Asthma is the most common childhood chronic disease. The National Heart, Lung, and Blood Institute estimates that in an average classroom of 30 children, 3 have asthma. But how do you diagnose children who are not old enough to accurately describe their experiences, or demonstrate years of symptoms?
According to pediatric asthma researcher Arthur Owora, PhD, preschool and toddler years are often a missed opportunity for early asthma interventions.
“Many young children have wheezing or breathing symptoms, but it can be hard to know which children will outgrow those symptoms and which will go on to develop persistent asthma,” said Owora, associate professor of pediatrics at the IU School of Medicine and research scientist at Regenstrief Institute. “Our goal was to see whether a machine learning tool built into the electronic health record could help pediatricians make that distinction earlier.”
In a study recently published in Scientific Reports, Owora's team tested an AI tool called a Passive Digital Marker (PDM). The tool uses machine learning to evaluate electronic health data, and the researchers wanted to see if its analysis could improve a pediatrician’s ability to accurately spot children at high risk for asthma.
In a pilot randomized clinical trial, 34 pediatricians in Indiana evaluated the health data of 10 children from birth to age 3. Some of the doctors assessed a child’s risk of developing asthma independently, and others conducted the assessment with support from the PDM tool. The researchers then checked how accurately the doctors predicted which children would go on to have asthma between the ages of 6 and 11.
Owora's team discovered that pediatricians who used the PDM tool were significantly more accurate in predicting which children would later develop asthma, especially for children who were truly at higher risk.
“The tool uses information that is already in the child’s medical record, so families would not need extra blood tests, special lung testing or additional procedures for the tool to work,” Owora said. “The child’s medical history may already contain important clues. The challenge is that those clues can be scattered across years of visits. This tool helps bring them together in a way that is easier for clinicians to use.”
Owora said that while the tool will help clinicians identify children who need closer attention sooner, additional testing is needed. Next steps will involve using the PDM tool in real pediatric clinics during routine care to learn whether it helps children receive earlier diagnoses, more timely treatment and better long-term outcomes.
“Ultimately, my goal is for tools like this to support — not replace — clinicians, so that families get clearer answers earlier in a child’s asthma journey,” Owora said. “For families, the hope is fewer years of uncertainty and fewer missed opportunities to prevent worsening symptoms.”
IU School of Medicine’s Bowen Jiang and Yash Shah are co-authors on the study, and the research was supported by funding from the National Institutes of Health.
