epocrates logo
epocrates logo
epocrates logo
  • 0

Journal Article Synopsis

J Allergy Clin Immunol

AI closes the prediction gap for asthma after early childhood eczema

April 28, 2026

card-image

Clinical takeaway: For young children with early‑onset atopic dermatitis, validated machine‑learning risk scores may help clinicians identify those most likely to develop persistent asthma or allergic rhinitis and prioritize closer monitoring, early referral, and preventive strategies.

Atopic dermatitis in infancy is a well‑recognized entry point to the “atopic march,” yet clinicians have lacked reliable tools to predict which children will later develop asthma or allergic rhinitis. Investigators addressed this gap by applying machine‑learning methods to large‑scale real‑world health data.

In a retrospective birth cohort from Kaiser Permanente Southern California, researchers analyzed electronic health records for 10,688 children diagnosed with atopic dermatitis before age 3. They developed and validated two prediction approaches for moderate‑to‑severe persistent asthma and allergic rhinitis between ages 5 and 11: a comprehensive EHR model and a simplified model using routinely available clinical features.

For asthma prediction, both models showed strong discrimination, with areas under the curve of 0.893 (comprehensive) and 0.892 (simplified). At a high specificity threshold (95%), sensitivity reached 40.4% with a 39.3% positive predictive value (PPV) for the comprehensive model and 36.2% sensitivity with 33.8% PPV for the simplified version.

Performance was more modest for allergic rhinitis (AUC 0.793 and 0.773), but PPVs in higher‑risk strata were notably high—exceeding 70% at 90% specificity. Calibration was acceptable across models, with the strongest agreement seen in children classified as highest risk.

Overall, the findings suggest that machine‑learning tools built on early‑life clinical data could support more proactive, individualized management of children on an atopic trajectory.

Source: Chen, W, et al. (2026, April 17). J Allergy Clin Immunol. Machine Learning Prediction of Asthma and Allergic Rhinitis in Children with Early-Onset Atopic Dermatitis

Trending icon

TRENDING THIS WEEK

EPOCRATES CME

View Catalog

view all CME activities
learn more about epocrates plus

Clinical FAQs

Check out the answers to frequently asked questions about our clinical content.

Download Epocrates from the App StoreDownload Epocrates from the Play Store
About UsFeaturesBusiness SolutionsHelp & Feedback
© 2026 epocrates, Inc.   Terms of UsePrivacy PolicyEditorial PolicyDo Not Sell or Share My Information