Deep neural networks detect suicide risk from textual facebook posts
Abstract Detection of suicide risk is a highly prioritized, yet complicated task. Five decades of research have produced predictions slightly better than chance (AUCs = 0.56–0.58). In this study, Artificial Neural Network (ANN) models were constructed to predict suicide risk from everyday language o...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2020-10-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-020-73917-0 |
| _version_ | 1818845946285391872 |
|---|---|
| author | Yaakov Ophir Refael Tikochinski Christa S. C. Asterhan Itay Sisso Roi Reichart |
| author_facet | Yaakov Ophir Refael Tikochinski Christa S. C. Asterhan Itay Sisso Roi Reichart |
| author_sort | Yaakov Ophir |
| collection | DOAJ |
| description | Abstract Detection of suicide risk is a highly prioritized, yet complicated task. Five decades of research have produced predictions slightly better than chance (AUCs = 0.56–0.58). In this study, Artificial Neural Network (ANN) models were constructed to predict suicide risk from everyday language of social media users. The dataset included 83,292 postings authored by 1002 authenticated Facebook users, alongside valid psychosocial information about the users. Using Deep Contextualized Word Embeddings 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 (0.621 ≤ AUC ≤ 0.629), the MTM produced significantly improved prediction accuracy (0.697 ≤ AUC ≤ 0.746), with substantially larger effect sizes (0.729 ≤ d ≤ 0.936). Subsequent content analyses suggested that predictions did not rely on explicit suicide-related themes, but on a range of text features. The findings suggest that machine learning based analyses of everyday social media activity can improve suicide risk predictions and contribute to the development of practical detection tools. |
| first_indexed | 2024-12-19T05:37:44Z |
| format | Article |
| id | doaj.art-f2eff099f81a47a5be67597ff0d20008 |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-12-19T05:37:44Z |
| publishDate | 2020-10-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-f2eff099f81a47a5be67597ff0d200082022-12-21T20:34:05ZengNature PortfolioScientific Reports2045-23222020-10-0110111010.1038/s41598-020-73917-0Deep neural networks detect suicide risk from textual facebook postsYaakov Ophir0Refael Tikochinski1Christa S. C. Asterhan2Itay Sisso3Roi Reichart4The Hebrew University of JerusalemThe Hebrew University of JerusalemThe Hebrew University of JerusalemThe Hebrew University of JerusalemThe Faculty of Industrial Engineering and Management, Technion—Israel Institute of TechnologyAbstract Detection of suicide risk is a highly prioritized, yet complicated task. Five decades of research have produced predictions slightly better than chance (AUCs = 0.56–0.58). In this study, Artificial Neural Network (ANN) models were constructed to predict suicide risk from everyday language of social media users. The dataset included 83,292 postings authored by 1002 authenticated Facebook users, alongside valid psychosocial information about the users. Using Deep Contextualized Word Embeddings 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 (0.621 ≤ AUC ≤ 0.629), the MTM produced significantly improved prediction accuracy (0.697 ≤ AUC ≤ 0.746), with substantially larger effect sizes (0.729 ≤ d ≤ 0.936). Subsequent content analyses suggested that predictions did not rely on explicit suicide-related themes, but on a range of text features. The findings suggest that machine learning based analyses of everyday social media activity can improve suicide risk predictions and contribute to the development of practical detection tools.https://doi.org/10.1038/s41598-020-73917-0 |
| spellingShingle | Yaakov Ophir Refael Tikochinski Christa S. C. Asterhan Itay Sisso Roi Reichart Deep neural networks detect suicide risk from textual facebook posts Scientific Reports |
| title | Deep neural networks detect suicide risk from textual facebook posts |
| title_full | Deep neural networks detect suicide risk from textual facebook posts |
| title_fullStr | Deep neural networks detect suicide risk from textual facebook posts |
| title_full_unstemmed | Deep neural networks detect suicide risk from textual facebook posts |
| title_short | Deep neural networks detect suicide risk from textual facebook posts |
| title_sort | deep neural networks detect suicide risk from textual facebook posts |
| url | https://doi.org/10.1038/s41598-020-73917-0 |
| work_keys_str_mv | AT yaakovophir deepneuralnetworksdetectsuicideriskfromtextualfacebookposts AT refaeltikochinski deepneuralnetworksdetectsuicideriskfromtextualfacebookposts AT christascasterhan deepneuralnetworksdetectsuicideriskfromtextualfacebookposts AT itaysisso deepneuralnetworksdetectsuicideriskfromtextualfacebookposts AT roireichart deepneuralnetworksdetectsuicideriskfromtextualfacebookposts |