New technology that translates text from implanted neural array signals to a computer aims to improve communication that’s ‘as effortless as speaking.’
A man who wasn’t able to speak since 2003 after a car accident left him paralysed at the age of 20 was able to express himself thanks to new revolutionary technology.
Neurosurgeons implanted electrodes on the brain surface of a 38-year-old man known as Pancho, and a software algorithm “learned” the patient’s speech patterns as he tried to speak.
Pancho, who the scientists called Bravo-1 for privacy reasons, was able to reach fifteen words a minute with 75 percent accuracy, enough for normal conversation.
The first full sentence he said was, “My family is outside,” scientists said.
A medical-first, the achievement was announced in a report on Wednesday in The New England Journal of Medicine.
"This tells us that it's possible," Dr Edward Chang, chairman of neurological surgery at University of California, San Francisco, who’s lab led the research, said.
"I think there's a huge runway to make this better over time,” he added, stating that normal speech is in the order of 120 to 150 words per minute.
People with paralysis commonly use devices that transcribe eye or head movements. Another commonly used technology is a device that allows leading a cursor with thoughts.
For the first time, a device allows brain activity to be measured into full sentences, without the need to spell words letter-by-letter. Chang Lab says their goal is to improve assistive neurotechnology by restoring communication that’s as effortless as speaking.
Chang Lab’s study took 81 weeks to complete.
Bravo-1, who had been communicating by pointing at a screen or through small head gestures, agreed to participate in the study three years ago. Scientists first implanted a 128-electrode array, but they were not sure if his brain’s mechanism for speech was still active, the New York Times reported.
But over the course of 48 sessions, they were able to record 22 hours of cortical activity as Bravo-1 attempted to say individual words from a vocabulary set of 50 words including “hungry,” “music,” “computer.” With only 50 words, he could build over 1,000 sentences.
They had him saying specific sentences such as “please bring my glasses,” and “my nurse is right outside.”
It was all possible thanks to an extensive team effort.
The project lead, David Moses’s real-time time decoding framework combined speech detection, word classification, and language modelling. Jessie R. Liu then trained the real-time detector to be able to recognise the attempted speech activity in the brain. The word classifier, developed by Sean Metzger, functioned as a kind of autocorrect language-prediction system similar to those available on smartphones.
In interviews with the Times over several weeks, Bravo-1 said the brain implant’s ability to recognise his words is “a life-changing experience.”
“I just want to, I don’t know, get something good, because I always was told by doctors that I had 0 chance to get better,” Pancho said during a video chat from the Northern California nursing home where he lives, using a head-controlled mouse that allows him to type only key by key.
“Not to be able to communicate with anyone, to have a normal conversation and express yourself in any way, it’s devastating, very hard to live with,” he said in an email to the Times.
“It’s very much like getting a second chance to talk again,” he wrote during the research sessions.