Related Experiment Video
Updated: Jun 17, 2026
Virtual Agent for Real-Time Motivational Interviewing by Integrating Adaptive Nonverbal Behavior and Language Models
Published on: December 23, 2025
Mapping mental health and suicide-related discourse on TikTok: An AI-based social media listening study
Julia Marti- Ochoa1, Ana Freire2, Nia Plamenova Djourovaand2
1Department of Economics and Business, University of Lleida, Lleida, Spain.
Purpose:
To examine engagement patterns around mental health and suicide-related discourse on TikTok, a platform widely used by youth and Generation Z, by comparing problem-focused (e.g., depression, loneliness, suicide) with recovery-, support-, and prevention-oriented hashtags (e.g., #mentalhealth, #youarenotalone, #selflove). We further assess whether engagement differs by sentiment, discrete emotions, creator type, geotagging, verification status, and publication year.
Methodology:
We collected 12,587 TikTok posts published between 2020 and 2025 from 27 Spanish, English, and recovery-, support-, and prevention-oriented hashtags and analyzed them using an automated low-code pipeline integrating AI-based sentiment and emotion models. Preprocessing was conducted in Python under privacy-by-design principles; identifiers were anonymised, and engagement differences tested with t-tests and ANOVA.
Findings:
Posts classified as neutral or negative, and those expressing sadness and fear, were associated with higher engagement than positive content. Engagement was higher for nano- and micro-creators, unverified accounts, and posts without geotags, consistent with the prominence of perceived authenticity and proximity on the platform. Engagement peaked in 2020-2021, declined during 2022-2024, and increased again in 2025. Problem-focused hashtags related to depression, loneliness, and suicide showed the highest engagement; several recovery-, support-, and prevention-oriented hashtags, particularly #mentalhealth, #youarenotalone, #selflove, and #suicideprevention, also attracted substantial engagement.
Originality:
TikTok simultaneously hosts visible well-being-oriented discourse, while highly negative emotional content is associated with higher levels of engagement. Interpreted through the Werther/Papageno dual framework, these patterns indicate that while negative emotional tone attracts interaction, the specific topical focus (hashtag) is a far stronger driver of engagement. Mapping how engagement varies across emotional tone, creator attributes, and hashtag communities can inform youth-centered communication strategies and improve understanding of content visibility and interaction patterns on the platform.
