OBJECTIVES: Modeling techniques are widely used in pharmacoeconomics studies. Computational intelligence (CI) is an example of modeling approach successfully applied in various areas of science and technology. The aim of the study was to assess the usefulness of CI tools in pharmacoeconomics analysis. METHODS: Database contained 100 patients with Non-small Cell Lung Cancer (NSCLC) in IIIB and IVth stage. Every patient was described by 30 features (pharmacotherapy and diagnostics path). The pharmacotherapy characteristics included chemotherapy schemes based on cisplatin or carboplatin with vinorelbine, gemcitabine, etoposide and the additive therapy. The output value had binary characteristic (35 weeks of survival as a threshold). Data Mining Software WEKA was used. Support vector classifier (SMO), naive Bayes classifier (NB), and decision trees (RandomForest, J48) were applied. The quality of models was assessed based on their generalization abilities. The 10-fold cross validation procedure was applied. RESULTS: The best results obtained for each one of above mentioned tools were as follows: SMO – 80% of all positive, 70% of good positive and 86% of good negative; NB – 69%, 70%, 68%; RandomForest – 75%, 68%, 79%; J48 - 74%, 57%, 84% respectively. Using best obtained models, the in silico tests with various chemotherapy schemes were applied. Simultaneously, the cost-effectiveness studies with modeled survival were performed as the effectiveness measure of simulated in silico chemotherapeutic schemes. The results confirm the literature information about the clinical and economical efficacy of abovementioned chemotherapy schemes (i.e. no statistical significance in clinical outputs between vinorelbine – cis-platine and gemcitabine – cis-platine but the vinorelbine based scheme was more cost-effective). The experiment with in silico cytostatics dose reduction from 100% to 0 showed that BSC therapy could be the valuable alternative for palliative chemotherapy. CONCLUSIONS: Computational intelligence was found to be powerful and flexible tool allowing reliable models creation to perform in silico search for optimal therapy.