A new machine learning-based tool is helping doctors identify children most likely to develop chronic asthma in the future with greater accuracy.
This is done by analyzing pre-existing health information in the electronic health record, according to an experimental clinical study conducted by a researcher from the American Regenstrief Institute, and published in the journal "Scientific Reports".
Asthma is one of the most common chronic diseases among children, but predicting which children with wheezing or early respiratory symptoms will later develop the disease is a challenge for doctors. Some children experience a reduction in these symptoms over time, while others require ongoing monitoring and treatment, making early prediction of risk levels crucial for improving healthcare.
The study evaluated a machine learning-based clinical decision support tool, known as the "negative digital index," which analyzes routinely collected electronic health record data, such as respiratory symptoms, allergies, medication use history, respiratory infections, and family history, and then categorizes children into two groups: high-risk and low-risk for developing chronic asthma.
The tool does not require additional tests or new questionnaires, as it relies on the information already in the child’s health file, and provides the doctor with a simplified assessment that helps him make the clinical decision.
Dr. Arthur H. Owora, a research scientist at the Regenstriff Institute and lead author of the study, said the tool is not intended to replace the medical assessment performed by a pediatrician, but rather to help collect and analyze years of available health information in the electronic record, to provide a clearer picture of a child's risk of developing asthma.
The study results showed that pediatricians who used this tool were able to predict future asthma with greater accuracy than those who relied solely on traditional assessment, achieving an accuracy rate of 83% compared to 61%. The greatest improvement was in the doctors' ability to identify children who later developed chronic asthma.
The researchers emphasized that the goal of developing the tool is to support medical decisions, not replace clinical expertise, by helping doctors take advantage of large amounts of accumulated health information and transform it into a clear and easy-to-use assessment.
They pointed out that the study was based on standardized patient cases and not on direct application in clinics, so more research is still needed to determine the extent to which the use of this tool improves patient outcomes within the daily practice of pediatrics.
