It’s called mindwriting, or mental writing, and it’s a new software that could revolutionize the written translation of thoughts by offering new communication opportunities to people who have lost the use of their arms. Developed by scientists from Stanford University, the innovative system was featured in the journal Nature. The researchers paired artificial intelligence software with a device, a brain-computer interface, or BCI, implanted in a man with total body paralysis, called T5. The program was able to decode the information obtained from the BCI and convert the participant’s thoughts into digital writing. This method, the authors point out, allowed the patient to write at twice the speed of previous instruments designed by the same team and described in an article published in 2017 in the journal eLife.
Experts hope these discoveries could lead to further advancements that benefit millions of people who have lost the use of their upper limbs or the ability to express themselves due to spinal cord injury, stroke, or amyotrophic lateral sclerosis. “Our approach – says Jaimie Henderson, professor of neurosurgery at Stanford University – has allowed a paralyzed person to copy a text at a speed of about 18 words per minute, a speed almost comparable to that characteristic of able-bodied individuals of the same age. , which is about 23 words per minute. ” As for free writing, however, which required small reflection intervals, T5’s composing speed was about 15 words per minute. The team placed two BCI chips on the left side of the participant, who in 2007 lost the motility of every part of the body below the neck. Artificial intelligence algorithms, designed at Stanford University’s Neural Prosthetics Translational Lab, decoded the signals received by the chips and hypothesized the hand movement thought by T5.
“The brain – adds Frank Willett, a researcher at the Howard Hughes Medical Institute – retains the ability to direct movements until a decade after the body has lost the ability to perform those actions. We also know that complicated intentional movements that involve movements quick and fast, such as handwriting, can be interpreted relatively easily and efficiently by artificial intelligence algorithms. ” The ability to interpret the movement associated with the various characters have in fact improved markedly in recent years.
The phrasing error rate for T5 was approximately one every 18.5 characters, while in free composition it was one every 11.5 characters. Using an automatic correction function, the error rate dropped to one and two percent for copying and composing, respectively. “These rates are quite low compared to other BCIs – concludes Krishna Shenoy, another author of the article and professor at Stanford University – we hope that our work will offer a possibility for immobilized patients, so that they can communicate more effectively. “.