Little by little, science is getting closer to fulfilling the dream that those who cannot, can. Hand in hand with artificial intelligence, machine language, algorithms and a lot of technology, neuroscience is bringing closer the day when spinal cord injured people can walk; that people with Parkinson’s stop blocking themselves; that those who lost sensitivity they can play again; or that those who lost their sight, feel the light again. Now, American neuroscientists have designed a brain-machine interface (BCI) that allows you to communicate by typing on a keyboard with just your mind. As detailed in the scientific journal Nature Neurosciencethey have tested it successfully two patients with paralysis, who have managed to type at great speed and almost without errors. There are only two, the system is in its beginnings, but it brings that dream a little closer.
At one end of the BCI there are some plates with hundreds of microelectrodes placed directly in the brain. But not anywhere, but in the areas of the motor cortex that previous and own work has identified as responsible for fine finger movement. At the other end, there is a screen on which a QWERTY keyboard appears. In between, a lot of science.
“It is not based on the trial and error method, but on a process of calibration in which the participant attempts to write a series of predefined sentences“, says the researcher at the Center for Neurotechnology and Neurorecovery at Massachusetts General Hospital (United States) and first author of the study, Justin Jude. “The machine learning algorithm then learns what patterns of neural activity (recorded) at the electrodes correspond to each of the 30 movements possible of the fingers,” he adds. The English QWERTY keyboard has 26 letters. The other four movements were for the period, the comma, the closing question mark and the space key.
After thirty exercises to train the system, the results are impressive. One of the participants, who has been called T18, a 48-year-old man with a spinal cord injury at the level of the cervical spine, managed to a mind-typing speed of 110 characters per minutealmost the same speed as people in his age group without tetraplegia. He only had to think about typing. The BCI was able to discriminate that extending the index finger was always to hit the R or the T, that lowering it on the keyboard corresponded to the F and raising it towards the palm indicated that T18 wanted to write the V or the B. The number of errors was only 1.6%.
“The The most frequent errors were those that occurred between keys controlled by adjacent fingers or between movements that controlled different keys with the same finger,” says Jude. To reduce them, they turned to language models that anticipated the real intention of T18, “similar to autocorrection when typing on a smartphone keyboard, in order to generate the phrase that the user probably intended,” adds the researcher, also from Harvard Medical School (United States).
The other participant in the trial, T17, is a 33-year-old man with amyotrophic lateral sclerosis (ALS) who did not write as quickly, and did so with more errors. His lower performance could have been due to the state of his illness, with tetraplegia, the need for mechanical ventilation and anarthria, the total inability to articulate words caused by the inability to activate the speech muscles. But it could also be that they used fewer electrodes, 128 compared to 384 in T18, which would have affected their precision, something already foreseen in the study design. Still, despite his captivity syndrome almost complete and the limitation of his BCI, T17 managed to write 47 words per minute and not those of a dictation, but what he wanted to say.
T17 and T18 are two of several patients who are participating in a larger project called BrainGatewhat are you looking for all possible approaches to facilitate communication for those who cannot communicate. One such approach was to ask the participant to think about writing. The results of that one, which they called T5, were very good, achieving manual writing of 90 characters per minute thanks to two chips in the area of the brain dedicated to writing. The clinical trial is ongoing with nine participants across the United States.
The same inserts with those 384 microelectrodes are used by Eduardo Fernández, director of the Bioengineering Institute of the Miguel Hernández University of Elche. But on that basis, his team is designing BCI not for typing, but for those who have lost their sight, see again. As he recalls, “one of the most serious and challenging problems associated with some neurological disorders such as ALS and strokes is the loss of communication capacity caused by a lack of strength in the muscles responsible for speech or those for the mobility of the arms, hands and fingers.” On many occasions, as in the cases of T17 and T18, “Intellectual capacity and thinking usually remain intact“, highlights the scientist. And that is what we must try to take advantage of.
“This study introduces a new strategy to facilitate communication for these people,” says Fernández. “Unlike other approaches in which subjects are asked to imagine the movement of a cursor, the experimental paradigm is simpler, since They only have to think about moving a finger on a regular computer keyboard. In this way, users simply have to try to move their own fingers as if they were in front of a physical keyboard,” he adds.
Although there are only two participants, what is relevant for Fernández, who was not involved in this study, is that “it confirms that the neural representations for fine motor skills remain intact in the brain, even years after paralysis.” And for those concerned about external mind control, the neuroscientist settles the issue: “This type of technology is not capable of mind reading in the sense of extracting information from a subject involuntarily; “What it does is allow users to communicate by using brain signals instead of muscles.”
