Deep Neural Networks Detect Suicide Risk from Textual Facebook Posts

Abstract

Background: Detection of suicide risk is a highly prioritized, yet complicated task. In fact, five decades of suicide research produced predictions that were only marginally better than chance (AUCs = 0.56 – 0.58). Advanced machine learning methods open up new opportunities for progress in mental health research. In the present study, Artificial Neural Network (ANN) models were constructed to predict externally valid suicide risk from everyday language of social media users. Method: The dataset included 83,292 postings authored by 1,002 authenticated, active Facebook users, alongside clinically valid psychosocial information about the users. Results: Using Deep Contextualized Word Embeddings (CWEs) for text representation, two models were constructed: A Single Task Model (STM), to predict suicide risk from Facebook postings directly (Facebook texts → suicide) and a Multi-Task Model (MTM), which included hierarchical, multilayered sets of theory-driven risk factors (Facebook texts → personality traits → psychosocial risks → psychiatric disorders → suicide). Compared with the STM predictions (.606 ≤ AUC ≤ .608), the MTM produced improved prediction accuracy (.690 ≤ AUC ≤ .759), with substantially larger effect sizes (.701 ≤ d ≤ .994). Subsequent content analyses suggest that predictions did not rely on explicit suicide-related themes, but on a wide range of content. Conclusions: Advanced machine learning methods can improve our ability to predict suicide risk from everyday social media activities. The knowledge generated by this research may eventually lead to the development of more accurate and objective detection tools and get individuals the help they need in time.

Americans’ distress early in the COVID-19 pandemic: Protective resources and coping strategies.

“…Evaluations have found the data obtained on MTurk to be high quality, replicable, and valid across comparisons with frequently used academic platforms and student and professional samples (Kees et al, 2017;Sheehan, 2018). Although some research suggests that MTurk respondents report higher levels of depression than the general population (e.g., Ophir et al, 2020), other studies indicate mental health of MTurk workers approxi-…”

Proceedings of the Eighth Workshop on Computational Linguistics and Clinical Psychology

“…Tracking naturalistic language use on the internet is an effective method of measuring how people cope with trauma and experience emotions over time (Vine et al, 2020). Research has, for example, used both dictionary-based and openvocabulary analyses of online language use (including social media, online forums, and search engine activity) to understand how individuals anticipate and then cope with traumatic events such as suicide attempts Ophir et al, 2020;Roy et al, 2020), relationship dissolution (Seraj et al, 2021), illnesses such as breast cancer (Verberne et al, 2019) and autoimmune disease (Jordan et al, 2019), and mental health conditions such as anxiety (Ireland and Iserman, 2018) and depression (Eichstaedt et al, 2018).…”

Tracking Mental Health Risks and Coping Strategies in Healthcare Workers’ Online Conversations Across the COVID-19 Pandemic

“…Tracking naturalistic language use on the internet is an effective method of measuring how people cope with trauma and experience emotions over time (Vine et al, 2020). Research has, for example, used both dictionary-based and openvocabulary analyses of online language use (including social media, online forums, and search engine activity) to understand how individuals anticipate and then cope with traumatic events such as suicide attempts (De Choudhury et al, 2016;Ophir et al, 2020;Roy et al, 2020), relationship dissolution (Seraj et al, 2021), illnesses such as breast cancer (Verberne et al, 2019) and autoimmune disease (Jordan et al, 2019), and mental health conditions such as anxiety and depression (Eichstaedt et al, 2018).…”