Frontal-thalamo-cerebellar circuitry of attention deficit via imaging-genetic-environmental analyses

摘要/总结： 注意力缺陷（AD）是精神健康和神经系统疾病的一个报告问题。它作为一种极端存在 在一般人群中连续分布的特征。AD作为ADHD的一个重要组成部分， 在多个神经认知领域，特别是在注意力/警惕性，工作记忆，处理 速度和响应可变性。迄今为止，对AD的研究大多集中在额顶环的研究上，而对其了解较少 关于额丘脑小脑回路（FCC）与AD的关系。为了扩展我们对AD神经机制的了解， 本研究旨在描述FCC改变与神经认知和AD症状的关系， ABCD队列中的成像数据、基因组学、神经认知、环境数据。首先，在目标1中，我们将应用高级深度 学习算法以建模FCC（灰质、白色物质、静息）中的多模态脑图像数据之间的关系 状态fMRI功能连接，和情绪N-back任务fMRI激活）和神经认知措施，在四个 基线时的领域（注意力/警惕性、工作记忆、处理速度和反应变异性）。然后我们将 将迁移学习技术应用于神经认知的潜在神经成像特征以估计AD。然后， 在目标2中，我们将关注FCC神经影像学特征的纵向变化与 神经认知和AD的关系我们将应用先进的机器学习方法，就像在目标1中识别FCC一样。 神经认知纵向变化的动态特征，进而转化为AD。AD症状和体征 这些变化还受到遗传特征和环境因素的影响。在目标3中，我们将应用多元数据挖掘 算法来提取与FCC神经成像特征相关的遗传因素，并建立AD的预测模型， 除了FCC之外，还使用提取的遗传因素、社会人口统计学和环境因素来改变AD 多模态神经影像学特征。最后，在目标4中，我们将使用Year验证FCC-遗传-环境-AD模型 4.在ABCD队列中随访数据，并使用独立的PNC队列验证FCC-遗传-AD模型。这些发现 从这项研究将详细说明FCC的关键区域的变化，每个神经认知领域和贡献 每个神经认知结构域对FCC神经元特征介导的AD症状的影响。脑、基因和环境模型 将有助于识别由于一般人群中FCC改变而具有AD风险的亚群，并有助于指定 由于FCC异常而有AD恶化风险的精神障碍患者。

Abstract/Summary: Attention deficit (AD) is a reported concern across mental health and neurological disorders. It exists as an extreme condition of a continuously distributed trait in the general population. AD as a key component of ADHD is often associated with impairments in multiple neurocognitive domains, particularly in attention/vigilance, working memory, processing speed, and response variability. To date, most investigations on AD focus on frontal-parietal circuity, and less is known about the frontal-thalamo-cerebellar circuitry (FCC) relates to AD. To extend our knowledge on neural mechanisms of AD, this study aims to delineate FCC alteration in relation to neurocognition and AD symptoms, leveraging longitudinal brain imaging data, genomics, neurocognition, environmental data in ABCD cohort. First, in Aim 1 we will apply advanced deep learning algorithms to model the relationship between multimodal brain image data in FCC (gray matter, white matter, rest state fMRI functional connectivity, and emotional N-back task fMRI activation) and neurocognitive measures in the four domains (attention/vigilance, working memory, processing speed, and response variability) at baseline. And then we will apply the transfer learning techniques to the latent neuroimaging features underlying neurocognition to estimate AD. Then, in Aim 2 we will focus on the relation between longitudinal changes of FCC neuroimaging features and the changes in neurocognition and AD in two years. We will apply advanced machine learning methods just as in Aim 1 to identify FCC dynamic features underlying longitudinal changes in neurocognition, and then transfer to AD. AD symptoms and symptom changes are also affected by genetic profiles and environmental factors. In Aim 3 we will apply multivariate data mining algorithms to extract genetic factors associated with FCC neuroimaging features, and build a prediction model for AD and changes of AD using genetic factors extracted, social demographic and environmental factors, in addition to FCC multimodal neuroimaging features. Lastly, in Aim 4 we will validate the FCC-genetic-environmental-AD model using Year 4 follow up data in ABCD cohort and validate the FCC-genetic-AD model using an independent PNC cohort. The findings from this study will specify alterations in crucial regions of FCC underlying each neurocognition domain and contribution of each neurocognition domain to AD symptoms mediated by FCC neuronal features. Brain, gene, and environmental model of AD will help identify a subpopulation with risk for AD due to FCC alterations in the general population, and help specify patients across the boundaries of mental disorders who are risk for worsening AD due to FCC abnormalities.