(CNN) – Researchers said Wednesday that a brain transplant has helped restore a man’s speech pattern for more than 15 years after a stroke robbed him of his ability to speak.

They call this a neuroprosthesis, and while he is now a patient, a team from the University of California, San Francisco (UCSF) hopes that their device will help the paralyzed to communicate with others.

“To the best of our knowledge, this is the first successful demonstration of direct word decoding of the brain function of a paralyzed and speechless person,” said Dr. Edward Chang, a UCSF neurosurgeon who led the research team.

“There is great promise in restoring communication through the use of the brain’s natural speech machines,” Chang said in a statement.

The team fitted a series of electrodes to the area of ​​the brain that controlled the speech of a man suffering from stroke, which paralyzed him and made him unable to speak in his 20s.

“Since he was injured, his head, neck and limbs have had very little movement, and he communicates using a stylus attached to a baseball cap to mark the characters on a screen,” UCSF said in a statement.

“His cognitive function remained the same,” the group wrote in their report, published in the New England Journal of Medicine.

The man, now in his 30s, was asked to use a limited vocabulary when adjusting a device using computer algorithms to translate the electrical activity of his brain. These words were then projected onto the computer screen.

Contains video of the man using the UCSF device. “Good Morning” is promoted by the computer screen. The answer is “hello” and after a few seconds the text on the screen is also written.

Neurosurgeon Edward Chang performs brain surgery at UCSF.

“How are you today?” The patient responds reluctantly: “I’m fine.”

“We decoded the sentences from the participant’s cortical activity in real time at an average rate of 15.2 words per minute, with an average word error rate of 25.6%,” the panel wrote.

“We found 98% of attempts to generate participants’ personal words, and categorized words with 47.1% accuracy using cortical references that were consistent over the 81-week study period, ”the researchers said.

Finally, the patient, who was asked not to be identified, helped the team create a vocabulary of 50 words that included some words such as “yes,” “no,” “family,” “clean,” and “nurse.” These were extended to whole sentences like “No, I’m not thirsty”.

This is not a permanent solution: the electrode is a large device that sits on top of the skull and cannot be used continuously. But researchers say this is not a test miracle.

“In previously announced brain-computer interface applications, decoding models often require daily overhaul before being implemented with a user,” the researchers wrote. This device, they said, is more stable.

“This is an important technical milestone for a person who is unable to communicate naturally, and demonstrates the ability of this approach to give voice to those with severe stroke and speech loss,” said David Moses, a postdoctoral engineer at Chang’s lab. Studio.

Participant’s speech Explains the location of the echoji electrode in the motor cortex and the positions of the head used to connect the electrode to the computer.

“This test is only a beginning. It was the first participant in the trial and the first tests that were part of this experiment to show that it was possible,” Chang said.

“On the hardware side, we need to create systems with higher data resolution to quickly record more information from the brain. On the algorithm side, there should be systems that can translate these complex brain signals into spoken words, not text, but oral and audible spoken words,” he said.

“Perhaps one of the most important priorities is to expand the vocabulary not limited to the 50 words we start with, but one that can be generalized to all words in English, for example. We need to make sure what we do. This participant can be found with others for a wider patient population.”

Other teams have tried to talk to paralyzed people.

In 2017, a team from the University of Dubingen in Germany used a cap with EEG sensors to help patients with paralysis caused by amyotrophic lateral sclerosis (ALS) express some simple thoughts.