Improving momentary suicide risk identification through adaptive time sampling

通过自适应时间采样提高瞬时自杀风险识别

• 批准号: 10575138
• 负责人: Brooke A Ammerman
• 金额: $ 23.48万
• 依托单位: UNIVERSITY OF NOTRE DAME
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-12-16 至 2024-11-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10575138
• 关键词: Acute; Address; Algorithms; Benchmarking; Calibration; Data; Data Analyses; Data Collection; Detection; Development; Ecological momentary assessment; Feeling suicidal; Friends; Health; Heterogeneity; Hour; Individual; Intervention; Knowledge; Literature; Machine Learning; Mental Health; Methods; Mission; Modeling; National Institute of Mental Health; Outcome; Participant; Performance; Persons; Phase; Public Health; Reporting; Research; Resolution; Risk; Risk Factors; Sampling; Schedule; Scheme; Statistical Models; Suicide; Suicide prevention; System; Time; Training; United States National Institutes of Health; Update; Validation; Variant; Work; adaptive intervention; algorithm training; arm; assessment application; data streams; effective intervention; experience; high risk; implementation intervention; improved; innovation; insight; mortality; novel; personalized predictions; prospective; reducing suicide; response; societal costs; suicidal risk; suicide rate; web app

PROJECT SUMMARY/ABSTRACT Suicide rates have risen sharply over the past 20 years1. There is a need to more precisely identify proximal risk indicators for the development of near-term suicide risk in order to effectively intervene. Studies utilizing ecological momentary assessment (EMA) to collect data at several intervals per day have demonstrated that suicidal ideation (SI) and proximal risk factors change rapidly across the course of the day2. However, prior EMA studies examining SI dynamics implement stable assessments, with intervals of several hours between SI assessments across the duration of a study period3 for all participants. This one-size-fits-all approach to SI assessment fails to capture the nuanced within-person variability of the timescale of the development of acute suicide risk. In turn, we lack even a basic understanding of within-person variability in the time varying relationship between SI and its proximal risk factors. The proposed study aims to address the limitations of current assessment approaches in proximal suicide risk research through the development of a personalized, adaptive time sampling system. The specific objectives of the proposed research are to: (1) develop a novel, adaptive time assessment system that more efficiently and accurately identifies when an individual is at highest risk for SI; and (2) advance the understanding of SI and its theoretically-informed proximal risk factors at finer timescales. Data collected according to varied timing schedules in the first phase will be used to train an algorithm that generates predictions of suicide risk, predictions that will be adaptively use to determine assessment timing during the second phase of data collection. Aim 1 is to develop the adaptive time assessment system, followed by assessing the predictive accuracy of the adaptive sampling system (Aim 2) and identifying variations in person-specific effects of the relationship between SI and theoretically-informed risk factors (Aim 3). The research team (PIs: Ammerman, Jacobucci; Co-I: Cheng; Consultants: Burke) has access to world-class expertise, with extensive experience in EMA data collection in high-risk samples, machine learning for suicide prediction, longitudinal data analysis, collecting and modeling continuous data streams, and the development of adaptive assessment platforms. To meaningfully reduce suicide rates, a more nuanced understanding of suicidal thoughts and associated risk factors is required. Our adaptive assessment platform will more efficiently assess suicidal thoughts and risk factors, allowing for a closer approximation of the true associations. Indeed, there is a need to identify near-term risk factors prior to suicidal thought occurrences to successfully deliver an intervention and prevent suicidal outcomes. These findings will support the successful implementation of just-in-time adaptive interventions through increased precision of suicide risk detection and targeted intervention timing. Given the grave personal and societal cost of suicide, this work has important public health implications.

项目摘要/摘要 在过去的20年里，自杀率急剧上升。有必要更准确地识别 制定近期自杀风险指标，以便有效干预。研究 利用生态瞬时评估(EMA)每天在几个时间间隔收集数据已经证明 自杀意念(SI)和近端风险因素在一天中变化迅速。然而，之前 检查SI动态的EMA研究实施了稳定的评估，SI之间的间隔为几个小时 在整个研究期间对所有参与者进行评估3。这种一刀切的SI方法 评估未能捕捉到急性胰腺炎发生的时间尺度的细微差别 自杀风险。反过来，我们甚至对人内在时间变化中的可变性缺乏基本的理解 SI与其近端危险因素的关系。 这项拟议的研究旨在解决目前近端自杀评估方法的局限性。 通过开发个性化、自适应的时间采样系统进行风险研究。具体目标 建议研究的目的是：(1)开发一种新的、自适应的时间评估系统，该系统可以更有效地和 准确识别一个人何时是SI的最高风险者；以及(2)促进对SI及其 在更精细的时间尺度上从理论上了解近端风险因素。根据不同的时间收集的数据 第一阶段的时间表将用于训练一个算法，该算法生成自杀风险预测、预测 这将在数据收集的第二阶段自适应地用来确定评估时间。目标1是 开发自适应时间评估系统，然后评估自适应时间的预测精度 抽样系统(目标2)，并确定SI和SI之间关系的特定于人的影响的差异 理论上知情的风险因素(目标3)。 研究团队(PI：Ammerman，Jacobucci；Co-I：Cheng；Consulters：Burke)拥有世界级的 专业知识，在高危样本的EMA数据收集、自杀的机器学习方面拥有丰富经验 预测、纵向数据分析、连续数据流的收集和建模，以及 适应性评估平台。 为了有意义地降低自杀率，对自杀想法和相关的更细微的理解 风险因素是必需的。我们的适应性评估平台将更有效地评估自杀想法和风险 因素，允许更接近真实的关联。事实上，有必要确定近期目标。 自杀念头发生前的风险因素，以成功实施干预并预防自杀 结果。这些发现将支持适时适应性干预措施的成功实施 通过提高自杀风险检测的准确性和有针对性的干预时机。考虑到严重的个人 和自杀的社会成本，这项工作具有重要的公共卫生影响。