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

失范评估：利用项目反应理论提高效率和效用

• 批准号: 9022355
• 负责人: Gerasimos Fergadiotis
• 金额: $ 14.83万
• 依托单位: PORTLAND STATE UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-09-22 至 2018-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/9022355
• 关键词: Address; Affect; Agreement; Algorithms; Anomia; Aphasia; Archives; Binding; Chi-Square Tests; Classification; Cognitive deficits; Communication impairment; Computational algorithm; Computers; Data; Databases; Decision Making; Development; Diagnosis; Diagnostic; Disease; Electronics; Engineering; Evaluation; Generations; Goals; Healthcare; Impairment; Individual; Logistics; Measurement; Measures; Modeling; Mosses; Names; National Institute on Deafness and Other Communication Disorders; Nature; Outcome; Participant; Patients; Performance; Persons; Philadelphia; Plant Roots; Policies; Property; Protocols documentation; Psycholinguistics; Psychometrics; Rehabilitation therapy; Research; Retrieval; Sampling; Severities; Strategic Planning; Symptoms; System; Testing; Theoretical model; Time; United States National Institutes of Health; base; handheld mobile device; health data; improved; novel therapeutics; prevent; prospective; public health relevance; response; simulation; theories; tool; trait

 DESCRIPTION (provided by applicant): Anomia is a core feature of aphasia, a disorder that affects approximately 80,000 people each year. Sensitive metrics of change are needed for the development and evaluation of new anomia treatments and to provide the basis for rehabilitation reimbursement and policy decisions. To be optimally useful, these metrics must (i) be validated within a modern framework, such as item response theory, to enhance their psychometric properties, (ii) yield information about severity and underlying cognitive deficits with low response burden, (iii) support repeated assessments without posing threats to internal validity, and (iv) predict functioning at the activity level of the ICF. The long-term goal of thisline of research is to maximize the precision, efficiency, and utility of word retrieval assessment for quantifying change, evaluating new therapeutic protocols, and diagnosing impairments in terms of current theoretical models. To achieve this goal, this proposal will focus on engineering a theoretically driven, computer-adaptive system using IRT for assessing anomia. IRT is a state-of-the-art psychometric framework with important theoretical and practical benefits. First, it transforms observed responses to test items into latent trait scores with interval properties. Second, it formalizes how measurement precision varies as a function of a person's ability. Third, it supports algorithms for computer adaptive testing (CAT), in which only the most informative items are selected and administered for a particular patient, thus reducing testing time. It also permits the generation of equivalent test forms for assessing the same person on multiple occasions with scores that are expressed on a single common scale. This proposal has the following Specific Aims. Specific Aim 1a is to validate a CAT version of the Philadelphia Naming Test (PNT-CAT) that is based on a 1-parameter logistic IRT model for quantifying overall severity. Specifically, this aim will determine the extent to which CAT-PNT scores are precise, unbiased, and correlated with performance on the full PNT. Specific Aim 1b is to determine the ability of an augmented CAT algorithm to predict error proportions on the full PNT. The goal of Specific Aim 2 is to verify the equivalence and precision of PNT-CAT alternate forms in order to determine stability when no change is expected. Specific Aim 3 will determine the relationship of the full PNT with discourse performance and assess any potential degradation of the strength of the relationship when CAT-PNT - instead of the full PNT - are used. Our overarching goal is consistent with the NIH Roadmap's recognition of the "pressing need to better quantify clinically important symptoms and outcomes". Further, it directly addresses goals identified in the NIDCD Strategic Plan for 2012-2016, including the development of effective and efficient diagnostic tools by using "electronic health data to inform the conscientious, explicit, and judicious use of current best evidence in making decisions about health care options to improve outcomes for individuals with communication disorders."

 描述（由申请人提供）：失语症是失语症的核心特征，失语症是一种每年影响大约 80,000 人的疾病，需要敏感的变化指标来开发和评估新的失语症治疗方法，并为康复报销和政策提供基础。为了发挥最佳作用，这些指标必须（i）在现代框架（例如项目反应理论）内进行验证，以增强其心理测量特性，（ii）产生有关严重性和潜在认知缺陷的信息。响应负担低，(iii) 重复支持而不会对内部有效性评估构成威胁，以及 ​​(iv) 预测 ICF 活动水平的功能。该研究的长期目标是最大限度地提高精度、效率和效率。词检索评估在量化变化、评估新的治疗方案以及根据当前理论模型诊断损伤方面的效用为了实现这一目标，本提案将重点关注使用 IRT 来设计一个理论驱动的计算机自适应系统来评估失常症。 IRT 是一种最先进的心理测量框架，具有重要的理论和实践优势，首先，它将观察到的测试项目反应转化为具有区间属性的潜在特质得分；其次，它正式化了测量精度如何随个人的变化而变化。第三，它支持计算机自适应测试（CAT）算法，其中只为特定患者选择和管理信息最丰富的项目，从而减少测试时间，以评估同一个人。在多个场合用分数表示该提案具有以下具体目标。具体目标 1a 是验证费城命名测试 (PNT-CAT) 的 CAT 版本，该版本基于 1 参数逻辑 IRT 模型，用于量化整体严重性。目标将确定 CAT-PNT 分数的精确性、公正性以及与完整 PNT 性能的相关程度。 具体目标 1b 是确定增强 CAT 算法预测错误比例的能力。完整 PNT 的目标是验证 PNT-CAT 替代形式的等效性和精确性，以便在预计不会发生变化时确定稳定性。特定目标 3 将确定完整 PNT 与话语表现的关系并评估任何变化。当使用 CAT-PNT（而不是完整的 PNT）时，关系的强度可能会降低。我们的总体目标与 NIH 路线图认识到的“迫切需要更好地量化临床重要症状和结果”相一致。它直接解决了 NIDCD 2012-2016 年战略计划中确定的目标，包括通过使用“电子健康数据，在做出医疗保健决策时认真、明确和明智地使用当前最佳证据，从而开发有效和高效的诊断工具”改善患有沟通障碍的个人的结果的选择。”