Estimation of Re-hospitalization Risk of Diabetic Patients based on Radial Base Function (RBF) Neural Network Method Combined with Colonial Competition Optimization Algorithm

Diabetes is the costliest gland disease in the world. Given the high rates of diabetic people, the necessity of reducing the costs of early re-hospitalization and increasing re-admissions within 30 days after discharge have drawn the attention of researchers and other health sector authorities to find ways to reduce potential and preventable hospital re-admissions. The objective of this paper is to estimate the risk of re-hospitalization of diabetic patients. In order to achieve this goal, the data were first pre-processed, and then, radial base function neural network combined with colonial competition optimization algorithm was used to estimate the risk of re-hospitalization of diabetic patients. Moreover, this risk was estimated using back propagation neural network algorithm and the radial base function neural network algorithm. The accuracy of the proposed method is 99.91. This method shows higher performance compared to radial base function neural network method and back propagation neural network without feature selection.