NUVA: A Naming Utterance Verifier for Aphasia Treatment

Anomia (word-finding difficulties) is the hallmark of aphasia, an acquired language disorder most commonly caused by stroke. Assessment speech performance using picture naming tasks a key method for both diagnosis and monitoring responses to treatment interventions people with aphasia (PWA). Currently, this assessment conducted manually therapists (SLT). Surprisingly, despite advancements in automatic recognition (ASR) artificial intelligence technologies like deep learning, research on developing automated systems task has been scarce. Here we present NUVA, utterance verification system incorporating learning element that classifies 'correct' versus' incorrect' attempts from aphasic stroke patients. When tested eight native British-English speaking PWA system's accuracy ranged between 83.6% 93.6%, 10-fold cross-validation mean 89.5%. This was not only significantly better than baseline created study one leading commercially available ASRs (Google speech-to-text service) but also comparable some instances two independent SLT ratings same dataset.