Performance of Univariate Forecasting on Seasonal Diseases: The Case of Tuberculosis

Abstract

The annual disease incident worldwide is desirable to be predicted for taking appropriate policy to prevent disease outbreak. This chapter considers the performance of different forecasting method to predict the future number of disease incidence, especially for seasonal disease. Six forecasting methods, namely linear regression, moving average, decomposition, Holt-Winter's, ARIMA, and artificial neural network (ANN), were used for disease forecasting on tuberculosis monthly data. The model derived met the requirement of time series with seasonality pattern and downward trend. The forecasting performance was compared using similar error measure in the base of the last 5 years forecast result. The findings indicate that ARIMA model was the most appropriate model since it obtained the less relatively error than the other model.

Seasonal Variations in Notification of Active Tuberculosis Cases in China, 2005–2012

“…A study indicated that ARIMA model is the most appropriate model for forecasting seasonality pattern of seasonal diseases since it has the more relatively accuracy than models of linear regression, moving average, decomposition, Holt-Winter’s and artificial neural network. [24] Compared with ARIMA, maybe the X-12-ARIMA program is more appropriate model. The chief source of the X-12-ARIMA program is the extensive set of time series model building facilities built into the program for fitting the regARIMA models.…”

Forecasting Tuberculosis Incidence in Iran Using Box-Jenkins Models

“…Zhang et al used the ARIMA model to fit the changes of the incidence of tuberculosis in parts of China and to predict the incidence in future, which can be used as a scientific basis for the prevention and treatment of tuberculosis in this country (15). Permanasari et al investigated the performance of six different forecasting methods, including linear regression, moving average, decomposition, Holt-Winter’s, ARIMA, and artificial neural network (ANN) for monthly tuberculosis data forecasting and showed that the ARIMA model was the most appropriate model (28). Na et al in a study showed that the ARIMA model can be used to appropriately fit the changes of the incidence of pulmonary tuberculosis in Sichuan province of China and for short-term predictions (29).…”

Trends, seasonality and forecasts of pulmonary tuberculosis in Portugal

“…Of all the forecasting methods available for time series analysis, ARIMA models have been shown to outperform other techniques when applied to TB data. 17 The selected SARIMA models presented the best fit to PTB time series; they explained most of the data variance and provided an effective model for the seasonality and correlation structure of the time series. Furthermore, SARIMA models confirmed the presence of seasonality and a downward trend detected using STL decomposition.…”

“…9,15 Moreover, taking into account the correlation between nearby observations in time series allows the prediction of short-term future incidence rates and the suitable planning of control strategies. 4,5,16,17 No published studies with these methodologies were found concerning PTB trend and seasonality or forecasts in Portugal. The present study aims to supply a descriptive overview of trend and seasonality of PTB monthly incidence rates in Portugal from 2000 to 2010, disaggregated by high/low-incidence areas and population subgroups (sex and age groups).…”