Potential test to identify autism in toddlers

This is going to scare hell out of too many moms. They showed movies to a bunch of kids age 14 to 42 months. One side of the screen showed children dancing or doing yoga, and the other side showed a computer screen saver (in their terms, "dynamic social images" and "dynamic geometric patterns) and bounced an infrared light beam off the kids' eye to track what they were looking at.

• 2% of typical  kids preferred to look at the screen saver (1 out of 51)
• 9% of developmentally delayed kids preferred the screen saver
• 40% of kids already diagnosed with autism spectrum disorder preferred the screen saver. They also had a different pattern of eye movements, changing the gaze less often.

I wonder if it's dose related, meaning does the percentage get higher with more severe autism?

But using the figures as given, as a test for autism, this method gives 2% false positives and 60% false negatives. I'd say this falls into the category of interesting to know, and if you run across a kid who gets fixated on the center computer screen saver more than other kids, it would be worthwhile screening the kid for ASD.