OBJECTIVES: The aim of the present analysis was to optimize the prediction of treatment response in patients with allergic rhinitis. To determine an optimal prediction accuracy, we used different Pattern Recognition (PR) approaches and evaluated their added value compared to standard predictive models as Logistic Regression, and Linear Discriminant Analysis. METHODS: In order to optimize the prediction of treatment response, 76,981 case reports of patient with allergic rhinitis from ten post-marketing-studies in Germany were analyzed by means of PR methodology. The processing steps applied within this study are: (a) feature extraction (genetic algorithm based synthetic feature calculation), (b) dimensionality reduction (correlation filter based feature selection, wrapper based feature selection, Principle Component Analysis based feature transformation), (c) classification (Support Vector Machine, Decision Tree, K-Nearest Neighbor, Random Forest, Artificial Neural Network, Bagging, Boosting Logistic Regression, Linear Discriminant Analysis), and (d) validation (leave-one-sample-out cross validation). RESULTS: The AdaBoost Support Vector Machine classifier with correlation filter based feature selection achieved the highest unweighted mean recall rate (mean of sensitivity and specificity; URR) of 62.8%. The standard Logistic Regression approach yielded 50.4% (-12.4%), the Linear Discriminant Analysis 50.6% (-12.6%). CONCLUSIONS: In comparison to standard learning schemes as e.g. Logistic Regression, and Linear Discriminant Analysis applying advanced PR methods improves substantially the prediction of treatment response in patients with allergic rhinitis. Due to the achieved added value and the superiority of Pattern Recognition methods within several benchmarking studies, advanced PR methods should be primarily considered for modeling and prediction tasks within the field of pharmacoeconomics.