AI ‘locked in’ person to write with his mind

engineering careers  AI ‘locked in’ person to write with his mind

A team of researchers have managed to double the speed at which paralyzed people can communicate by allowing them to simply ‘imagine’ writing with their hand.

The experiment aimed to help those with so-called ‘locked-in’ syndromes (tetraplegia, a form of complete paralysis normally caused by a stroke or neurological disease).

How can paralysied people communicate?

Up to now, fully paralysed patients have been able to communicate with the outside world in a number of ways.

They can use software which tracks eye movements to electrodes implanted in the parts of the brain involved in motion.

This modern approach can allow them to move a cursor onscreen and type using a virtual keyboard. However, even this is is slow with top speeds for typing remaining at only 39 characters a minute (three times slower than handwriting).

How the team used AI to allow paralysed people to write


The team instead took a different approach. Using the same tech of implanted electrodes to monitor brain activity they instead let volunteers train a computer program ( a netural network ).

The AI software was able to interpret their brain activity as written characters as they imagined moving their arms to write each letter of the alphabet.

The software was able to interpret the imagined motion and trace the intended trajectory of a pen tip to create letters on the screen.

The program was eventually able to read out whole sentences and achieved an impressive 95% accuracy rate at 66 characters per minute.

The next step for the team is to refine the tech to increase the speed at which people can write. This can be done by both refining the software but also by simply allowing the AI to learn more through practice.

The findings also have wider implications for how we understand the brain and how it processes fine motor movements.

Presented as “Motor cortical representation and decoding of attempted handwriting in a person with tetraplegia” at the annual meeting of the Society for Neuroscience.