You are viewing the SNL 2018 Archive Website. For the latest information, see the Current Website.

Poster B43, Thursday, August 16, 3:05 – 4:50 pm, Room 2000AB

Predicting naming responses based on pre-articulatory electrical activity in individuals with aphasia

Janina Wilmskoetter1,2, John Delgazio2, Lorelei Phillip3, Roozbeh Behroozmand3, Ezequiel Gleichgerrcht2, Julius Fridriksson3, Leonardo Bonilha2;1Department of Health Sciences and Research, College of Health Professions, Medical University of South Carolina, Charleston, SC, 2Department of Neurology, College of Medicine, Medical University of South Carolina, Charleston, SC, 3Department of Communication Sciences and Disorders, University of South Carolina, Columbia, SC

Background: Anomia – trouble with naming – is among the most common symptom of language processing due to neurological disorders, including post-stroke aphasia. Virtually all individuals with aphasia, independently of aphasia type or syndrome, present word-finding deficits with varying degrees. Importantly, naming errors are unpredictable and vary across testing sessions. Thus, they are not item specific, but likely relate to the insufficient emergence of appropriate patterns of neural activity. In the study presented here, we sought to investigate whether pre-articulatory neural activity can be systematically used to predict individualized naming error responses in individuals with aphasia. Methods: We performed 64-channel high density electroencephalography (EEG) on one individual with chronic post-stroke aphasia (59 year old male, 13 years since stroke) during naming of 80 concrete images. Time between image (stimulus) presentations was 8 seconds and the subject’s responses were audio-recorded, transcribed and classified into correct and incorrect responses. Using Curry 8, we pre-processed the EEG signal and calculated stimulus-locked event-related potentials (ERPs) for a time range of 0ms to 1500ms after stimulus presentation. We applied deep machine learning with recurrent convolutional neural networks to predict correct and incorrect responses. Results: After preprocessing of the EEG signal, 69 of the 80 stimuli met the criteria to be included for analysis. Of those, the patient named 49 correctly and 20 incorrectly. Using the pre-articulatory EEG signal, we were able to predict correct and incorrect responses with an accuracy of >70%. Conclusions: Our findings indicate that it is possible to predict correct and incorrect naming responses based on pre-articulatory neural activity. Future research is needed to improve prediction accuracy, and extend applicability to other individuals with aphasia. We believe that that this line of research has the potential to guide the development of new treatment approaches that take neural activity into consideration.

Topic Area: Language Disorders