Thought-to-text record has just been set with a new brain portal interface

In a new breakthrough in Brain-Computer Interface (BCI) technology , scientists implanted a chip in the brain of a 65-year-old man and helped him regain the ability to communicate. After nearly 15 years of paralysis.

The unnamed patient in the study was nicknamed T5. He was paralyzed from the neck down after a spinal cord injury in 2007. The accident deprived him of not only his ability to move but also his voice.

But now, with a BCI chip implanted in the brain, T5 is able to communicate via computer at a transmission rate of 18 words per minute, at an accuracy of 94-99%. This is the record that a BCI communication system has managed to date – approximately the average typing speed of the elderly on a mobile phone.

Brain-Computer Interface (BCI) technology.

Scientists really have the secret to getting to that number. Unlike previous brain-computer interfaces that only converted the patient's mere thoughts into characters, the scientists asked T5 to imagine he was holding a pen and writing.

The handwriting signals are decoded by the computer and converted into characters on the keyboard. With an AI system and an auto-correction system, this man will be able to communicate with near-perfect accuracy.

Part of the Brain Gate project

The new research is part of a large project called BrainGate , or " Brain Gate ". In it, scientists are looking for paralyzed patients and implanting intracranial microelectrodes (microelectrodes) into their motor cortex.

These intracranial microelectrodes can record signals fired from a patient's neurons, thereby reproducing them into signals that help them control a computer mouse, a wheelchair or even an arm robot outside his body by thinking.

In this new trial with patient T5, the scientists wanted to help the man regain his ability to hold a pen to write, even though his hands and all of his limbs have been paralyzed for more than 10 years. .

T5 had an intracranial microelectrode implanted with a computer, and while sitting still, he was asked to imagine himself picking up a pen again and writing each letter of the alphabet. An AI algorithm will try to record the signals emitted by the T5 brain and reconstruct them on the screen, decoding the letters, along with movements such as lifting the pen, spacing, dot, etc. comma or delete…

This paralyzed man wrote 18 words per minute using a brain-computer interface.

" In previous studies, we have known how well the brain retains the ability to control muscle movement, even though the paralyzed person's body has lost the ability to perform these movements. that behavior after decades ," said Frank Willett, study author and an expert in neural prostheses at Stanford University.

Willett's idea was that handwriting, a complex orbital movement, with fast speed and very subtle resolution could also be decoded by a brain-computer interface. For the elderly, their scribbling speed is even higher than the typing speed of the keyboard.

Therefore, if the BCI can translate these ingenious movement control thoughts in the paralyzed person's brain into text, they can communicate at a much faster rate than manipulating a mouse pointer.

Set communication record with brain-computer interface

That has been proven with T5. After connecting his brain to a computer, the patient was able to write at a rate of 90 characters per minute (equivalent to about 18 words). The accuracy that the BrainGate interface achieved this time is up to 94%, and if autocorrect is enabled, it can be up to 99%.

That's not only significantly faster than previous BCI tests (using a virtual keyboard and pointing the mouse), but it's also roughly on par with the typing speed of smartphone users in this age group – the old people can only type about 115 characters or 23 words per minute.

One advantage of handwriting in the alphabet is that each letter is very different in shape. Therefore, it is not difficult for AI algorithms to decipher them.

" We have learned that the motion intends complex involves changing the speed and trajectory curve, like handwriting, can be explained easily and quickly than by algorithms AI creation that we're using. And it's faster than even simpler intended movements like moving the cursor in a straight line at a steady rate ," says Wallett.

The success with patient T5 has opened up a new direction of research, offering great promise for people who have lost their voice and mobility.

As a next step in this study, Wallett and colleagues will continue to upgrade their handwriting decoding system, to recognize extended characters, such as capital letters, brackets and symbols complex.

The sensitivity of the brain gate will also continue to be upgraded. Wallett is recruiting new volunteers to participate in its trial, which promises to open the door to communication rehabilitation for even more paralyzed people.

The study was published in the journal Nature.