Peripartum Depression Prevention: Algorithmic Identification and Digital Solutions

围产期抑郁症预防：算法识别和数字解决方案

• 批准号: 10523267
• 负责人: Tamar Krishnamurti
• 金额: $ 15.98万
• 依托单位: UNIVERSITY OF PITTSBURGH AT PITTSBURGH
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-08-15 至 2025-06-30
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10523267
• 关键词: Address; Adherence; Advisory Committees; Affect; Algorithms; Artificial Intelligence; Behavior Control; Black Populations; Black race; Caring; Child; Childbirth; Clinical; Cognitive Therapy; Data; Data Set; Detection; Development; Developmental Delay Disorders; Diagnosis; Discrimination; Early Intervention; Effectiveness; Electronic Health Record; Enrollment; Ensure; Event; Evidence based treatment; Failure to Thrive; Feedback; First Pregnancy Trimester; Frequencies; Funding; Future; Health Services; Incidence; Individual; Infant; Intervention; Light; Machine Learning; Measures; Medical center; Mental Depression; Mental Health Services; Methods; Modeling; Mothers; National Institute of Mental Health; Nurse Practitioners; Outcome Measure; Output; Participant; Pathway interactions; Patients; Perinatal; Persons; Physicians; Population; Population Heterogeneity; Postpartum Period; Pregnancy; Premature Birth; Prevention; Prevention approach; Prevention strategy; Preventive service; Process; Provider; Race; Randomized; Randomized Controlled Trials; Recommendation; Recording of previous events; Risk; Risk Factors; Services; Societies; Time; Touch sensation; United States; Universities; Woman; Work; acceptability and feasibility; antepartum depression; arm; artificial intelligence algorithm; base; behavioral health; clinical practice; community engagement; cost; depression prevention; digital; digital mental health; evidence base; experience; feasibility testing; graph learning; human centered design; implementation framework; improved; individual patient; innovation; insight; large datasets; learning algorithm; maternal depression; patient population; peripartum depression; prediction algorithm; pregnant; primary outcome; prospective; racial difference; recruit; risk prediction; routine care; standard of care; tool; treatment as usual; uptake; usability

Project Summary/Abstract Background: Depression during pregnancy and the postpartum period affects up to 15% of US mothers, imposing costs on mother, child, and society. Significantly more Black individuals meet the criteria for depression than white individuals in the US, yet they are less likely to receive mental health care, highlighting disparities in diagnosis and treatment. A United States Preventive Services Task Force recommendation suggests that pregnant people at risk for depression be proactively engaged in behavioral health services. For any depression prevention approach to be scalable and sustainable, those at risk of depression must be accurately identified prior to depression onset and subsequently connected to an evidence-based treatment that is feasible, acceptable, and usable. Methods Aim 1 will apply the PC Kernel Conditional Independence algorithm to two large prospective observational datasets. The output will be models of the potential causal pathways of perinatal depression onset that can be used to predict individual patient depression risk based on factors that can 1) be queried from the existing electronic health record (EHR), and 2) can be combined with EHR data to more precisely predict subsequent maternal depression above and beyond standard of care screeners. This will create a minimal set of data needed for risk prediction and intervention. Aim 2 will convene an expert panel of 5 OB/GYN providers and two community engagement studios, one with Black pregnant individuals and one with white pregnant individuals, to obtain feedback on the implementation needs and acceptability of the risk prediction algorithms. Aim 3 will randomize 60 participants (50% Black individuals) who are at- risk for future depression in the first trimester of pregnancy to a digital CBT (treatment) or usual care (control), using a non-traditional, highly scalable approach to trial enrollment. Potential Impact: This work will identify the data and processes necessary for a subsequent randomized controlled trial of an acceptable, scalable, and largely digital strategy for depression detection and prevention among pregnant people. It is highly innovative as it will be the first study of its kind to identify risk prior to depression onset using a scalable approach to engaging a diverse population of pregnant patients paired with digital mental health care provision, incorporating the perspectives and experiences of the patient population for whom the model is serving into the development process.

项目摘要/摘要 背景：妊娠期和产后期抑郁影响高达15%的美国人 母亲，给母亲、孩子和社会带来了代价。明显更多的黑人遇到 在美国，抑郁症的标准比白人更低，但他们患抑郁症的可能性更小 精神卫生保健，强调在诊断和治疗方面的差异。一个美国 预防服务工作组的建议建议，孕妇有患上 抑郁症患者应积极从事行为健康服务。对于任何抑郁症的预防 为了具有可伸缩性和可持续性，必须准确地识别那些有抑郁风险的人 在抑郁症发作之前，随后与循证治疗有关，这是 可行、可接受和可用的。 方法目标1将PC内核条件独立算法应用于两个较大的 预期观测数据集。输出将是潜在的因果路径的模型 围产期抑郁发作可用于预测个体患者的抑郁风险 可以从现有的电子健康档案(EHR)中查询的因素，以及2)可以 结合EHR数据，更准确地预测随后的母亲抑郁 超出标准的关怀筛查员。这将创建风险预测所需的最小数据集 和干预。AIM 2将召集由5个OB/GYN提供商和两个社区组成的专家小组 订婚工作室，一个是黑人孕妇，另一个是白人孕妇 个人，以获得关于实施需求和风险可接受性的反馈 预测算法。AIM 3将随机选择60名参与者(50%为黑人)，他们的年龄- 怀孕前三个月接受数字CBT(治疗)或常规治疗的未来抑郁风险 关怀(控制)，使用非传统、高度可扩展的方法进行试验登记。 潜在影响：这项工作将确定后续 一种可接受的、可扩展的、主要数字化的抑郁症治疗策略的随机对照试验 孕妇的检测和预防。这项研究具有很高的创新性，因为这将是第一项研究 使用可扩展的方法在抑郁症发作之前识别风险 多样化的怀孕患者群体与数字精神卫生保健服务相匹配， 结合模型所针对的患者群体的观点和经验 服务于开发过程。