Assessment of anomia: Improving efficiency and utility using item response theory

项目概要/摘要 失语症是失语症的一个核心特征，这种疾病影响着至少 250 万美国人。命名评估 能力是失语症研究和临床实践的基础，但目前可用的命名测试 很大程度上未能利用过去 50 年来心理测量理论和实践的重要进展 年。需要敏感的指标来支持对失范症背后的认知机制的调查， 临床决策和治疗研究。为了发挥最佳作用，这些指标必须在以下范围内进行验证： 严格的心理测量框架，以 (i) 有效地提供有关总体严重程度和 统一框架中的潜在认知缺陷，（ii）支持重复评估而不构成威胁 内部有效性或测量精度，(iii) 以对象和动作命名为目标，以及 (iv) 轻松集成 进入计算机化自适应测试平台。目前所有的贫血测试都至少缺乏其中一项功能。至 为了弥补这一差距，该项目追求三个目标。 首先，简化诊所和实验室之间的信息共享，并增强开展稳健的能力 元分析研究，五个对象图片命名测试将在项目反应理论中等同 框架，并且他们的分数将用相同的指标表示。此外，为了避免需要大 在未来项目开发的样本中，将完善预测大多数可图形名词难度的模型 并交叉验证。正如长度的测量不取决于使用的尺子一样，测量 失范严重程度的确定并不取决于给出的命名测试。 其次，为了量化失常症的潜在认知缺陷，我们建立了一种认知心理测量模型 贫血症将被完善和评估，以产生一个平衡临床效用、模型数据的测量模型 适合并忠实于当前理论。 第三，鉴于人们日益认识到动作命名的重要性，计算机自适应动作 将使用项目响应理论通过对动作命名响应进行建模来开发命名测试；回归 相关目标属性的动词项目难度参数；并进行真实数据模拟来评估 计算机自适应测试引擎的精度。 这项工作将产生一个强大而灵活的失智症评估工具，可用于研究和 临床实践。根据用户规格，它将最大化诊断信息和测量 精度，同时最大限度地减少响应负担。该项目的目标直接解决了 NIDCD 2017-2021 年战略计划，包括开发和完善工艺、技术和 用于改进诊断的仪器，以帮助治疗和改善临床结果。

Project Summary/Abstract Anomia is a core feature of aphasia, a disorder affecting at least 2.5 million Americans. Assessment of naming ability is foundational to both research and clinical practice in aphasiology, but currently available naming tests have largely failed to capitalize on important advances in psychometric theory and practice over the past 50 years. Sensitive metrics are needed to support investigations of the cognitive mechanisms underlying anomia, clinical decision making, and treatment research. To be optimally useful, these metrics must be validated within a rigorous psychometric framework to (i) efficiently provide information about both overall severity and the underlying cognitive deficits in a unified framework, (ii) support repeated assessments without threatening internal validity or measurement precision, (iii) target both object and action naming, and (iv) easily integrate into computerized adaptive testing platforms. All current anomia tests lack at least one of these features. To remedy this gap, this project pursues three aims. First, to ease sharing of information across clinics and labs, and enhance the ability to conduct robust meta-analytic studies, five object picture-naming tests will be equated within an item response theory framework, and their scores will be expressed on the same metric. Further, to obviate the need for large samples in future item development, a model for predicting the difficulty of most picturable nouns will be refined and cross-validated. Just as the measurement of length does not depend on which ruler is used, measurement of anomia severity need not depend on which naming test is given. Second, to quantify the underlying cognitive deficits of anomia, a cognitive-psychometric model of anomia will be refined and evaluated to produce a measurement model that balances clinical utility, model-data fit, and fidelity to current theory. Third, given the increasingly recognized importance of action naming, a computer adaptive action naming test will be developed using item response theory by modeling action naming responses; regressing verb item difficulty parameters on relevant target properties; and, conducting real-data simulations to assess the precision of the computer adaptive test engine. This work will produce a powerful and flexible anomia assessment tool with utility for both research and clinical practice. Depending on user specifications, it will maximize diagnostic information and measurement precision while minimizing response burden. The aims of this project directly address goals identified in the NIDCD Strategic Plan for 2017-2021, including developing and refining techniques, technology, and instrumentation for improved diagnosis to aid in treatment and improve clinical outcomes.