Using Machine Learning Approaches to Examine Emotion-Related Brain Activity and Substance Use Among Adolescents

项目摘要/摘要 物质使用和物质使用障碍是世界范围内导致死亡和残疾的主要原因。 重要的是，大多数患有物质使用障碍的成年人在青少年时期就开始使用物质，使 青春期是物质使用障碍发展的重要时期.因此，至关重要的是要确定 与青少年药物使用有关的因素，符合NIDA目标1.1。一个与以下因素相关的因素 青少年物质使用是负性情绪加工。当青少年经历了重大的生物学和 在发育过程中的心理社会变化他们可能会经历改变的负面情绪处理 如果没有适当的管理，就会导致物质的使用。不幸的是，对于神经细胞如何 青少年在负性情绪加工上的水平差异与物质使用有关。此外， 现有的关于负面情绪处理和物质使用的神经成像研究采用了单变量 方法，而不是多元方法。与单变量方法相比，多变量方法具有更多的优点 灵敏地检测跨体素的神经激活模式，并产生更普遍和可复制的结果 总体而言，发现。此外，本研究还采用了标准化的负面情绪范式， 高实验控制性，但可能不太可能反映负面情绪处理，因为它发生在现实中 世界。为了解决文献中的这些空白，拟议的研究将使用多变量机器学习 标准化阴性神经激活模式的分类方法(即多体素模式分析) 情绪加工任务与一种新的自然主义负性情绪加工任务的区别 不使用药物的青少年，以及预测药物使用障碍风险 各种因素。此外，这项拟议的研究将使用机器学习方法来检查 情绪相关的神经激活与物质使用有关的这些任务。这项研究将进行 对赞助商(卓别林)竞争续签补助金(RO1)的326名12-13岁青少年的抽样调查 DA033431-06A1)。这项拟议研究的知识将被用来识别与情绪相关的神经生物学 青少年药物使用的标志。最终，这些标记物可以用来针对高危青少年 需要药物使用预防和干预努力。拟议研究的目标将是 在旨在发展情感领域多学科专业知识的研究培训计划内完成 神经科学，特别是多元机器学习方法和物质发展模型 使用。培训计划包括完成相关课程，参加有针对性的讲习班， 发展和神经科学领域专家的个人指导，以及科学写作和 演示体验。

Project Summary/Abstract Substance use and substance use disorder are leading causes of death and disability worldwide. Importantly, most adults with substance use disorder begin using substances as adolescents, making adolescence an important period for the development of substance use disorder. Thus, it is critical to identify factors related to substance use among adolescents, consistent with NIDA objective 1.1. One factor linked to adolescent substance use is negative emotion processing. As adolescents undergo significant biological and psychosocial changes across development they may experience altered negative emotion processing that can lead to substance use if not appropriately regulated. Unfortunately, there is limited understanding in how neural level differences in negative emotion processing is related to substance use among adolescents. Moreover, extant neuroimaging research on negative emotion processing and substance use has employed univariate methods instead of multivariate methods. In contrast to univariate methods, multivariate methods are more sensitive in detecting patterns of neural activation across voxels and yield more generalizable and replicable findings overall. In addition, this research has employed standardized negative emotion paradigms, which have high experimental control, but may be less likely to reflect negative emotion processing as it occurs in the real world. To address these gaps in the literature, the proposed study will use multivariate machine learning approaches (i.e., multivoxel pattern analysis) to classify patterns of neural activation in a standardized negative emotion processing task and in a novel naturalistic negative emotion processing task that differentiate substance using adolescents from non-using adolescents, as well as predict substance use disorder risk factors. Additionally, the proposed study will use machine learning approaches to examine sex differences in emotion-related neural activation to these tasks in relation to substance use. This research will be conducted on a sample of 326 12-13 year old adolescents from Sponsor’s (Chaplin) competing renewal grant (RO1 DA033431-06A1). Knowledge from the proposed study will be used to identify emotion-related neurobiological markers of adolescent substance use. Ultimately, these markers can be used to target at-risk adolescents in need of substance use prevention and intervention efforts. The goals of the proposed study will be accomplished within a research training plan aimed at developing multidisciplinary expertise in affective neuroscience, particularly in multivariate machine learning methods, and developmental models of substance use. The training plan includes completion of relevant coursework, attendance at targeted workshops, individual mentorship by experts in the field of development and neuroscience, and scientific writing and presentation experience.