‘Virtual language therapist’ set to benefit aphasia patients

By Andrew Czyzewski, medwireNews Reporter

Portuguese researchers have developed an online speech recognition tool for the treatment of word naming difficulties in patients with aphasia.

Lead author Alberto Abad (Institute for Systems and Computer Engineering, Lisbon) and colleagues found that their tool achieved high word verification rates relative to manual human word naming for different types of patients and acoustic conditions.

"In practice, the platform is intended not only to serve as an alternative, but most importantly, as a complement to conventional speech-language therapy sessions, permitting intensive and inexpensive therapy to patients, besides providing to the therapists a tool to assess and track the evolution of their patients," they comment in the journal Computer Speech & Language.

Aphasia is a communication disorder caused by the damage of one or more language areas of the brain. The most common symptom of the disorder is a difficulty to recall words or names.

Recently, Abad and colleagues have presented the first prototype of an online platform that incorporates speech recognition technology named VITHEA (Virtual Therapist for Aphasia treatment).

Adbad and colleagues tested their VITHEA tool in a total of 16 patients (in two groups of eight) focusing on two measures: the word naming score (WNS) and the word verification rate (WVR).

The WNS was computed for every speaker as the number of positive word detections divided by the total number of exercises. Meanwhile, the WVR was computed for every speaker as the number of coincidences between the manual and automatic result (true acceptances and true rejections) divided by the total number of exercises.

Each patient was assessed by a human evaluator and by the automatic detector.

In the first dataset of eight patients, the Pearson's correlation coefficient between human and automatic WNS scores was 0.904 and the WVR for the VITHEA was 0.800.

In the second dataset, the Pearson's correlation coefficient between human and automatic WNS scores was 0.972, but with a lower WVR of 0.66 for VITHEA.

Adbad and colleagues attribute this to electrical noise, due to the failure of one of the components of the two-microphone recording set-up.

"In spite of the different patient characteristics and speech quality conditions of the collected data, encouraging results have been obtained thanks to a calibration method that makes use of the patients word naming ability to automatically adapt to the patients speech," the researchers conclude.

Licensed from medwireNews with permission from Springer Healthcare Ltd. ©Springer Healthcare Ltd. All rights reserved. Neither of these parties endorse or recommend any commercial products, services, or equipment.