Facial skin is particularly exposed to external factors and as such is subject to various changes that affect its health. The most important monitoring parameters whose values indicate skin condition are skin pH, sebum and transepidermal water loss, compared to age and sex. Artificial neural networks are computer models that were created by the model of the structure and functioning of neurons. They can recognize patterns, manage data and learn. Along with the improvement of artificial intelligence, the application of artificial intelligence in the diagnosis of skin changes is also being improved. In this paper, a database was created for 1000 participants in the study, 200 healthy volunteers, and 800 dermatological patients with problematic facial skin health conditions. An expert system has been developed, with idea to classify patients with problematic facial skin (majority class). A pre-fed artificial neural network (ANN) was selected for the development of the expert system in this study. A modified learning algorithm was used for the problem of unbalanced data set, treating the minority class as noise or interference, during the training process.