Title : Ai-powered prediction of eczema severity and flare risk in pediatric atopic dermatitis using blood biomarkers: A proof-of-concept study
Abstract:
Atopic Dermatitis (AD) is a chronic inflammatory skin condition with a significant pediatric burden. Traditional assessment tools, such as the Eczema Area and Severity Index (EASI), though widely adopted, are time-consuming and prone to inter-observer variability. This study presents a novel machine learning-based web application aimed at objectively predicting eczema severity and potential flare-ups using two blood-derived inflammatory markers: Eosinophil Relative Count (ERC) and Eosinophil-Lymphocyte Ratio (ELR). A dataset of 25 pediatric patients diagnosed with AD was used to train and validate a Random Forest Regressor for EASI score prediction and a classification model to assess flare-up risk—defined as a ≥20% increase in ERC or ELR. The model's performance was benchmarked against linear regression, with the AI approach demonstrating improved accuracy and robustness. The application empowers clinicians with real-time, objective tools to monitor disease progression and tailor treatments based on inflammatory profiles. This proof-of-concept highlights the potential of integrating AI with dermatological biomarkers to streamline pediatric eczema management and reduce diagnostic subjectivity, laying the foundation for more extensive multicenter validation studies.
