Write just thinking? A team of researchers from Stanford University (USA) has presented software that translates a common idea in the field of science fiction to reality: a microchip inserted into the brain is capable of decoding the brain signals related to handwriting and a neural network of tiny electrodes about the size of a pill inserted into the brain translate this mental writing into text that is output on a computer screen.
This new brain-machine interface (BCI) can convert ‘mental writing’, the act of imagining yourself writing on a sheet of paper, into text on a screen in real time.
BrainGate’s innovative system of being able to translate what the brain thinks into text can ultimately point to ways to help people with disabilities like paralysis communicate using only their thoughts.
In fact, a 65-year-old man had two grids of tiny electrodes implanted on the surface of his brain designed to read electrical activity in the part of the brain that controls hand and finger movements. Although the patient was paralyzed from the neck down, he imagined writing letters gently with his hand. Then, using an algorithm, the researchers discovered the neural patterns that went with each imagined letter and transformed those patterns into text on a screen .
Therefore, solely with his brain activity, the patient was able to produce 90 characters, or 15 words per minute; in essence, the average writing rate of people of the participant’s age when using a smartphone, according to Krishna Shenoy, a researcher at the Howard Hughes Medical Institute of Stanford University in the journal Nature that collects the study.
“The big surprise is that even years and years after the spinal cord injury , where you have not been able to use your hands or fingers, we can still hear that electrical activity. He’s still very active, “Shenoy clarifies.
This innovation could be of great help for those people who are completely paralyzed in the face of soon being able to access a new method of communication. Previous brain-machine interfaces were limited to a maximum of about 40 characters per minute, while the new handwriting decoding BCI can reach a writing speed of 90 characters per minute with an accuracy of 94% . This performance rivals that of healthy peers texting on their smartphones, as we’ve seen.
Simple operation
The participant had to imagine that he was writing sentences as if his hand was not paralyzed by imagining that he was holding and pressing a pen on a sheet of paper. The machine learning algorithm known as a neural network translated the specific brain activity associated with this movement into text in real time, tracing the path of this imaginary pen.
Drawbacks
Brain surgery is explicitly invasive. In addition, the machine learning system is not universal and has to adapt to the cognitive nuances of each user, which requires a “very computer-intensive” process. Although we are far from seeing it in the everyday life of any user, the potential of a system like this could one day transform the way we write forever.
Referencia: F.R. Willett et al. High-performance brain-to-text communication via handwriting. Nature. Published online May 12, 2021. doi: 10.1038/s41586-021-03506-2.
