Bilingual AI brain implant helps stroke survivors communicate in Spanish and English

Scientists at the University of California, San Francisco developed a bilingual brain implant which uses artificial intelligence to help a stroke survivor communicate in Spanish and English for the first time.

Nearly a dozen scientists at the university’s Center for Neural Engineering and Prosthetics worked for several years to design a decoding system that could transform human brain activity into sentences in both languages ​​and display them on a screen.

One article published on May 20 in Nature Biomedical Engineering Outlining his research identifies the man as Pancho. At age 20, he became severely paralyzed as a result of a stroke he suffered in the early 2000s. Pancho can moan and grunt, but cannot articulate clear words. He is a native Spanish speaker who learned English as an adult.

Under the leadership of Dr. Edward Changneurosurgeon who serves as co-director of the Center for Neural Engineering and Prosthetics, Pancho received a neural implant in February 2019allowing scientists to begin tracking his brain activity.

Using an AI method known as a neural network, researchers were able to train Pancho’s implant to decode words based on the brain activity produced when he tried to articulate them. This AI training method essentially allows the brain implant, known scientifically as a brain-computer interface device, to process data in a way somewhat similar to the human brain.

In 2021, technology significantly helped restore Pancho’s ability to communicate, but only in English.

“Speech decoding has been demonstrated primarily for monolinguals, but half the world is bilingual, with each language contributing to a person’s personality and worldview,” Chang’s research group said in X. “It is necessary to develop decoders that allow bilinguals to communicate in both languages.”

However, the 2021 research served as the basis for developing the decoding system that later made Pancho’s brain implant bilingual in Spanish and English.

Allowing a language change based on preference

After discovering that Pancho’s brain had “cortical activity” in both languages Years after he became paralyzed, scientists realized they could leverage this to train a bilingual brain implant without the need to train separate language-specific decoding systems.

“We leveraged this discovery to demonstrate transfer of learning across languages. Data collected in a first language could significantly speed up training a decoder in the second language,” Chang’s research group said in Xbecause is based on brain activity produced by “the intended movements of the participant’s vocal tract, regardless of language.”

In 2022, scientists sought to prove this. They again used the artificial neural network to train Pancho’s brain implant on the distinct neural activity produced by his bilingual speech.

According to his findings, Pancho was able to use the bilingual decoding system that powers his brain implant to “participate in a conversation, switching between [both] languages ​​based on preference.”

Ultimately, the study shows “the feasibility of a bilingual speech neuroprosthesis,” or bilingual brain implant, and gives insight into how this type of technology has the “potential to restore more natural communication” among bilingual speakers with paralysis, according to the May 20 report. article.

For more information from NBC Latino, sign up for our weekly newsletter.

This article was originally published in NBCNews. with