Neuroprosthesis converts brain waves to speech in paralysed man with anarthria

A newly developed neuroprosthetic device has directly translated the brain waves of a paralysed person with anarthria into complete sentences, according to a study by researchers from the University of California San Francisco (UCSF).

“In a person with anarthria and spastic quadriparesis caused by a brain-stem stroke, words and sentences were decoded directly from cortical activity during attempted speech with the use of deep-learning models and a natural-language model,” the researchers said.

In the study, a subdural, high-density, multielectrode array was implanted over the area of the sensorimotor cortex that controls the vocal tract in a 36-year-old man with anarthria (loss of the ability to articulate speech) and spastic quadriparesis, which was caused by a brain-stem stroke.

The researchers recorded 22 hours of cortical activity over the course of 48 sessions while the participant tried to say individual words from a vocabulary set of 50 words. They used deep-learning algorithms to create computational models that detected and classified words from patterns in the recorded cortical activity.

These computational models, as well as a natural-language model yielding next-word probabilities given the preceding words in a sequence, were applied to decode full sentences as the participant attempted to say them. The computer algorithm was trained to differentiate among 50 words that could generate over 1,000 sentences.

Sentences from the participant’s cortical activity were decoded in real time at a median rate of 15.2 words per minute, with a median error rate of 25.6 percent. Post hoc analyses revealed 98 percent attempts by the participant to produce individual words. [N Engl J Med 2021;385:217-227]

USCF researchers were able to classify words with 47.1-percent accuracy via cortical signals that were stable throughout the 81-weeks study period. Words were decoded with a median accuracy of 74 percent at 15 words per minute. The neuroprosthetic device had a best performance of up to 93-percent accuracy at 18 words per minute. [https://www.youtube.com/watch?v=_GMcf1fXdW8&t=38s]

“This is the first time [that] brain activity in the speech cortex of a paralysed person has been decoded into full words,” the researchers said. “The advances pave the way for new technology to restore natural and rapid communication for people with severe paralysis.”

The result was described as an “important technological milestone” for an individual who lost the ability to speak naturally, according to David Moses, one of the lead authors of the study and a postdoctoral engineer at UCSF.

"It demonstrates the potential for this approach to give a voice to people with severe paralysis and speech loss," he said. [https://bit.ly/3zbgfiF]

The novel neuroprosthetic device has the potential to exponentially improve autonomy and quality of life in paralysed persons with speech loss.

“An approach that decodes words and sentences directly from the cerebral cortical activity of such patients may represent an advancement over existing methods for assisted communication,” the researchers said.