AI Algorithm Might Detect Autism Early Using Pupil Dialation, Heart Rate

Autism spectrum disorders, such as Rett syndrome, are usually diagnosed once a child is a few years old, but now, neuroscientists believe they have developed an AI system that can detect these conditions earlier in children by detecting abnormalities in pupil dilation and heart rate.

Abnormal Pupil Dilation and Heart Rate an Early Indicator of Autism?

Typically, a child is diagnosed with an autism spectrum disorder (ASD) once they are a few years old and behavioral, speech, and occupational therapies prove ineffective. Now, thanks to a machine-learning algorithm developed by neuroscientists in collaboration with Boston Children's Hospital, diagnosing ASDs like Rett Syndrome might happen much sooner than is currently possible.

RELATED: RESEARCHERS DEVELOP BLOOD AND URINE TESTS FOR EARLY AUTISM DETECTION

Michela Fagiolini, PhD, and Pietro Artoni, PhD, published a new paper in PNAS this week that can identify abnormalities in a child's pupil dialation and heart rates that can predict an ASD months, or even years, earlier than is currently the case.

"We want to have some readout of what's going on in the brain that is quantitative, objective, and sensitive to subtle changes," Fagiolini said. "More broadly, we are lacking biomarkers that are reflective of brain activity, easy to quantify, and not biased. A machine could measure a biomarker and not be affected by subjective interpretations of how a patient is doing."

The researchers, in collaboration with Takao Hensch, PhD, and Charles Nelson, PhD, from Boston Children's Hospital, began looking for predictive indicators in children's pupil dilation and heart rates on the hypothesis that persons with an ASD have altered behavioral states. Earlier work has found that the cholinergic circuits in the brain, which are partially responsible for arousal, are particularly affected, which in turn would lead to abnormal pupil dilation and heart rates.

The researchers measured the pupil dilation and heart rates of mice with the genetic mutations that cause Rett's Syndrome or CDKL5 disorder and found that the mice had abnormal instances of pupil dilation and heart rate well before behavioral symptoms of the disorder became apparent.

Further, restoring a working copy of the MeCP2 gene--the gene whose mutation is responsible for Rett Syndrome--to just the neurons in the cholinergic brain circuits prevented the development of pupil dilation abnormalities as well as behavioral symptoms of the disorder.

Identifying Rett Syndrome in Young Girls

After observing mice with the specific genetic deficiencies for about 60 hours, the researchers developed a neural network to detect the pupillary abnormalities. They then took this AI and tested 35 girls with Rett syndrome, with 40 girls without an ASD as a control. The AI was able to identify Rett Syndrome with about 80% accuracy in one and two year old girls.

"These two biomarkers fluctuate in a similar way because they are proxies of the activity of autonomic arousal," said Artoni. "It is the so-called 'fight or flight response."

Taken together with earlier tests that Fagiolini and Nelson developed for Rett syndrome, the researchers believe that these biomarkers could be an effective and affordable screening tool for infants that can alert parents and doctors to imminent developmental challenges.

"If we have biomarkers that are non-invasive and easily evaluated, even a newborn baby or non-verbal patient could be monitored across multiple timepoints," Fagiolini said.