features are highly valued by social media users, particularly young adults. Youth are classified as ardent social media users in a variety of research conducted globally [3]. Malaysian youth are likewise reported to be avid social media users. While there is no debate about the use of social media [4], it is also necessary to exercise caution regarding the medium's hazards. Spam hoaxes, cyberbullying, online harassment, and sexting are just a few of the dangers [1]. Youth who use social media risk privacy assaults and depression [1] . These dangers can be avoided if youth possess an acceptable level of social media competency. Presently, social media platforms such as Facebook and Instagram have become the primary sources of information for assisting people of modern society in adjusting to their new lifestyle. Indeed, the global population of social media users is predicted to reach 3.02 billion by 2021, and one of the important uses of social media may be to encourage healthy lifestyles and to enhance people's management of their health status [5]. In Malaysia specifically, a 2018 poll performed by the Malaysian Communications and Multimedia Commission (MCMC) discovered that many Malaysian internet users, particularly young users, shared content online via social media (61.8 percent) . According to the survey, 97.3 percent of Malaysians use Fac

FAQs

What is the purpose of using social media data in suicide and self-harm prediction?

The study aims to leverage context- and time-rich online networking data to improve prediction of suicidal ideation and self-harm behaviors.

Which dataset and platform are used for model development?

Models are developed using a previously published Instagram dataset of youths containing both suicide-indicative and non-suicidal posts.

How are machine learning models evaluated and interpreted in the study?

The study compares out-of-sample, cross-validated performance against earlier efforts and uses model explanation methods to investigate predictor relevance and subject-level patterns.

Aug 4, 2026

Upload by Finn