<p>Forecasting the seasonality and trend of pulmonary tuberculosis in Jiangsu Province of China using advanced statistical time-series analyses</p>

Abstract

Objective Forecasting the seasonality and trend of pulmonary tuberculosis is important for the rational allocation of health resources; however, this foresting is often hampered by inappropriate prediction methods. In this study, we performed validation research by comparing the accuracy of the autoregressive integrated moving average (ARIMA) model and the back-propagation neural network (BPNN) model in a southeastern province of China. Methods We applied the data from 462,214 notified pulmonary tuberculosis cases registered from January 2005 to December 2015 in Jiangsu Province to modulate and construct the ARIMA and BPNN models. Cases registered in 2016 were used to assess the prediction accuracy of the models. The root mean square error (RMSE), mean absolute percentage error (MAPE), mean absolute error (MAE) and mean error rate (MER) were used to evaluate the model fitting and forecasting effect. Results During 2005–2015, the annual pulmonary tuberculosis notification rate in Jiangsu Province was 56.35/100,000, ranging from 40.85/100,000 to 79.36/100,000. Through screening and comparison, the ARIMA (0, 1, 2) (0, 1, 1) 12 and BPNN (3-9-1) were defined as the optimal fitting models. In the fitting dataset, the RMSE, MAPE, MAE and MER were 0.3901, 6.0498, 0.2740 and 0.0608, respectively, for the ARIMA (0, 1, 2) (0, 1, 1) 12 model, 0.3236, 6.0113, 0.2508 and 0.0587, respectively, for the BPNN model. In the forecasting dataset, the RMSE, MAPE, MAE and MER were 0.1758, 4.6041, 0.1368 and 0.0444, respectively, for the ARIMA (0, 1, 2) (0, 1, 1) 12 model, and 0.1382, 3.2172, 0.1018 and 0.0330, respectively, for the BPNN model. Conclusion Both the ARIMA and BPNN models can be used to predict the seasonality and trend of pulmonary tuberculosis in the Chinese population, but the BPNN model shows better performance. Applying statistical techniques by considering local characteristics may enable more accurate mathematical modeling.

<p>An Advanced Data-Driven Hybrid Model of SARIMA-NNNAR for Tuberculosis Incidence Time Series Forecasting in Qinghai Province, China</p>

“…In the current work, to attenuate or avoid this issue: firstly, we divided our data into training and testing subsamples. Then, we selected suitable ranges for the knots in the hidden layer and the lagged inputs based on those reported by a body of earlier literature; 7,[18][19][20][21] and next, every time we ran the network, the performance measures on both subsamples of MAPE and RMSE were computed and compared, as such, until the minimum values were found on both subsamples simultaneously.…”

“…Initially, the stationarity of the TB incidence series was examined. 7 The SARIMA approach is designed to model series with stationarity. Thereby the augmented Dickey-Fuller (ADF) test was employed to verify the assumption of stationarity in the TB morbidity series.…”

“…Thereby the augmented Dickey-Fuller (ADF) test was employed to verify the assumption of stationarity in the TB morbidity series. 7 With nonstationary TB series, log transformation or differencing was adopted to help accomplish stationarity. Then, model parameters were identified and the autocorrelation function (ACF) and partial ACF (PACF) plots to select the plausible parameters for the SARIMA approach were used.…”

“…Here y(t) represents the simulations and projections produced by applying the last P sample points from the same m season besides the earlier inputs at lags p. In earlier publications, it was demonstrated that it can attain a satisfactory result and can attenuate overfitting to a great extent when the ANNs with 2-7 lagged inputs and 2-20 nodes in the hidden layer were used to conduct forecasting. 7,[18][19][20][21] Thus, in this study, to identify the bestpredicting NNNAR network, we adopted a strategy of repeated attempts to seek the suitable number of the lagged inputs and the hidden neurons in the ranges mentioned above. Among all the possible methods, NARNN network that has the lowest mean absolute percentage error (MAPE) and rootmean-square error (RMSE) values between the outputs and targets and reveals uncorrelated sample ACFs and PACFs in the residuals without leading to obvious overfitting, is the best fitting.…”

“…In recent decades, coinciding with increasingly rapid advances in the technology of computers, the use of statistical techniques for modelling and forecasting has become widespread. Of these statistical techniques, the autoregressive integrated moving average (ARIMA) method based on an assumption of linearity is currently the most extensively applied to analyze and evaluate the morbidity or mortality time series of contagious diseases, such as TB, 7 scarlet fever, 8 human brucellosis, 9 pertussis, 10 etc. Yet the incidence series of contagious diseases over time includes not only linearity but also nonlinearity due to their secular trend, cyclic pattern, seasonality and stochastic fluctuation, and therefore may result in a limited ability to extract the nonlinear clues using the ARIMA model.…”