@article{ref1,
title="Application of machine learning analysis for prediction of repeated suicide attempts among the attempters: a nationwide population study in Taiwan",
journal="Journal of suicidology (Taipei)",
year="2023",
author="Huang, Min-Wei and Jian, Bo-Lin and Fan, Kai-Sen and Wu, Chia-Yi and Lee, Ming-Been and Chan, Chia-Ta and Chen, Chun-Ying",
volume="18",
number="1",
pages="487-492",
abstract="Background and purpose: The use of big data analysis in suicide prevention has gained increasing interest, and machine learning techniques can be used to predict potential suicide behavior. These include medical, behavioral, and semantic models that can analyze large amounts of historical data to identify the potential occurrence of such behavior. Risk factors for suicide include mental illness, past suicide attempts, drug or alcohol abuse, family history, loneliness, lack of support network, major life transitions, social exclusion, poverty, and impulsive behavior. In Taiwan, the National Suicide Surveillance System (NSSS) for suicide attempters has been established since 2006, and studies have identified characteristic factors in the database that are useful for suicide risk analysis. The study aimed to develop predictive models based on machine learning to provide sensitive and effective measures for the prevention of reattempts among suicide attempters. <br><br>METHODS: All the suicide attempters reported to the NSSS in 2020 were included in the study. The model's results were compared using various machine learning techniques, and the Ensemble Learning model had the highest recognition rate of 75%. <br><br>RESULTS: There were 323,701 attempters recruited in the study. The results indicated the significance of five feature values in repeated-suicide rate prediction (i.e., age at the time of report, psychiatric patients registered for community aftercare, history of psychiatric disorders, reporting unit, and perpetrator of sexual assault) and the Boosted Trees were suitable for building models for this type of data. <br><br>CONCLUSION: The findings from the machine learning analysis provided evidence for the government, related departments, and non-governmental organizations to collaborate for the implementation of effective suicide prevention strategies.<p /> <p>Language: zh</p>",
language="zh",
issn="2790-1645",
doi="10.30126/JoS.202303_18(1).0011",
url="http://dx.doi.org/10.30126/JoS.202303_18(1).0011"
}