MAPS: Mobile Assessment for the Prediction of Suicide

项目摘要 自杀是青少年死亡的第二大原因。除了死亡，16% 青少年报告说，他们每年都认真考虑自杀，8%的人有一次或多次尝试。尽管有这些 令人震惊的统计数字，很少有人知道的因素，赋予迫在眉睫的自杀风险。因此，发展有效 改善自杀想法和行为（STBs）短期预测的方法至关重要。 目前，我们最可靠的STB预测指标是人口统计学或临床指标， 预测值较低。然而，有一个关于短期预测自杀风险的新兴文献， 确定了一些有希望的候选人，包括快速升级：（a）情绪困扰，（B）社会 功能障碍（即，欺凌，拒绝），和（c）睡眠障碍。然而，先前的研究在两个关键方面受到限制。 首先，他们几乎完全依赖自我报告。其次，大多数研究都没有集中在这些评估上， 风险因素，使用密集的纵向评估技术，能够捕捉动态 风险状态的变化。这些是基本的限制。虽然自杀意念可能先于 自杀未遂前的社会情绪变化通常发生在几分钟内， 小时这项研究将利用实时监测方法的最新发展， 青少年对智能手机技术的自然使用。具体而言，我们现在有能力用途：（a） 智能手机技术进行密集的纵向评估，监测假定的风险因素， 最小的参与者负担和（B）现代计算技术，以开发用于STB的预测算法。 该项目将包括从门诊招募的13-18岁的高危青少年（n = 200）， 住院患者：（a）近期有自杀意念的自杀者（n = 70），（B）目前没有自杀意念的自杀者（n = 70）， （c）无STB病史的精神病对照组（n = 60）。轻松评估 风险状态（EARS）将用于持续测量与关键风险领域相关的变量， 痛苦，社交功能障碍和睡眠障碍-通过被动监测参与者的智能手机 使用.首先，我们将在最初的2周内测试组间风险因素的差异，并确定 来自移动的电话的风险因素在多大程度上改善了对自我报告的 指标其次，我们将使用统计技术来测试风险因素是否在短期内得到改善 STB的预测（例如，自杀企图、住院治疗），以及 超越临床评估。第三，计算机器学习技术-基于先验和 将开发利用全方位密集纵向数据的预测模型 通过主动和被动监测方法收集的数据来预测组成员和STB结果。 最终，通过利用智能手机技术，我们的目标是改善短期STB预测，并提供 为临床医生和患者提供可靠、可扩展和可操作的工具，减少不必要的生命损失。

Project Summary Suicide is the second leading cause of death among adolescents. In addition to deaths, 16% of adolescents report seriously considering suicide each year, and 8% make one or more attempts. Despite these alarming statistics, little is known about factors that confer imminent risk for suicide. Thus, developing effective methods to improve short-term prediction of suicidal thoughts and behaviors (STBs) is critical. Currently, our most robust predictors of STBs are demographic or clinical indicators that have relatively weak predictive value. However, there is an emerging literature on short-term prediction of suicide risk that has identified a number of promising candidates, including rapid escalation of: (a) emotional distress, (b) social dysfunction (i.e., bullying, rejection), and (c) sleep disturbance. Yet, prior studies are limited in two critical ways. First, they rely almost entirely on self-report. Second, most studies have not focused on assessment of these risk factors using intensive longitudinal assessment techniques that are able to capture the dynamics of changes in risk states. These are fundamental limitations. While suicidal ideation may precede an attempt by years, socio-emotional changes preceding a suicide attempt often occurs within the time span of minutes to hours. This study will capitalize on recent developments in real-time monitoring methods that harness adolescents' naturalistic use of smartphone technology. Specifically, we now have the capacity to use: (a) smartphone technology to conduct intensive longitudinal assessments monitoring putative risk factors with minimal participant burden and (b) modern computational techniques to develop predictive algorithms for STBs. The project will include high-risk adolescents (n = 200) aged 13-18 years recruited from outpatient and inpatient clinics: (a) recent suicide attempters with current ideation (n = 70), (b) current suicide ideators with no attempt history (n = 70), and (c) a psychiatric control group with no STB history (n = 60). Effortless Assessment of Risk States (EARS) will be used to continuously measure variables relevant to key risk domains—emotional distress, social dysfunction, and sleep disturbance—through passive monitoring of participants' smartphone use. First, we will test between-group differences in risk factors during an initial 2-week period, and determine the extent to which risk factors derived from mobile phones improves discrimination over self-reported indicators. Second, we will use statistical techniques to test whether the risk factors improve short-term prediction of STBs (e.g., suicide attempts, hospitalization) during the 6-month follow-up period above and beyond clinical assessments. Third, computational machine learning techniques—based on a priori and learned features—will develop predictive models that utilize the full range of intensive longitudinal data collected by the active and passive monitoring methods to predict group membership and STB outcomes. Ultimately, by leveraging smartphone technology, we aim to improve the short-term STB prediction and provide clinicians and patients with reliable, scalable and actionable tools that will reduce the needless loss of life.