Using data envelopment analysis‐neural network model to evaluate hospital efficiency

Due to an increasing amount of public resources dedicated to healthcare systems, it is important to measure the efficiency of hospitals. Systematically analysing hospital systems is one important way to discover and improve inefficiencies. The purpose of this study is to propose a data envelopment analysis (DEA)‐artificial neural network (ANN)‐based model to measure and evaluate the efficiency scores of hospitals. DEA is straightforward but requires time, knowledge and more process time than ANN. By combining these two methods, it is possible to lessen the shortcomings of DEA. In the proposed model, DEA classifies each hospital as either efficient or inefficient. Input and output variables of DEA are used for the inputs, and the efficiency scores of the hospitals are defined as the outputs of the ANN system. After the system is trained, the ANN model is applied to the test data to classify each hospital into efficient or inefficient. The results are then compared with each other, and discriminant analysis (DA) is compared with ANN. Results show that a well‐trained ANN performs good classification and even gives better solutions than DA. Also, ANN shows the advantage of using less CPU time and computer resources than the DEA, especially in large data sets.