ObjectiveThe upward trend in using plant materials introduced essential oils (EOs) as a valuable, novel, bioactive antifungal agent and as an alternative to standard treatment protocol of denture stomatitis caused by Candida species. Therefore, the aim was to evaluate the antifungal activity of different EOs and to present the response surface methodology (RSM) and artificial neural network (ANN) as possible tools for optimizing and predicting EOs antifungal activity. MethodsMinimum inhibitory concentration (MIC) and Minimum fungicidal concentration (MFC) of the EOs against 3 species Candida spp. (C. albicans, C. tropicalis, C. glabrata) isolated in patients with DS were determined, together with optimization and prediction based on non-terpene and terpene content in EOs, using two mathematical models RSM and ANN-GA. ResultsThe highest concentrations of EO M. alternifolia inhibited (1.6–2.8 μg/ml) and fungicided (3.5–6.0 μg/ml) all three investigated Candida spp. while the lowest concentrations of EO C. limon inhibited (0.2 – 0.5 μg/ml) and fungicided (0.6–1.1 μg/ml). Among the three types of Candida, C. glabrata was the most sensitive. The RSM modelling proved that MICs and MFCs statistically depend on non-terpene and terpene content in different EOs (<0.0001). Both models showed that a citrus oil (EO C. limon) with 89% content of terpenes and limonene as major constituent was more antifungal efficient. ConclusionsThe investigated EOs showed a broad spectrum of anticandidal activity, also confirmed using the RSM and ANN-GA models. Since EOs can be cytotoxic in higher concentrations, models may be used for qualitative and quantitative dosage predictions of the antifungal activity of EOs.