Smartphone-based digital phenotyping to detect high-risk affect states in body dysmorphic disorder

躯体变形障碍(BDD)与极高的自杀企图风险(22%-28%)有关 物质使用障碍(49%)，强调了BDD风险检测的关键重要性。负面影响 国家--尤其是焦虑和羞耻--是有充分证据的自杀和药物使用的危险因素 BDD，为风险检测和干预提供明确的目标。此K23旨在开发和验证 对焦虑、羞耻和一般负面情绪的不显眼、时间敏感和生态上有效的衡量标准 BDD中的州，使用基于智能手机的数字表型。被动(即，不显眼)智能手机 负面情绪状态的测量将基于GPS、加速计和通信日志，用于 检测焦虑(回避、仪式)、羞耻(社交退缩、孤立)和一般的行为特征 负面情绪(集中的回避、仪式、退缩和孤立特征)。我们将收集被动和 85名患有BDD的成年人的活跃(即生态瞬时评估[EMA])智能手机数据，并将使用 EMA将负面情绪作为结果进行评级，以构建和验证被动的预测性统计模型 数据。我们还将测试以下假设：被动智能手机对负面影响状态的测量可以 显著预测BDD患者第二天的自杀意念和物质使用，高于普通临床 风险指数。该项目综合了候选人在BDD中基于情绪的自杀风险方面的专业知识和 她从事智能手机研究的经验。在此基础上，这款K23将提供关键的 在关键领域开展新的培训，以启动候选人的独立研究生涯：(1)数字表型， 包括统计学习和纵向分析；(2)EMA方法；(3)自杀评估和 物质使用；(4)职业发展，包括R01写作；和(5)基于技术的自杀和 物质使用研究。培训目标将在出色的指导和机构支持下实现 马萨诸塞州总医院和哈佛医学院。Sabine Wilhelm博士，BDD和 临床研究，将担任首要导师。Jukka-Pekka Onnela博士，数字表型专家 和它的统计方法，以及EMA情绪和自杀研究专家迈克尔·阿梅博士将 担任共同导师。该咨询将在EMA和物质使用方面提供补充指导 团队：Bettina Hoeppner博士和A.Eden Evins博士。根据NIMH战略目标2，这款K23将产生 可扩展、不显眼的工具，用于检测高危人群中自杀和药物使用的急性、可修改的风险因素 人口。此外，消极情绪状态也是跨诊断风险因素。作为下一步的证据- 概念K23，候选人将申请R01，以进一步验证被动移动检测的负面影响 国家及其跨诊断预测风险的能力。这项研究计划可以使(1)个性化 针对高危影响州的及时干预，以减少自杀和药物使用；(2)不引人注目 监测风险的变化；以及(3)对风险过程进行大规模、生态有效的纵向研究。

Body dysmorphic disorder (BDD) is associated with extremely high risk for suicide attempts (22-28%) and substance use disorders (49%), underscoring the critical importance of risk detection in BDD. Negative affect states - particularly anxiety and shame - are well-documented risk factors for suicide and substance use in BDD, offering clear targets for risk detection and intervention. This K23 aims to develop and validate unobtrusive, time-sensitive, and ecologically valid measures of anxiety, shame, and general negative affect states in BDD, using smartphone-based digital phenotyping. Passive (i.e., unobtrusive) smartphone measurement of negative affect states will be based on GPS, accelerometer, and communication logs, used to detect behavioral features of anxiety (avoidance, rituals), shame (social withdrawal, isolation), and general negative affect (aggregated avoidance, rituals, withdrawal, and isolation features). We will collect passive and active (i.e., ecological momentary assessment [EMA]) smartphone data in 85 adults with BDD and will use EMA ratings of negative affect as outcomes, to build and validate predictive statistical models from passive data. We will also test the hypotheses that passive smartphone measures of negative affect states can significantly predict next-day suicidal ideation and substance use in BDD, above and beyond common clinical indices of risk. This project synthesizes the Candidate’s expertise in emotion-based risk for suicide in BDD with her experience conducting smartphone research. Building from this foundation, this K23 will provide critical new training in key areas to launch the Candidate’s independent research career: (1) digital phenotyping, including statistical learning and longitudinal analysis; (2) EMA methods; (3) assessment of suicide and substance use; (4) career development, including R01 writing; and (5) ethics of technology-based suicide and substance use research. Training goals will be accomplished with stellar mentorship and institutional support at Massachusetts General Hospital and Harvard Medical School. Dr. Sabine Wilhelm, a leader in BDD and clinical research, will serve as the primary mentor. Dr. Jukka-Pekka Onnela, an expert in digital phenotyping and its statistical approaches, and Dr. Michael Armey, an expert in EMA research of emotions and suicide, will serve as co-mentors. Complementary guidance in EMA and substance use will be provided by the advisory team: Drs. Bettina Hoeppner and A. Eden Evins. In line with NIMH Strategic Objective 2, this K23 will yield scalable, unobtrusive tools to detect acute, modifiable risk factors for suicide and substance use in a high-risk population. Moreover, negative affect states are transdiagnostic risk factors. As a next step to this proof-of- concept K23, the Candidate will apply for an R01 to further validate passive mobile detection of negative affect states and their ability to predict risk transdiagnostically. This program of research can enable (1) personalized just-in-time interventions targeting high-risk affect states, to reduce suicide and substance use; (2) unobtrusive monitoring of changes in risk; and (3) large-scale, ecologically-valid longitudinal research of risk processes.