Integrative computational models of latent behavioral and neural constructs in children: a longitudinal developmental big-data approach

项目摘要 认知系统中调节适应和应对变化的能力的损害 刺激和目标是精神病理学的一个标志。确定潜在的认知和神经因素， 驾驶功能障碍的行为动力学是精神病学研究的主要目标。无论多么传统 方法无法揭示控制这些动态过程的潜在构造。新颖的计算方式 需要方法来揭示潜在的行为动力学和与精神病理学相关的特征，以及 他们的神经回路基础，在研究领域标准(RDoC)框架内。大多数，如果不是全部的话，精神病 障碍源于神经发育，并与认知大脑的非典型成熟有关 网络。认知是一个动态的过程，它依赖于灵活的抑制控制和目标导向的信念 这种影响时时刻刻的预期，以及从先前的决定中学习和适应的能力。 在精神病理学的背景下，开发认知的动态潜在行为模型是重要的， 由于抑制控制、绩效监控和信念更新方面的缺陷与多个 精神障碍包括多动症、自闭症和精神分裂症。我们的首要目标是发展和 一种新的一体化计算方法--分层潜变量动力学(HLVD)的验证 发现健壮的潜在行为结构及其神经回路基础。拟议的研究将 利用纵向青少年行为和认知发展(ABCD)研究，该研究具有 产生了空前数量的“大数据”(N&GT；5,000)，用于绘制认知和大脑发展图表 随着时间的推移，儿童和青少年。至关重要的是，HLVD将用于识别和验证新的潜在结构 被认为是外化症状的重要维度预测因素的行为动力学 和发育性精神病理学。建议的研究将大大加深我们对 RDoC构建并提供对潜在行为动力学和与以下各项相关的特征的新见解 发育中的大脑中的精神病理学。我们的研究与NIMH倡议的使命高度相关 RFA-MH-19-242，旨在加快对神经发育和精神疾病风险轨迹的研究 生病了。我们的创新方法最终将有助于开发生物标记物，用于早期检测和 治疗精神疾病。

Project Abstract Impairments in cognitive systems that regulate the ability to adaptively engage with and respond to changing stimuli and goals are a hallmark of psychopathology. Identifying the underlying cognitive and neural factors that drive dysfunctional behavioral dynamics is a primary goal for psychiatric research. However conventional methods are unable to reveal latent constructs that govern these dynamic processes. Novel computational approaches are required to reveal latent behavioral dynamics and traits associated with psychopathology, and their neural circuit basis, within the Research Domain Criteria (RDoC) framework. Most, if not all, psychiatric disorders have a neurodevelopmental origin and are associated with atypical maturation of cognitive brain networks. Cognition is a dynamic process, which relies on flexible inhibitory control, goal-directed beliefs that impact moment-to-moment expectation, and the capacity to learn and adapt from prior decisions. Developing dynamic latent behavioral models of cognition is significant in the context of psychopathology, because deficits in inhibitory control, performance monitoring and belief updating are implicated in multiple psychiatric disorders including ADHD, autism, and schizophrenia. Our overarching goal is to develop and validate Hierarchical Latent Variable Dynamics (HLVD), a novel integrative computational approach for discovering robust latent behavioral constructs and their neural circuit bases. The proposed studies will leverage the longitudinal Adolescent Behavioral and Cognitive Development (ABCD) study, which has generated unprecedented amounts of “Big Data” (N>5,000) for charting cognitive and brain development in children and adolescents over time. Crucially, HLVD will be used to identify and validate novel latent constructs of behavioral dynamics that are expected to be significant dimensional predictors of externalizing symptoms and developmental psychopathology. The proposed studies will significantly enhance our understanding of RDoC constructs and provide new insights into latent behavioral dynamics and traits associated with psychopathology in the developing brain. Our studies are highly relevant to the mission of the NIMH initiative RFA-MH-19-242, which seeks to accelerate research on neurodevelopment and trajectories of risk for mental illness. Our innovative approach will ultimately aid in the development of biomarkers for early detection and treatment of psychiatric disorders.