![[2510.14889] Detecting Early and Implicit Suicidal Ideation via Longitudinal and Information Environment Signals on Social Media](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2510.14889] Detecting Early and Implicit Suicidal Ideation via Longitudinal and Information Environment Signals on Social Media
Summary
This article presents a computational framework for detecting early and implicit suicidal ideation on social media by analyzing user interactions and posting histories.
Why It Matters
Understanding and detecting suicidal ideation early is crucial for intervention. This research highlights the potential of social media signals as predictive indicators, which can inform the development of better mental health support systems.
Key Takeaways
- The study improves detection of suicidal ideation by 10% using a novel computational framework.
- Peer interactions on social media provide valuable signals for identifying at-risk individuals.
- The approach utilizes a fine-tuned DeBERTa-v3 model for enhanced predictive accuracy.
Computer Science > Social and Information Networks arXiv:2510.14889 (cs) [Submitted on 16 Oct 2025 (v1), last revised 21 Feb 2026 (this version, v3)] Title:Detecting Early and Implicit Suicidal Ideation via Longitudinal and Information Environment Signals on Social Media Authors:Soorya Ram Shimgekar, Ruining Zhao, Agam Goyal, Violeta J. Rodriguez, Paul A. Bloom, Navin Kumar, Hari Sundaram, Koustuv Saha View a PDF of the paper titled Detecting Early and Implicit Suicidal Ideation via Longitudinal and Information Environment Signals on Social Media, by Soorya Ram Shimgekar and 7 other authors View PDF HTML (experimental) Abstract:On social media, several individuals experiencing suicidal ideation (SI) do not disclose their distress explicitly. Instead, signs may surface indirectly through everyday posts or peer interactions. Detecting such implicit signals early is critical but remains challenging. We frame early and implicit SI as a forward-looking prediction task and develop a computational framework that models a user's information environment, consisting of both their longitudinal posting histories as well as the discourse of their socially proximal peers. We adopted a composite network centrality measure to identify top neighbors of a user, and temporally aligned the user's and neighbors' interactions -- integrating the multi-layered signals in a fine-tuned DeBERTa-v3 model. In a Reddit study of 1,000 (500 Case and 500 Control) users, our approach improves early and imp...
