Evaluation of neurobiological models of adolescent maltreatment through machine learning

通过机器学习评估青少年虐待的神经生物学模型

• 批准号: 9108475
• 负责人: Matthew Price
• 金额: $ 19.1万
• 依托单位: UNIVERSITY OF VERMONT & ST AGRIC COLLEGE
• 依托单位国家: 美国
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-06-01 至 2020-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/9108475
• 关键词: Adolescence; Adolescent; Affect; Age; Amygdaloid structure; Anhedonia; Anterior; Area; Base of the Brain; Behavior; Behavioral; Biological Neural Networks; Brain; Brain region; Child Abuse; Child Abuse and Neglect; Childhood; Chronic; Classification; Clinical Psychology; Collection; Complex; Computational Science; Computing Methodologies; Corpus striatum structure; Data; Databases; Development; Diagnosis; Diagnostic; Discipline; Disease; E-learning; Engineering; Evaluation; Functional Imaging; Functional Magnetic Resonance Imaging; Funding; Future; Genetic Programming; Goals; Hippocampus (Brain); Image; Impairment; Intervention; Investigation; Knowledge; Longevity; Machine Learning; Manuscripts; Measures; Mental Depression; Mental disorders; Mentored Clinical Scientist Development Award (K08); Mentors; Mentorship; Methods; Modeling; National Institute of Mental Health; Neurobiology; Neurosciences; Nucleus Accumbens; Participant; Positioning Attribute; Prefrontal Cortex; Preparation; Psychopathology; Public Health; Recording of previous events; Reporting; Research; Research Ethics; Rewards; Role; Sampling; Series; Severities; Societies; Stress; Structure; Symptoms; System; Testing; Theoretical model; Therapeutic; Training; Translating; United States National Institutes of Health; Variant; base; career; cingulate cortex; cognitive control; computer science; conditioned fear; experience; gray matter; imaging probe; improved; learning strategy; maltreatment; multidisciplinary; neuroimaging; neuromechanism; novel therapeutics; programs; public health relevance; putamen; research study; response; reward processing; young adult

 DESCRIPTION (provided by applicant): The goal of this Mentored Clinical Scientist Development Award (K08) is to provide the candidate with extensive training in 1) machine learning, 2) neuroimaging and 3) developmental psychopathology to conduct independent investigations on the neurobiology that underlies psychopathology that results from childhood maltreatment. There is strong evidence that maltreatment during childhood results in differences across multiple circuits in the developing brain. The extent that affected circuits are involved in psychopathology, however, is unclear. Determining which circuits are involved is necessary to develop interventions that target specific neurobiological systems. The primary aims of the proposed studies are to clarify the involvement of multiple brain features (e.g., structures and functional activations) in maltreatment-related psychopathology. These aims will be accomplished with machine learning. Machine learning is a set of computer-based learning methods in which meaningful theoretical models are derived from empirical data. This proposal builds on the candidate's prior training in clinical psychology, neurobiological models of traumatic stress, and advanced computational methods to develop expertise in: 1) computational methods that include machine learning, genetic algorithms, and artificial neural networks, 2) the collection of multi-modal neuroimaging data and 3) the effect of maltreatment on development. Additional professional development training will include grantsmanship, manuscript preparation, and research ethics. To achieve these aims, the candidate has assembled an accomplished mentorship team of experts that span a range of disciplines including developmental psychopathology, computational science, and neuroscience. The training and mentorship will allow the candidate to conduct a series of research studies with two primary aims. The first is to use machine learning to develop a model that integrates structural and task-based functional MRI data to differentiate adolescents with a history of maltreatment from controls. Identifying the features that best differentiate these groups will determine which circuits are implicated in maltreatment-related psychopathology. These models will be constructed with data from an existing high-dimensional database that was obtained by project mentors. The second aim is to obtain pilot data on the association between variations in brain regions and the severity of psychopathology in a sample of maltreated adolescents. When this research is completed, the brain features most affected by maltreatment will be identified and the relation between these brain features and observable symptoms will be quantified. Such knowledge will provide a set of targets for treatment and assist in our classification of maltreatment-related mental illness. The experience and data obtained from this project will position the candidate to pursue future NIH funding to further examine the neurobiology of maltreatment- related psychopathology in adolescence.

 描述（由申请人提供）：该指导临床科学家发展奖（K08）的目标是为候选人提供1）机器学习，2）神经成像和3）发育精神病理学方面的广泛培训，以对儿童虐待导致的精神病理学基础的神经生物学进行独立调查。有强有力的证据表明，童年时期的虐待会导致发育中的大脑中多个回路的差异。受影响的电路参与的程度 然而，精神病理学尚不清楚。确定涉及哪些回路对于开发针对特定神经生物学系统的干预措施是必要的。拟议研究的主要目的是阐明多个大脑特征的参与（例如，结构和功能激活）在虐待相关的精神病理学。这些目标将通过机器学习来实现。机器学习是一组基于计算机的学习方法，其中有意义的理论模型来自经验数据。该提案建立在候选人之前在临床心理学，创伤应激的神经生物学模型和先进的计算方法方面的培训基础上，以发展以下方面的专业知识：1）包括机器学习，遗传算法和人工神经网络的计算方法，2）多模态神经成像数据的收集和3）虐待对发展的影响。额外的专业发展培训将包括语法，手稿准备和研究道德。为了实现这些目标，候选人已经组建了一个由专家组成的成熟的导师团队，涵盖了一系列学科，包括发展精神病理学，计算科学和神经科学。培训和指导将使候选人能够进行一系列研究，主要目的有两个。第一个是使用机器学习开发一个模型，该模型集成了结构和基于任务的功能性MRI数据，以区分有虐待史的青少年和对照组。识别最能区分这些群体的特征将确定哪些回路与虐待相关的精神病理学有关。这些模型将使用项目指导人员获得的现有高维数据库中的数据构建。第二个目的是获得试点数据之间的关联在大脑区域的变化和精神病理学的严重程度在一个样本的虐待青少年。这项研究完成后，将确定受虐待影响最大的大脑特征，并将量化这些大脑特征与可观察到的症状之间的关系。这些知识将提供一套治疗目标，并有助于我们对与虐待有关的精神疾病进行分类。从这个项目中获得的经验和数据将使候选人能够在未来寻求NIH的资助，以进一步研究青少年虐待相关精神病理学的神经生物学。