With evidence that computer software may be able to spot signs of autism in young children just as well as trained experts, researchers are looking to make the technology widely available via an app.
Current behavioral tests require a clinician to observe an infant’s response when a toy is shaken or a ball rolls toward them, for example, counting the seconds it takes for a child to react appropriately.
Now researchers at Duke University have developed software to track an infant’s response to such stimuli by analyzing eye gaze, walking patterns and motor skills. Rather than count seconds like a clinician, the software can track a child’s reaction time down to tenths of a second, they said.
In a small trial of 12 children ages 5 to 18 months, researchers found the software was just as good at flagging behavioral signs of autism as experienced clinicians. What’s more, the technology performed better than students in training and medical professionals without expertise in autism screening, according to findings published online in the journal Autism Research and Treatment.
The team of autism and computer experts behind the study now say they are looking to test an app based on the software. Optimally, they’re hoping that schools and parents could use the technology to screen children for autism and determine if a more extensive clinical evaluation is warranted.
“We’re not trying to replace the experts,” said Jordan Hashemi of Duke who worked on the study. “We’re trying to transfer the knowledge of the relatively few autism experts available into classrooms and homes across the country.”