Predicting the Fit between the Respirator and Face based on facial anthropometric dimensions using neural-fuzzy method (used in crises)

Abstract

<strong>Objective:</strong> The occurrence of crises such as the outbreak of the new coronavirus (COVID-19) showed that the availability of a mask that fits the face is of great importance for individuals. The present study was performed to design a tool to assess the facial fitness of the mask based on face dimensions.<br /> <strong>Methods:</strong> A hybrid method is introduced which consists of modeling of a fuzzy system using a neural network, so that with only one-time training of this neuro-fuzzy system, ANFIS, it is possible to easily determine the fit of N95 respiratory mask only by applying the anthropometric dimensions of the face. Six anthropometric dimensions of the face were assigned as the inputs and respiratory mask fitness was assigned as the output of the ANFIS model.<br /> <strong>Results:</strong> The proposed neuro-fuzzy system, ANFIS, is designed in such a way that by specifying the input parameters for each individual, the fitness of the mask to the face can be predicted.<br /> <strong>Conclusion:</strong> According to the results of the probability predicted by the neuro-fuzzy system, using the data of the six dimensions of the face, in about 75 percent of the cases the fitness of the mask to the face of individuals can be predicted accurately; therefore, the designed ANFIS network can be used instead of the fitness test to predict the fitness of the respiratory mask to the face using the anthropometric data of the face of the individuals only when it is not possible to perform the fit testing.