This study aimed to determine conditional dependence relationships of variables that contribute to psychological vulnerability associated with suicide risk. A Bayesian network (BN) was developed and applied to establish conditional dependence relationships among variables for each individual subject studied. These conditional dependencies represented the different states that patients could experience in relation to suicidal behavior (SB). The clinical sample included 650 mental health patients with mood and anxiety symptomatology. Mainly indicated that variables within the Bayesian network are part of each patient’s state of psychological vulnerability and have the potential to impact such states and that these variables coexist and are relatively stable over time. These results have enabled us to offer a tool to detect states of psychological vulnerability associated with suicide risk. If we accept that suicidal behaviors (vulnerability, ideation, and suicidal attempts) exist in constant change and are unstable, we can investigate what individuals experience at specific moments to become better able to intervene in a timely manner to prevent such behaviors. Future testing of the tool developed in this study is needed, not only in specialized mental health environments but also in other environments with high rates of mental illness, such as primary healthcare facilities and educational institutions.

中文翻译：

---

识别自杀行为的心理脆弱性状态：应用于临床样本的人工智能贝叶斯网络。


本研究旨在确定导致与自杀风险相关的心理脆弱性的变量的条件依赖关系。开发并应用贝叶斯网络（BN）来为每个研究对象的变量之间建立条件依赖关系。这些条件依赖性代表了患者可能经历的与自杀行为（SB）相关的不同状态。临床样本包括 650 名具有情绪和焦虑症状的心理健康患者。主要表明贝叶斯网络内的变量是每个患者心理脆弱状态的一部分，并且有可能影响这种状态，并且这些变量共存并且随着时间的推移相对稳定。这些结果使我们能够提供一种工具来检测与自杀风险相关的心理脆弱状态。如果我们承认自杀行为（脆弱性、意念和自杀企图）存在不断变化且不稳定，我们就可以调查个人在特定时刻的经历，以便更好地及时干预以防止此类行为。未来需要对本研究中开发的工具进行测试，不仅要在专门的心理健康环境中进行测试，还要在其他精神疾病高发环境中进行测试，例如初级医疗机构和教育机构。