Externalizing outcomes in high risk youth

项目概要/摘要 公众经常关注暴力与心理健康问题之间的关系，但是 暴力行为是一种非常复杂的现象，受到无数社会、心理、环境、 和生物学影响。该项目旨在调查大样本中的这些影响 以前被监禁的青少年，跟踪他们并收集有价值的纵向数据。我们的团队 与这些年轻人及其家人合作，收集详细的心理、行为和神经影像 作为先前 NIH 资助的调查的一部分的措施。当前的项目旨在重新评估这些 个人（现在是年轻人）检查长期积极的（即停止吸毒/反社会） 行为）和负面（吸毒、反社会行为）结果。我们将收集新的神经影像 扫描与先前的 MRI 数据相结合，将有助于量化映射到的变化轨迹 对外化结果的坚持和抵制。先进的机器学习方法将 此外，还与结构、功能、网络和动态网络大脑测量结合使用 行为和心理测量。机器学习方法能够识别模式 高维数据并描绘最能预测特定情况的独特变量组合 结果变量。使用这些方法，我们打算定义预测结果的神经机制。我们 还旨在识别导致持续反社会行为更大风险的变量组合，以及 暴力。这项工作的转化价值将是澄清数据的信息模式，这些模式可能表明 可预防的结果。此外，指示特定脆弱性的神经测量将被确定为 具体的治疗目标和新颖的干预策略。通过识别特定的漏洞和 伴随积极成果而来的变化，我们将更接近理解认识和认识的最佳方式 防止代价高昂的暴力行为。

Project Summary/abstract Public attention has often been drawn to the relationship between violence and mental health issues, but violent behavior is a very complex phenomenon undergirded by myriad social, psychological, environmental, and biological influences. This project intends to investigate these influences among a large sample of previously incarcerated youth, following up with them and collecting valuable longitudinal data. Our team worked with these youth and their families collecting detailed psychological, behavioral, and neuroimaging measures as part of a previous NIH-funded investigation. The current project aims to re-assess these individuals (now young adults) to examine long-range positive (i.e., desistance from drug use/antisocial behavior) and negative (relapse to drugs, antisocial behavior) outcomes. We will collect new neuroimaging scans, which combined with prior MRI data, will be useful for quantifying trajectories of change that map to persistence and desistence from externalizing outcomes. Advanced machine-learning approaches will be utilized in conjunction with structural, functional, network, and dynamic network brain measures in addition to behavioral and psychological measures. Machine learning approaches are capable of identifying patterns in high-dimensional data and delineating the unique combinations of variables that are most predictive of specific outcome variables. Using these methods, we intend to define neural mechanisms that predict outcomes. We also aim to identify combinations of variables that confer greater risk for persistent antisocial behavior and violence. The translational value of this work will be to clarify informative patterns of data that may indicate preventable outcomes. Furthermore, neural measures indicative of specific vulnerability will be identified as specific targets for treatment and novel intervention strategies. By identifying specific vulnerabilities and the changes that accompany positive outcomes, we will be closer to understanding the best way to recognize and prevent costly violent behavior.