Identifying suicidal subtypes and dynamic indicators of increasing and decreasing suicide risk

摘要 过去 20 年来，美国总人口自杀率稳步上升。那些有 在美国武装部队服役的人员是高风险人群，其中发病率增长速度较快 与那些从未参军的人相比。新兴研究表明存在 自杀状态的几种亚型。不同亚型的个体可能会遵循不同的高风险途径 状态并可能以不同的方式对治疗干预做出反应。迄今为止，研究尚未检验 使用包括遗传、环境、医学和心理变量在内的综合数据集的类型学。 为了解决这一知识差距，我们建议利用南德克萨斯地区的存档数据集 创伤和复原力组织网络指导研究（STRONG STAR）存储库，其中包含 来自 4000 多名军事人员的遗传、环境、医疗和心理变量 部署前和部署后进行评估。使用此数据集，我们将 (a) 识别自杀式军事人员的亚组 (b) 确定自杀风险增加、减少和静态的不同模式。结果 分析将使我们能够识别自杀风险的离散基因型-表型表达，从而使我们能够 确定可用于改进风险检测和完善自杀预防的多种风险模型 干预措施。 新兴研究进一步表明，自杀风险随时间的变化本质上是非线性的。不幸的是， 大多数研究随着时间的推移自杀风险的出现都采用了研究和数据 分析方法无法准确捕捉非线性变化过程。为了解决这个问题 为了弥补知识差距，我们建议利用 STRONG 中包含的六项临床试验的存档数据集 STAR 存储库（N>800），每个存储库都包含抑郁症的重复评估（总共最多 13 次）， 创伤后应激障碍（PTSD）和自杀意念。以动力系统理论为基础的多变量潜在变化评分模型， 将用于对与低风险和高风险状态相关的非线性变化过程进行建模。结果 分析将产生后验概率，可以估计给定患者转变为高水平的可能性 给定时间点的风险状态，这可能会导致开发“警告系统”来识别谁 随着时间的推移和何时，风险会增加。 尽管拟议的研究使用从军事人员收集的存档数据，但拟议的方法可以 被转移到其他人群和环境中，从而导致检测和检测方面取得重大进展 自杀风险较高的人。

ABSTRACT The U.S. general population suicide rate has increased steadily over the past 20 years. Those who have served in the U.S. Armed Forces are a high risk subgroup among which rates have increased at a faster rate as compared to those who have never served in the military. Emerging research suggests the existence of several subtypes of suicidal states. Individuals in different subtypes may follow different pathways to high risk states and may respond to treatment interventions in different ways. To date, studies have not examined typologies using integrated datasets that include genetic, environmental, medical, and psychological variables. To address this knowledge gap, we propose to leverage an archived dataset from the South Texas Region Organization Network Guiding Studies of Trauma and Resilience (STRONG STAR) Repository, which contains genetic, environmental, medical, and psychological variables from over 4000 military personnel who were assessed before and after deployment. Using this dataset, we will (a) identify subgroups of suicidal military personnel and (b) identify different patterns of increasing, decreasing, and static suicide risk. Results of this analysis will enable us to identify discrete genotype-phenotype expressions of suicide risk, thereby enabling us to identify multiple risk models that can be used to improve risk detection and refine suicide prevention interventions. Emerging research further indicates the process of suicide risk over time is nonlinear in nature. Unfortunately, the majority of studies examining the emergence of suicide risk over time have employed research and data analytic methods that are unable to accurately capture nonlinear change processes. To address this knowledge gap, we propose to leverage archived datasets from six clinical trials included in the STRONG STAR Repository (N>800), each of which includes repeated assessments (up to 13 total) of depression, PTSD, and suicide ideation. Multivariate latent change score models, informed by dynamical systems theory, will be used to model nonlinear change processes associated with low risk and high risk states. Results of this analysis will yield posterior probabilities that can estimate the likelihood of a given patient transitioning to a high risk state at a given point in time, which could lead to the development of “warning systems” that identify who will experience increased risk over time, and when. Although the proposed study uses archived data collected from military personnel, the proposed methods can be translated to other populations and settings, thereby leading to significant advances in the detection and individuals with elevated risk for suicide.