The accuracy of the prediction models for surface roughness and micro hardness of denture teeth

Abstract

The paper aimed to compare the performance of artificial neural network (ANN) model with the results of in vitro experiments. For these experiments, maxillary molars of four different denture teeth were subjected to tea, coffee, cola, cherry juice, distilled water. Vickers microhardness and surface roughness values were measured. Subsequently, ANN model for the prediction of microhardness and surface roughness of different denture teeth were examined. A back-propagation ANN has been used to develop a model relating to the amount of microhardness and surface roughness. The independent variables of the model are distilled water, tea, filtered coffee, cola, cherry juice, time and denture teeth. Microhardness and surface roughness were chosen as the dependent variables. According to the results, a neural network architecture having one input layer with ten neurons, two hidden layers with six neurons, one output layer with two neurons and an epoch size of 48 gives better prediction. Prediction models for dental materials could also be supportive for in vitro studies.