AI Technology Allow Paralyzed Man To Type By Imagining Handwriting

By Anthony K

For a long time, humans have been fascinated by the thought of a device that can read their minds. Today, this is a reality; scientists have developed a system that combines a brain-computer interface with machine learning to read the handwriting inside the brain instead of on paper.

Image Courtesy of NPR

A group of scientists implanted sensors on a 65-year-old paralyzed man to detect neural activity relating to writing by hands. The 65-year-old man became paralyzed in 2007 after a spinal cord injury. He volunteered for the research 9 years after, at the Howard Hughes Medical Institute and Stanford University.  Following years of development, the man can now be connected to the system and, with the help of it, write up to 18 words in a minute with an accuracy of up to 94%. With the addition of an autocorrect feature to the system, the system records a 99% accuracy.

Image Courtesy of ZME Science

Details of the experiment are detailed in Journal Nature. The study, co-authored by a Howard Hughes Medical Institute investigator at Stanford University named Krishna Shenoy, asserts that the techniques used seem to be potentially more powerful than identical studies that have tried translating speech instead of handwriting.

Image Courtesy of Yahoo

By using handwriting to record from hundreds of distinct neurons, it’s possible to write any letter and thus any word that offers a vocabulary that can be applied in a majority of situations in life.

Such a system offers the hope of making it possible for paralyzed individuals to communicate and type effectively without utilizing their hands.  Already, the system’s proof of concept has been done with one patient.