Dyslexia, the underlying biology of which is not well understood, affects somewhere between 5% and 17% of the population. About one-fifth of dyslexics will eventually go on to read as well as a typical reader, while others will need extensive intervention. Now a new study out of Stanford University uses artificial intelligence to predict who might fall into which group.
Dyslexics do not, as many people think, see everything backwards; they have a problem with reading, particularly with reading aloud, often because they have difficulty recognizing the letters or group of letters associated with a sound. Typically they may see a “d” when there’s a “b” or be unable to distinguish words like “for” and “from.”
Sometimes this disability does not hold them back — Tom Cruise, Richard Branson, Whoopi Goldberg, Steven Spielberg and Charles Schwab are among the many famous and accomplished dyslexics. Other people, especially those who go undiagnosed, flounder in school and struggle in life. So far nobody has been able to explain the difference. “It has been a mystery why 20% or so of the dyslexic children learn to compensate by adulthood,” says Stanford psychiatrist Fumiko Hoeft. (More on Time.com: Study: Breast-Feeding Improves Academic Performance, Especially for Boys)
Finding the other 80% early enough, and giving them intensive help with reading, could be a key part of treating dyslexia and illiteracy more effectively. At the moment, children are diagnosed through a battery of standardized tests. But while these tests can measure reading ability, they can’t predict whether or not a child with dyslexia will be able to compensate for or overcome his or her difficulties in time.
In a study published Monday in the Proceedings of the National Academy of Sciences, Hoeft suggests that brain imaging could be a much more effective diagnostic tool than reading tests. She gathered data on 25 adolescent dyslexics, using standardized tests and two different kinds of brain imaging scans. Two and a half years later she would look at the children again. (More on Time.com: Free Play Won’t Make Your Child Smarter)
In the meantime, she took the data from the brain scans and crunched them through an algorithm used in artificial intelligence (AI) to program devices like smartphones to “learn.” Using the same predictive methods as in AI, she discovered that the presence of activity in certain parts of the brain were far more accurate in predicting whether a dyslexic would learn to read without major intervention. (More on Time.com: 10 Questions with Ray Kurzweil)
Her studies found that dyslexic readers who show more activity in the inferior frontal gyrus region of the right hemisphere, the bit right near the temple, while reading were more likely to eventually read as fluently as non-dyslexics. The corresponding part of the brain on the left hemisphere, it’s believed, is usually what is doing the work when we read. Hoeft found that when the left hemisphere doesn’t carry its load, if the right side of the brain steps up to compensate, the dyslexic child will eventually read normally 90% of the time. If not, it’s time for Plan B.
All of this, of course, raises a whole lot more questions: can we stimulate the brain, or use feedback systems to get it to compensate? (Hoeft says maybe.) If a dyslexic’s brain is not predisposed to compensate, are any of the extensive — and expensive — interventions currently being practiced going to make any difference? (Hoeft says her studies can’t tell what’s genetic and what’s been learned, so she can’t say.)
And finally, the big question for parents who want to spare their children the anguish of dyslexia, can we diagnose the condition any younger? Hoeft thinks that might be possible. “Although we still have to do the research and likely modify our models, it is the hope that we can identify children who will later develop dyslexia much earlier and more accurately,” she says, “so we can provide necessary interventions A.S.A.P.”