Early Prediction of Alopecia Areata Using Machine Learning Modeling of Neuro Stress Immune Signatures from Multiple Datasets

This study aimed to develop a predictive model for alopecia areata (AA) onset using machine learning on datasets from the Gene Expression Omnibus. By analyzing six AA-related datasets, the researchers identified key feature genes (KRT83, PPP1R1C, PIRT) and constructed predictive models using five machine learning algorithms. The XGBoost model was found to be the most effective, with SHapley Additive exPlanations (SHAP) used to interpret its predictions. The study highlights the role of tissue regeneration, immune dysregulation, and neuro-stress-immune interactions in AA pathogenesis. An online predictive tool was also developed, offering a clinically applicable method for early AA prediction.