An Adaptive Testing Platform for Optimizing RDoC Experimental Cognitive Measures

项目摘要 作为研究领域标准(RDoC)倡议的一部分，NIMH寻求改进神经元的测量 以及干预研究中使用的心理指标。RDoC工具必须精确测量认知和 神经系统产生可靠的发现。不幸的是，许多RDoC任务尚未完成 心理测量学评估，或改进，并容易混淆，可能导致不准确的主张 不同的缺陷，减弱的效应大小，以及相互矛盾的脑成像结果。正如FOA PAR所强调的那样- 18-930，迫切需要现代心理测量方法和工具来支持认知和 临床神经科学研究。该项目通过评估和改进方法来回应这一FOA 旨在管理RDoC任务的统计稳健变体，特别是在脑成像的背景下。 我们创建了一种定量方法，旨在管理计算机化自适应测试(CATS)在 认知和临床神经科学研究的背景。猫实时操纵刺激属性 为了提高测量精度，避免天花板和地板的影响，并最大化效果大小，即使在 认知功能水平差异很大的个人和群体。猫也会进行心理测试 对认知任务的调整，以便大脑功能异常可以独立于 业绩赤字。在试点工作中，我们将这种方法用于RDoC工作记忆任务，N- 回到过去，以表明该方法提高了认知和脑成像数据的可靠性。我们会 评估适应性测试的普适性和影响，超越N-back任务，用于翻译 和实验测试。精神分裂症患者和对照组将同时接受适应性和非适应性治疗 用于评估工作记忆的四种RDoC范例的自适应版本：延迟匹配到样本， Sternberg、自序指向和N-Back。此外，参与者将接受5个选项的管理 连续绩效任务和概率学习任务、控制和学习的翻译测量 分别进行了分析。这两组人都将接受功能神经成像，并对这些功能的适应性版本做出反应 任务。具体目标是：(1)确定适应性测试是否提高了精确度和效果大小 对RDoC任务产生的性能差异的估计；以及(2)确定适应性测试 提高了RDoC任务产生的大脑激活差异的可靠性和效果大小估计。而当 这些目标旨在评估一种方法，并解决与使用 RDoC工作记忆任务在精神病研究中作为神经探测器、中介和结果，结果还 在人群、大脑区域和网络以及认知域都有广泛的影响。通过寻址 对于低可靠性、弱效应大小和大脑激活的担忧是令人困惑的，这个项目将展示适应性 测试广泛地改善了认知神经科学的任务。为了促进自适应RDoC任务的快速部署， 我们将开发所用范例的免费版本和一个配套的R包‘catCog’。

Project Summary As part of the Research Domain Criteria (RDoC) initiative, the NIMH seeks to improve measures of neuronal and psychological targets for use in intervention research. RDoC tools must precisely measure cognitive and neuronal systems to produce reliable findings. Unfortunately, many RDoC tasks have not been psychometrically evaluated, nor refined, and are liable to confounds that can lead to inaccurate claims of differential deficit, weakened effect size, and contradictory brain imaging findings. As highlighted by FOA PAR- 18-930, there is a critical need for modern psychometric methods and tools designed to support cognitive and clinical neuroscience research. This project responds to this FOA by evaluating and refining a methodology designed to administer statistically robust variants of RDoC tasks, especially in the context of brain imaging. We have created a quantitative methodology designed to administer computerized adaptive tests (CATs) in the context of cognitive and clinical neuroscience research. CATs manipulate stimulus properties in real time in order to improve measurement precision, avoid ceiling and floor effects, and maximize effect size, even for individuals and groups with highly discrepant levels of cognitive functioning. CATs also perform psychometric adjustments to cognitive tasks so that brain functioning abnormalities can be interpreted independent of performance deficits. In pilot work, we have used this approach with an RDoC working memory task, the N- back, to show that the methodology improves the reliability of both cognitive and brain imaging data. We will evaluate the generalizability and impact of adaptive testing, beyond the N-back task, for use in translational and experimental testing. Patients with schizophrenia and controls will be administered both adaptive and non- adaptive versions of four RDoC paradigms used to assess working memory: delayed match-to-sample, Sternberg, self-ordered pointing, and N-back. Additionally, participants will be administered the 5-Choice Continuous Performance Task and Probabilistic Learning Task, translational measures of control and learning respectively. Both groups will undergo functional neuroimaging and respond to adaptive versions of these tasks. The specific aims are to: (1) Determine whether adaptive testing improves the precision and effect size estimates of performance differences produced by RDoC tasks; and (2) Determine whether adaptive testing improves the reliability and effect size estimates of brain activation differences produced by RDoC tasks. While these aims are designed to evaluate a methodology, and to address critical concerns related to the use of RDoC working memory tasks as neural probes, mediators, and outcomes in psychosis research, results also have broad implications across populations, brain regions and networks, and cognitive domains. By addressing concerns of poor reliability, weak effect size, and brain activation confounds, this project will show that adaptive testing broadly improves cognitive neuroscience tasks. To facilitate rapid deployment of adaptive RDoC tasks, we will develop freely available versions of the paradigms used and a companion R package ‘catCog’.