News

PSU professor awarded $2M grant to diagnose language disorder faster, easier
Author: Cristina Rojas, College of Liberal Arts and Sciences
Posted: October 11, 2018

With the help of a federal grant, a Portland State University professor is trying to make it faster and easier to diagnose an increasingly common language disorder known as anomia.

Gerasimos Fergadiotis, a speech-language pathologist in PSU's College of Liberal Arts and Sciences, and Steven Bedrick, an assistant computer science professor at Oregon Health and Science University, were awarded a five-year, $2 million grant from the National Institutes of Health. Work began Sept. 1, but the pair previously received funding during a pilot phase.

Anomia refers to a condition where an individual has difficulty expressing the words they want to say, mainly the result of strokes or brain injuries. For example, they might want to say "milk" but instead say "bilk" — a word that, while incorrect, is phonologically related to the target— or they might want to say "dog" but instead say "cat" — an incorrect but semantically related word.

Often the severity of the language deficits is determined by a time-consuming process of a speech-language pathologist eliciting, transcribing, and analyzing a patient's language sample. Fergadiotis hopes that within five years, a patient could sit in front of a computer, name a series of pictures and computer algorithms would then analyze the sample and determine the nature of the errors, helping to shed light on the impaired cognitive processes underlying language production of stroke survivors.

"The goal is not to replace speech-language pathologists, but to provide them with a tool that's going to give them clinically relevant information in a fraction of the time it would ordinarily take them," he said. "By creating a profile of the errors a person is producing, we can then tailor the treatment to their specific deficits."

Fergadiotis said the algorithms could also be used to re-analyze hundreds, if not thousands, of samples that have already been collected from patients.

"It's extremely time-consuming to go back and re-analyze language samples in extant databases, but if you have an automated tool that's effective, efficient and reliable, then we're hoping this can be used to accelerate research and discovery."