中国临床心理学杂志2023，Vol.31Issue(6)：1442-1447,6.DOI:10.16128/j.cnki.1005-3611.2023.06.029
自杀行为预测中的机器学习方法:一项系统综述
Machine Learning Approaches to the Prediction of Suicidal Behavior:A Systematic Review
摘要
Abstract
Objective:To summarize the characteristics,application scenarios and model performance of machine learn-ing in predicting suicidal behavior.Methods:Three databases(PubMed,EMBASE,Web of Science)were systematically searched for articles using machine learning algorithms to predict suicidal ideation,suicidal attempts,and suicidal death up to November 2022.Results:A total of 71 articles were included and 81 models were extracted.Predicting suicidal behav-ior models of machine learning was good(AUC interval of suicidal ideation was 0.70～0.97;AUC interval of suicidal at-tempt:0.63～0.96;AUC interval for suicide:0.69～0.93),but the positive predictive value was low.Current studies mostly used clinical or community samples,but lacked school samples.The model included mostly demographic and psychopatho-logical variables and lacked psychological,physiological,and social variables.Conclusion:Machine learning improves the ability to predict suicide,and is expected to be as auxiliary screening tool used in clinical practice.关键词
机器学习/自杀意念/自杀尝试/自杀死亡Key words
Machine learning/Suicidal ideation/Suicidal attampt/Suicidal death分类
医药卫生引用本文复制引用
钱春莲,何宇豪,张强,尤静,杨丽..自杀行为预测中的机器学习方法:一项系统综述[J].中国临床心理学杂志,2023,31(6):1442-1447,6.基金项目
国家社会科学基金一般项目:数字时代下青少年心理健康问题的形成机制、评估系统和网络干预研究(21BSH017) （21BSH017）
天津市教委科研计划专项任务项目:基于机器学习的大学生自杀行为的分类与识别(2021ZDGX03) （2021ZDGX03）
北京智源人工智能研究院面向可持续发展的公益研究计划(AI4SDGs). （AI4SDGs）
