Paralyzed volunteer can 'write' again thanks to mind-reading AIMarie Donlon | October 30, 2019
A fully paralyzed volunteer had his thoughts and handwriting come to life with help from mind-reading artificial intelligence (AI).
A team of researchers, led by Frank Willett, a biomedical engineer at the Cleveland Functional Electrical Stimulation (FES) Center at the Louis Stokes Cleveland VA Medical Center, has developed a computer model that can handwrite the thoughts of a paralyzed volunteer who is no longer capable of putting pen to paper.
Building upon related research where paralyzed volunteers have been implanted with electrodes for the purpose of enabling those patients to move a cursor to select letters on a display, researchers used the implanted electrodes to write the alphabet according to the thoughts and imagined motions of the paralyzed patient. Using a brain-computer interface, the volunteer was asked to imagine the motions involved with writing the letters of the alphabet. The associated brain activity trained the neural network, which is a series of algorithms that recognizes patterns, to decipher the commands and trace the imagined motions of the hand.
After training, the model could decipher the letters envisioned by the volunteer with a 95% rate of accuracy and at a speed of 66 characters a minute — for scale, average handwriting speed is 120 characters a minute.
The research team hopes that the technology will improve communications for fully paralyzed patients in the future.
The research appears in the Society of Neuroscience.