The emergence of machine learning techniques, especially in their applications to language, provides fertile ground for computational modeling, an avenue of research that has been historically underutilized in the study of communication sciences and disorders. Our lab is investigating methods of simulating abstract and concrete word processing with models trained on neuroimaging, behavioral or corpus data. The creation of these models will not only enhance our scientific understanding of language processing in the brain, but may also be of clinical relevance, as their performances when disturbed can be compared with and perhaps used to predict patient outcomes after stroke and in therapy.

A poster of our work can be seen here:
DiMercurio_Poster_CNS_Final