A New Hybrid Model Using an Autoregressive Integrated Moving Average and a Generalized Regression Neural Network for the Incidence of Tuberculosis in Heng County, China.

Abstract

It is a daunting task to eradicate tuberculosis completely in Heng County due to a large transient population, human immunodeficiency virus/tuberculosis coinfection, and latent infection. Thus, a high-precision forecasting model can be used for the prevention and control of tuberculosis. In this study, four models including a basic autoregressive integrated moving average (ARIMA) model, a traditional ARIMA-generalized regression neural network (GRNN) model, a basic GRNN model, and a new ARIMA-GRNN hybrid model were used to fit and predict the incidence of tuberculosis. Parameters including mean absolute error (MAE), mean absolute percentage error (MAPE), and mean square error (MSE) were used to evaluate and compare the performance of these models for fitting historical and prospective data. The new ARIMA-GRNN model had superior fit relative to both the traditional ARIMA-GRNN model and basic ARIMA model when applied to historical data and when used as a predictive model for forecasting incidence during the subsequent 6 months. Our results suggest that the new ARIMA-GRNN model may be more suitable for forecasting the tuberculosis incidence in Heng County than traditional models.