Using technology that links the brain and the computer, a person unable to move managed to transcribe his thoughts in writing.
- This is possible thanks to the brain’s ability to store neural activity in memory that activates movement even if the person is no longer able to move.
- The error rate reached 3.4%, a rate similar to state-of-the-art speech recognition systems.
- The system allowed the participant to write at a speed of 90 characters per minute.
A big step has just been taken. American researchers at Stanford University have managed to develop a technology that has enabled people paralyzed, due to spinal cord injury or neurological disorders, to be able to transcribe their thoughts in writing. In a study published on the 12th in the journal Nature they describe how they managed to re-establish communication with a person unable to move.
Two electrodes placed on the patient’s brain
Researchers have developed a technology imagined in the 1970s: the BCI, to Brain Computer Interface, a brain-computer interface. It is a direct connection system between the brain and a computer thanks to which it is notably possible to write letters, like on a computer keyboard, without having to use one’s hands. This is possible thanks to the ability of the brain to retain in memory the neural activity that activates the movements even if the person is no longer able to move, due to an accident or an illness.
The research was conducted with a 65-year-old volunteer who was paralyzed from the neck down. He was asked to copy 26 lowercase letters and several punctuation marks such as “>”, to indicate a space between two letters, and “~” to indicate period and the end of the sentence. To do this, the researchers placed two electrodes on the part of his brain associated with the movement of his right hand. The latter recorded the brain activity of about 200 individual neurons which reacted differently as he mentally wrote each individual character.
90 characters per minute
On the computer side, an algorithm of machine learning was designed to automatically recognize the neural patterns that the participant’s brain produces for each letter. After several tests, the error rate reached 3.4%, which is similar to state-of-the-art speech recognition systems. “This method is a marked improvement over existing communication BCIs which rely on using the brain to move a cursor to write words on a screen.”, rejoiced Frank Willett, researcher at Stanford University and lead author of the study.
The system allowed the participant to write at a speed of 90 characters per minute. The researchers welcomed this result, indicating that a person of the same age, ie 65, writes an average of 110 characters per minute on their smartphone. “By using handwriting to record from hundreds of individual neurons, we can write any letter, and therefore any word which provides a truly open vocabulary that can be used in almost any life situation.”, welcomed Doctor Shenoy, who supervised the study.
The next step for the researchers will be to test the system on a patient who has lost the ability to speak. They will also study how to complete the system by increasing the number of characters available, adding capital letters and numbers. Finally, they would like the patient to be able to edit or delete text.
.
