(Bloomberg) — Last August, Casey Harrell spoke the first clear words his five-year-old daughter could remember: a reiteration of his wedding vows to her mother, Levana Saxon. The adults in the room wept.
Most read by Bloomberg
This moment has been made possible by a wave of innovation in one of the most challenging areas of medicine: restoring the connection between the brain and the body when something—an accident or a disease—has broken that connection. While Elon Musk’s Neuralink Corp. is getting most of the attention and investor money in this area, academic labs and competing startups are making significant progress in restoring that broken connection.
“I’m using this in a very practical way right now,” said Harrell, 46, who was diagnosed with amyotrophic lateral sclerosis in 2019 and lost his ability to speak clearly a few years later. “This is real life for me,” he said, gasping as he described the day he was able to communicate with his own voice again.
His ability to speak is thanks to 256 tiny electrodes that researchers at the University of California, Davis, implanted in his brain last summer in a nearly five-hour surgical procedure. While the technology, known as a brain-computer interface, is often aimed at restoring the ability to move, the improvement in Harrell’s speech, described in the New England Journal of Medicine, underscores its greater potential.
Start-ups are proliferating because groundbreaking discoveries are revealing the complex signals that the nervous system uses to control the lips, jaw, tongue and larynx. Advances in artificial intelligence make it possible to decode these signals and thus restore communication. Similar to a prosthesis that replaces a missing body part, the industry has its own name for the device: speech neuroprosthesis.
“Given the complexity of language, the idea of restoring speech initially seemed unattainable,” wrote Edward Chang, chief of neurological surgery at the University of California, San Francisco, in an editorial accompanying the study. “Over the past decade, the concept of a speech neuroprosthesis has evolved from science fiction to reality.”
It’s still early days. The technology is expensive and bulky, requiring computers in Harrell’s home. It’s slow, helping him speak 33 words per minute, far less than the 160 words per minute of natural speech. And long-term performance is still unknown: Dutch researchers have described in the same issue of the journal the slow decay of a similar device used for seven years by a woman with severe ALS paralysis.
Destruction of nerve cells
Shortly after his daughter was born, Harrell learned he had ALS, also known as Lou Gehrig’s disease. The disease destroys nerve cells in the brain and spinal cord, leaving patients unable to control their muscles. His natural speech sounds like grunting or babbling to the untrained ear, although his mind is still sharp.
But today he is easy to understand. Electrodes in his brain track the activity of his neurons to predict what he wants to say. Then a speech generator, using a digitally reconstructed version of his pre-ALS voice, speaks for him – to the delight of all involved.
Rather than making him feel like a cyborg, the technology improves his interpersonal relationships. He can have conversations with friends that would have been unthinkable a year ago. His conversations with his wife go beyond purely business-related aspects. The implant gives him a sense of normality.
“People like me don’t often feel like doing something like this,” he says.
Electrical connection
Companies such as Neuralink, Paradromics Inc., Synchron Inc. and Precision Neuroscience Corp. are working on similar technologies. Neuralink’s device was first implanted in a paraplegic man earlier this year. He described the results as overwhelming. Synchron has fitted several patients with implants, although they do not contain as many electrodes.
Harrell’s electrodes sit in his precentral gyrus, a pathway involved in language that runs roughly from ear to ear across the top of the brain. They were implanted during a five-hour surgery in July 2023, weeks after a relative heard about the technology at UC Davis, where researchers were looking for their first patient.
“On the one hand, you don’t want to rush into brain surgery,” Harrell said during an interview at his home in May. “On the other hand, the risk you would be taking with this disease is definitely higher.”
Harrell, who uses a wheelchair, lacks the motor control to operate a traditional mouse. Before the surgery, he used a head mouse to type texts and emails, which he can now compose using the device.
“It gives him the power to say what he wants to say, when he wants to say it,” said David Brandman, the neurosurgeon at UC Davis who implanted the device. “It’s a whole new level.”
Track movement
Given all the jokes and observations, it seems as if the BCI can read Harrell’s mind. But that’s not the case, according to Brandman and his colleague Sergey Stavisky, who co-direct the Neuroprosthetics Lab at UC Davis.
Instead, it tracks the attempted movements associated with the words he’s trying to say. As the brain tries to activate muscles in the tongue, larynx and throat – instructions that are held up as Harrell’s neurons die due to ALS – the implanted electrodes pick up the messages.
The signals are processed and decoded by four computers running in his Oakland apartment. That facility may get smaller over time, says Nicholas Card, a postdoctoral fellow in neurosurgery at UC Davis and lead author of the study.
A recurrent neural network first predicts the likelihood of which phonemes, or sound units, he is trying to pronounce. A preliminary language model assembles these sound units into possible words and phrases, then a second, more refined language model creates the most likely group of words. After a phrase is decoded, Harrell can play an audio version in his own voice.
It doesn’t always work perfectly: When a Bloomberg reporter was there, it mistakenly displayed “dirty” instead of “nerdy.” In those cases, Harrell can tell it to try again by pressing a button on the screen using a wearable eye tracker from Tobii AB.
Improved accuracy
Rudimentary versions of this technology date back to 2004, when Matt Nagle, a Massachusetts man who was paralyzed after a knife attack, first had it implanted in his brain. While innovations since then have focused primarily on speed, the real breakthrough in Harrell’s device is accuracy.
After less than two hours of training, most of which focused on a vocabulary of 125,000 words, the BCI’s accuracy was 90%. Within a few days, this increased to 95% and later reached 97%.
Although his decryption progress is slower, it is enough time for him to attend talks at home and help plan strategies for his employer, the Sunrise Project, an Australian non-profit organization that works on climate change projects that also target financial institutions.
The commercial potential of the technology is limited by the small number of patients and the high costs. But that could change, and investors are watching developments.
“We constantly follow developments in research, especially at universities,” said Konstantine Buhler, a venture capitalist at Sequoia Capital, who praised UC Davis’ “very impressive” work. “Academic institutions are well positioned to fund science and research, while venture capital is most helpful when a technology moves from ‘research’ to ‘development.'”
Harrell’s device stands out in one respect: After training, his wife can operate it herself. This is in contrast to many implants that require medical professionals to operate and are used primarily in laboratories.
So far, he’s the only patient to receive the UC Davis technology, which connects via ports on his head. It leaves him vulnerable to infection and can be unsettling for those seeing the juxtaposition of wires and brain for the first time. Harrell says none of that matters.
“People like me don’t have time to wait for the perfect product,” he says. “For now, it’s good enough for me.”
(Updated with duration of operation in fourth paragraph.)
Most read by Bloomberg Businessweek
©2024 Bloomberg L.P.
