Advanced Automated Assessment of Cognitive Changes Associated with Brain Injury and Neurological Disease

The proposed work will focus on developing an automated language-based assessment system, AVANT, that is primarily self-administering with clearly presented directions to a patient using avatar-based verbal instructions. Using speech recognition and a conversational avatar representing a virtual clinician, AVANT will enhance current approaches to neuropsychological assessment and allow for avatar-administered assessments, based on various perceptual modalities (auditory, visual, gestural), that augment a real clinicians ability to assess patients. The aim is the implementation of speech recognition using a virtual clinician that is sufficiently accurate to permit computer administration of verbally-based neuropsychological tasks including word-list learning, confrontation naming, and aural comprehension. Ultimately, AVANT will employ extended natural language capability through use of the Soar cognitive architecture to administer additional language-based tasks. In this work, the team plans to have the virtual clinician administer neuropsychological tests and accept patient responses as real clinicians now do.