Neuroscientists have been able to translate the cognitive signals associated with writing text in real time. The new technique is faster than the previous method, allowing a paralyzed person to send a text at a rate of 90 characters per minute.
See also: AI: You need sleep like the human brain
The researchers in collaboration with BrainGate have developed a system that could eventually "allow people with severe mobility problems such as paralysis to communicate via text, email or other forms of writing", according to Jaimie henderson, his co-director Neural Prosthetics Translational Laboratory at Stanford University and one of its authors new study.
The signs brain thought-provoking and writing-related thoughts have been translated into real-time text, allowing a paralyzed person to send text at a rate of 16 words per minute. The system uses brain implants and a machine learning algorithm to decode the brain signals associated with writing.
The BrainGate consortium has made significant contributions to the development of brain-computer interfaces (BCIs) over the years, including an advanced brain-controlled robotic arm introduced in 2012. Work on the development of the new author brain-computer interface has been directed by Frank Willett, a researcher at Stanford University and supervised by a neuroscientist Krishna Shenoy from the Medical Institute Howard Hughes and Henderson.
In 2017, Shenoy and his colleagues developed a thought-text system that significantly improved prior art, allowing monkeys to send text at 12 words per minute.
The participant in the experiment was a 65-year-old man who had been paralyzed from the shoulders and down for the last 10 years after a spinal cord injury.
"Two sensors, each measuring 4 × 4 mm, about the size of an aspirin, with 100 electrodes of hair thickness, were placed in the outer layers of the motor cortex - the area that controls movement on the opposite side of the body.Henderson explained. «These electrodes can record signals from about 100 neurons"And the resulting signals"processed by a computer to decode the brain activity associated with writing individual letters».
During the experiment, the paralyzed patient tried to move his hand to write words. He thought that "wrote the letters on top of each other with a pen, on a yellow boardWhile a decoder typed each letter as "were identified by the neural networkSaid Henderson. The team used the symbol> to indicate spaces between words, "because otherwise there would be no way to locate the intention to leave a gap", He added.
The system was able to distinguish individual letters with approximately 95% accuracy. Henderson said the rate of 16 words per minute is about three-quarters of the speed typically seen in people over 65 when typing in smartphone their.
The results are promising, but the system has limitations. First of all, it is extremely invasive, as it requires brain surgery and implants. Also, the system must learn the cognitive nuances of each user. The new approach requires a "specialized high-performance computer or a set of computers". Finally, the system requires a technician to configure the brain-computer interface and run the software.
Despite these limitations, Henderson envisions a fully mature version that will be "wireless, always available and self-regulating." All of these goals are achievable, he said, but would require investing resources that could ideally be provided by a company rather than an academic lab.
Looking ahead, the team hopes to study how the brain coordinates movements and understand how speech is created by the brain.