Predicting the risk of drug-induced urticaria in patients with an allergic history using artificial neural networks

Background: Drug-induced urticaria is a frequent hypersensitivity reaction. Identifying individuals at risk is crucial for clinical decision-making. Artificial neural networks (ANNs) offer a promising approach to predicting adverse drug reactions in allergic patients. Methods: We conducted a retrospective analysis using a dataset of patients with known allergic history. Various ANN architectures were trained and validated to predict drug-induced urticaria based on demographic, clinical, and pharmacological variables. Model performance was assessed using accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC-ROC). Results: The ANN model achieved high predictive accuracy, outperforming traditional statistical methods. Key predictive variables included previous allergic reactions, drug type, and comorbidities. The model demonstrated robust generalizability in external validation. Conclusions: ANNs provide an effective tool for predicting drug-induced urticaria in allergic patients. Their implementation could enhance personalized medicine strategies and improve patient safety. Further prospective studies are needed to confirm these findings in broader populations.