Brain implants and AI help voiceless patients to speak up

Developments in existing tech

The computer system, called brain-computer interface (BCI), is not new. Scientists have used it to acquire and analyze brain signals and translate them into commands, which are then relayed to an output device to carry out a desired action.

In 2019, Edward Chang, part of the research team, translated the brain signals of five epilepsy patients using a device with the help of electrodes and AI. The study was conducted with research participants with intact speech. In 2021, Chang and his team decoded the brain signals into the text of a man who had experienced a brainstem stroke. Since 2019, the research team has come far.

“Our goal is to restore a full, embodied way of communicating, which is really the most natural way for us to talk with others,” said Chang. “These advancements bring us much closer to making this a real solution for patients.”  

The research used phonemes

For the 2023 study, Chang implanted 253 electrodes attached to a piece of rectangular equipment onto the surface of a woman's brain with paralysis. The electrodes intercepted the brain signals and sent them to a bank of computers. Thus, the BCI.

Chang and his team trained the system’s AI algorithms to recognize her brain signals for speech and created a system that decodes words from phonemes - the building blocks of speech. For example, the word “Hello,” has four phonemes: “HH,” “AH,” “L” and “OW.”

The system’s accuracy and speed were enhanced as it only needed to learn 29 phonemes. But now the team had to create her voice. So they used a recording of her speaking at her wedding and trained a newly developed system to synthesize her speech. Then, the team created her facial animation using her facial expressions and the movement of her facial muscles.

The team built a customized machine learning system that combined their software to create facial expressions in the woman’s avatar as her brain signals were intercepted and converted into speech.

The next step for the team is to create a remote system without any wires, which wouldn’t have to be connected to the BCI.

Making major strides

In a separate case study by researchers at the Stanford School of Medicine, scientists gave a woman a way to express her thoughts using text. Two small implants were attached along the surface of 68-year-old Pat Bennett’s brain over speech-related regions in 2022. 

Bennett, who suffers from Amyotrophic Lateral Sclerosis (ALS), lost her ability to speak and can’t use the muscles of her lips, tongue, larynx, and jaws to enunciate phonemes.

Though Bennett understands these phonemes and her brain can direct her muscles to pronounce them, her muscles cannot carry out the command due to ALS.

About a month after Bennett’s surgery, the researchers trained the model to decode her incoherent speech into words. After four months, Bennett’s unintelligible words were converted into words on a computer screen at the speed of 62 words per minute. 

“We’ve shown you can decode intended speech by recording activity from a very small area on the brain’s surface,” said Jaimie Henderson, MD, the surgeon who performed the surgery on Bennett.

While the benefits of this technology could help those who have lost their speech due to an illness or a stroke, there’s some worry regarding the brain chips. There’s a clamor that these chips can read people’s thoughts. Elon Musk’s Neuralink is one such technology that has brought to the fore the skepticism around the further development of this technology. However, these views are unsubstantiated.