Multidimensional Adaptive Assessment of Language Impairment in Aphasia

Significance to VA: Aphasia is a neurogenic language impairment affecting over 2 million Americans, including more than 24,000 Veterans enrolled in the VHA. Structured tests of language performance are central to both clinical management of aphasia and its research. The Western Aphasia Battery Revised is the most commonly used aphasia assessment tool in the U.S., but its theoretical foundation lacks empirical support, which limits its ability to generate robust and valid clinical inferences. The global standard, the Comprehensive Aphasia Test, aligns better with current theories but also has limitations, including a lack of American norms and reliance on outdated psychometric methods, which do not maximize the validity and efficiency of its score estimates. This project aims to address these gaps using multidimensional extensions of Item Response Theory (IRT) within a structural equation modeling framework to develop a modern, efficient, and theoretically grounded aphasia assessment tool. Innovation and Impact: This project represents the first application of multidimensional IRT to the assessment of aphasia. These models, which have been used in educational tests such as the Graduate Record Examination, for decades, can support adaptive testing, which can substantially reduce testing burden while minimizing the loss of measurement precision. Reduced testing burden will improve care for Veterans with aphasia by allowing more time for counseling, treatment planning, and treatment. IRT models can also support test validity studies that can lead to more valid and interpretable score estimates and inferences. Specific Aims: Aim 1: Select an appropriate multidimensional IRT model for the Comprehensive Aphasia Test and establish norms for speakers of American English. The models tested will include at least five modality- oriented dimensions and these models will be compared statistically with models organized by processing domains representing the most salient core theoretical distinctions in aphasic language performance. Aim 2: Estimate the potential efficiency gains of multidimensional adaptive testing via simulations. We will use simulations to assess the extent to which adaptive testing can reduce testing burden while minimizing the loss of precision. We will quantify the loss of precision as a function of test length using model-based conditional standard error curves and marginal reliability coefficients. Aim 3: Develop a clinically useful software application for adaptive administration and scoring. We will develop a user-friendly software application that will provide adaptive administration and norm-referenced scoring of the Comprehensive Aphasia Test. We will trial this software in the context of our clinic and collect feedback from clinician and Veteran users Methodology: This study will use data collected from 342 Veterans and 262 non-Veterans with aphasia who have either been seen for aphasia rehabilitation in the VA Pittsburgh Program for Intensive Residential Aphasia Treatment and Education or participated in one or more VA- or NIH-funded research studies at VAPHS. The data set includes administrations of the Comprehensive Aphasia Test (CAT), a 198-item aphasia battery and associated connected speech samples. The CAT data will be cleaned and entered, the speech samples will be scored by a study speech pathologist, and the data will be used to estimate and compare the measurement models described above, develop American norms for the CAT, and conduct simulation studies of adaptive CAT administration. Path to Translation/Implementation: The software product of this work will be disseminated for use with Veterans with aphasia within VHA, and the other products will be used to support development of a next- generation theoretically and psychometrically modern, fully adaptive, culturally sensitive aphasia battery. Project Number: 1I01RD000601-01A1 | Fiscal Year: 2026 | NIH Institute/Center: Veterans Affairs (VA) | Principal Investigator: William Hula | Institution: VETERANS HEALTH ADMINISTRATION, PITTSBURGH, PA | Activity Code: I01 | Study Section: Sensory Systems & Communication Disorders[RRD3] View on NIH RePORTER: https://reporter.nih.gov/project-details/11242550