Suicide Risk Prediction Using Artificial Intelligence
DOI:
https://doi.org/10.31436/ijpcc.v10i2.468Keywords:
Suicide risk, Machine Learning, Reddit, Artificial IntelligenceAbstract
Over the past decade, social media has been attracting a growing number of people to the online space. Due to the increase in internet usage, a huge number of text data has been produced. Such data can reflectusers’ mental healthstatus, but it is still challenging to predictsuicide risk from data,due to the high complexity of texts.This research aims to predict the suicide risk from Reddit posts using artificial intelligence (AI). The data were collected from the Kaggle dataset, which includedpostingsof suicide subreddits.The datawere pre-processed throughnatural language processing techniques. Logistic regression, naive Bayes, and random forest models were then used for classifying the Reddit users, i.e., to predict if they are in a suicidal or non-suicidal mental state. These models were compared to identify an AI approach that provides the best performance among the three models. Then, the logistic regression model with doc2vec showed the highest precision of 0.92, recall 0.92, and F1score of 0.92.
References
World Health Organization, “Suicide,” World Health Organisation, Aug. 28, 2023. https://www.who.int/news-room/fact-sheets/detail/suicide
H.-C. Shing, S. Nair, A. Zirikly, M. Friedenberg, H. Daumé III, and P. Resnik, “Expert, Crowdsourced, and Machine Assessment of Suicide Risk via Online Postings,” ACLWeb, Jun. 01, 2018. https://aclanthology.org/W18-0603/
S. Ji, C. P. Yu, S. Fung, S. Pan, and G. Long, “Supervised Learning for Suicidal Ideation Detection in Online User Content,” Complexity, vol. 2018, pp. 1–10, Sep. 2018, doi: https://doi.org/10.1155/2018/6157249.
S. C. Shetty, “A Deep Learning Approach for Suicide Risk Assessment using Reddit,” norma.ncirl.ie, 2020. https://norma.ncirl.ie/4420/ (accessed Dec. 29, 2023).
Aldhyani, T. H., Alsubari, S. N., Alshebami, A. S., Alkahtani, H., & Ahmed, Z. A. (2022). Detecting and analyzing suicidal ideation on social media using deep learning and machine learning models. International journal of environmental research and public health, 19(19), 12635.
Ji, S., Yu, C. P., Fung, S. F., Pan, S., & Long, G. (2018). Supervised learning for suicidal ideation detection in online user content. Complexity, 2018.
O'dea, B., Wan, S., Batterham, P. J., Calear, A. L., Paris, C., & Christensen, H. (2015). Detecting suicidality on Twitter. Internet Interventions, 2(2), 183-188.
L. J. Richard and G. G. Koch, “The Measurement of Observer Agreement for Categorical Data,” Biometrics, vol. 33, no. 1, pp. 159–174, Mar. 1977, doi: https://doi.org/10.2307/2529310.
M. H. Zweig and G. Campbell, “Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine,” Clinical Chemistry, vol. 39, no. 4, pp. 561–577, Apr. 1993, Available: https://pubmed.ncbi.nlm.nih.gov/8472349/
L. Vergni and F. Todisco, “A Random Forest Machine Learning Approach for the Identification and Quantification of Erosive Events,” Water, vol. 15, no. 12, p. 2225, Jan. 2023, doi: https://doi.org/10.3390/w15122225.
