A ‘milestone’ in technology: Thanks to a brain interface, women can ‘talk’ again

“teacher” in technology
Women can “talk” again thanks to the brain interface

08/23/2023 at 7:42 PM

Incredibly efficient brain-computer interfaces are restoring the ability to ‘talk’ in stroke patients – even using their original voice. Research colleagues are impressed. When will there be systems suitable for everyday use?

For example, brain-computer interfaces should enable people to talk again after a stroke. Two such approaches have been presented by two research groups in the journal Nature. in study The patient was able to articulate an average of 62 words per minute through her thoughts Another Job One woman could pronounce 78 words per minute. This corresponds to about half the English speaking speed of about 150 words per minute.

In one study, the group led by Francis Willett of Stanford University in California treated patient Pat Bennett, who had amyotrophic lateral sclerosis (ALS). This leads to progressive muscle paralysis, which also affects speech.

Microelectrodes in the brain

The researchers implanted four arrays of tiny electrodes into areas of the brain associated with speech. Brain activities measured in this way were transmitted by cable to a computer system, which converted the signals into text.

The patient trained the system about twice a week by speaking pre-set texts in her mind, and the researchers modified the system to fit her brain patterns. Four months later, the woman was able to say 62 words per minute just by thinking about her. This speed exceeds previous systems by more than three times.

“For people who cannot speak, it means they can stay connected to the larger world, possibly continue to work and maintain relationships with friends and family,” Bennett said in a statement to Stanford University.

It can also be done without surgery

The second study, by a team led by Edward Chang of the University of California, San Francisco, also relies on a case study. This woman lost her ability to speak after suffering a stroke. The researchers used the electrocorticography method: brain signals are measured directly on the surface of the brain without the need for needles to be inserted into the brain, as is the case with microelectrode arrays. However, measuring electrodes must be applied to a larger area of ​​the brain in this method.

Zhang’s group decoded the signals sent from the brain to the muscle groups involved in speech. It also focused on 39 phonemes, called phonemes, for word recognition – this should speed up word recognition.

Reconstruct the original sound

The avatar then speaks on screen, moving its mouth and lips in real time according to the decoded sounds. There is even a reconstruction of the patient’s original voice: for the speech modification, the team used the audio recording of a speech the woman gave at her wedding.

In addition, the patient was able to use her thoughts to assign three facial expressions to the avatar, each of three different intensities: happy, sad, and surprised. “These advances bring us much closer to developing a real solution for patients,” Chang said in a statement from his university.

Higher accuracy in the Stanford study

This patient averaged 78 words per minute. However, the word error rate for a vocabulary of 1024 words was one fourth. The Stanford team achieved an error rate of just under 24% with their patient, with a vocabulary of 125,000 words.

The system is still far from a device that people can use in everyday life, stresses Stanford researcher Willett. “But it’s a huge step forward in quickly restoring communication for paralyzed people who can’t speak.”

“A turning point in development”

Commenting for Nature, Nick Ramsey of University Hospital Utrecht and Nathan Crone of Johns Hopkins University School of Medicine in Baltimore write about a turning point in the development of such technologies: “The two reports are evidence that communication using implantable brain-computer interfaces can be restored.” The next step now is to outfit the electrical implants with a wireless connection.

Torsten Zander of the Technical University of Brandenburg in Cottbus-Senftenberg sees a clear engineering progression. “The results presented are very promising in their practical application.” However, they have only been tested on people who have trained on them for several weeks.

Souriju Sokadar, Chair of the Clinical Neurotechnology Working Group at the Charité Berlin, talks about a milestone in the development of brain-computer interfaces. But he also stresses that the two patients were specially selected for the studies. “There is still a long way to go before widespread use of these technologies can be envisaged.”