February 9, 2023 – Autism might be detected in children almost from birth by reviewing their health records using an algorithm. a study by Duke University says.
“We can use the first 30 days of a child's medical care to say, 'This child is really at risk,'” said David Mandell, professor of psychiatry on the Perelman School of Medicine on the University of Pennsylvania, in USA today. He was not involved within the research.
Researchers analyzed the electronic medical records of 45,000 children who were treated as infants within the Duke University Health System between 2006 and 2020. They created an algorithm that might predict which babies later developed autism. These babies were more more likely to have seen an ophthalmologist or neurologist, had stomach or gastrointestinal problems, or received physical therapy.
“A large number of factors from the entire health profile of the infant” were incorporated into the models, said study co-author Dr. Matthew Engelhard, assistant professor of biostatistics and bioinformatics. “Each of these factors contributes incrementally.”
According to USA Today, the team “paid particular attention to the model's performance in groups of children who are often missed by traditional screening methods and therefore miss the benefits of early diagnosis. These include girls, children of color and children with a combined diagnosis of autism and ADHD,” Engelhard said.
The study could lead on to the algorithm getting used together with other tools to diagnose and help children earlier, said study creator Geraldine Dawson, who directs the Duke Center for Autism and Brain Development.
“We need to look at autism not just as a mental illness, but also as a condition that affects physical health,” she said. “This is one way to use this information: by improving early detection.”
Autism is an advanced condition that involves communication and behavioral problems with a spread of symptoms and skills. It generally is a mild disability or one which requires full-time care.