For the first time in medical history, doctors have helped a man who was paralyzed and mute for years communicate again by translating his brain waves into speech.
The 38-year-old man, living in the US, is nicknamed BRAVO-1 in a study conducted by the University of California. Fifteen years ago, he suffered a stroke that severed the nerve connecting his brain to his vocal cords.
BRAVO-1 is also paralyzed from the neck down, so he could only communicate by moving his head. A team of engineers designed a special chair with a touchscreen connected to a computer in front of him.
Normally, BRAVO-1 would wear a baseball cap with a long stick attached, allowing him to tap the touchscreen and select words to speak.
But now, with an electrode implanted directly inside his brain, BRAVO-1 can speak without the cap. His neural signals are transmitted straight to a computer for decoding, allowing him to speak at an average speed of 15 words per minute with 74% accuracy.
At times, the speed can reach up to 18 words per minute, with accuracy hitting 93%. For comparison, an average person speaks around 150 words per minute with 100% accuracy.
This shows that decoding BRAVO-1’s brain waves into speech is still not perfect. But scientists say the very first word he spoke was “a major technological milestone for someone unable to communicate naturally.”
“It demonstrates the potential of this method to give a voice back to people with severe paralysis and speech loss,” said Dr. David Moss, lead engineer from the University of California, in an interview with the Washington Post.
Helping a mute paralyzed man speak again
To enable BRAVO-1 to speak, scientists implanted an electrode array on his sensory cortex, the brain region controlling speech. The electrodes connect outside the skull via wires to a decoding computer.
Next, they trained a program to recognize and differentiate the electrical signals from BRAVO-1’s sensory cortex, turning his thoughts into words. This involved 48 sessions totaling 22 hours.
During these sessions, scientists recorded brain signals as BRAVO-1 silently read 50 words flashing on a screen. Then, they used deep learning algorithms to identify which signals corresponded to which words he intended to say.
To speed up communication, part of the algorithm predicts the next word BRAVO-1 might say, similar to autocomplete or spellcheck on your phone.
Scientists noted that typical spellcheck accuracy is about 2%. In this study, they boosted accuracy to 93% with a vocabulary of 50 words. This small dictionary includes essential daily words for someone like BRAVO-1, such as “water“, “family“, and “good“.
In a demonstration, researchers asked him questions like “How are you today?” and “Would you like some water?”
BRAVO-1 took a few seconds to think and then typed answers on the screen like “I’m doing well” and “No, I’m not thirsty.”
“To our knowledge, this is the first evidence that direct decoding of brain activity from a paralyzed, non-speaking person into full words can be done successfully,” said neurosurgeon Edward Chang, co-author of the study.
Five years ago, no one expected this research to succeed
Chang, now chair of neurosurgery at the University of California, emphasized that five years ago, no one believed such results were possible.
Most similar studies back then involved brain-computer interfaces only for epilepsy patients, used to diagnose seizure origins.
Going back another five years, some researchers could decode sounds or syllables from brain waves, but algorithm accuracy was too low to form complete words.
The success Chang and his team achieved with BRAVO-1 builds on a decade of advances in artificial intelligence, including speech recognition and spellcheck algorithms trained on billions of hours of data.
In neural signal decoding, there have also been major breakthroughs. For example, in May, a Stanford team helped a paralyzed man write on a screen by decoding brain waves.
In that case, the man imagined moving his hand and fingers to write, rather than thinking of letters being typed or selected on a touchscreen like BRAVO-1.
Commenting on the new study, Christian Herff, an associate professor of neural engineering at Maastricht University in the Netherlands, called what Chang and his team achieved a “giant leap“. “It really solved a major problem,” Herff said.
Previous studies showed brain wave decoding could help read thoughts of people who can speak. “But this is the first study to do it for a patient who cannot speak,” Herff emphasized.
In an interview, Chang said this work is the result of a decade-long effort. Ten years ago, he met a mute patient and began reflecting on the challenges they faced.
Thousands of people suffer similar fates each year due to neurological damage from stroke, cerebral palsy, trauma, and diseases like ALS, famously affecting scientist Stephen Hawking.
“Every day, I see patients who lost their ability to speak after stroke or brain injury. Silence devastates their lives. After all, speech is part of what makes us human. Losing it is truly brutal,” Chang said.
But with BRAVO-1’s success today, he hopes to help patients regain communication. “And this is really just the beginning,” Chang added.
Next, Chang and colleagues will improve deep learning algorithms to boost BRAVO-1’s speech accuracy and speed. They also plan to expand his vocabulary so he can discuss more topics.
Meanwhile, Herff suggested the team could convert BRAVO-1’s language into actual speech sounds, allowing him to speak aloud like Stephen Hawking did, but with added intonation and expression.
They are also developing wireless communication tech between brain electrodes and computers. If successful, this could free BRAVO-1 from the wires behind his skull.
Wireless brain-computer communication could also help mute people who aren’t paralyzed, letting them speak without being tied to a computer.
“Although the research is considered a success, we don’t claim to have perfected this technology. This is truly just the beginning,” Chang said.
The study was published in The New England Journal of Medicine.
Source: UCFS
