Having forecasts for infectious diseases can help support risk management and effective intervention against the outbreak of disease. The main object of this study was to obtain a suitable way of making predictions of disease incidents through the use of time-series analysis.
Methods Retrospective data of Japanese infectious diseases were collected for the period of 2000–2010. In this study, influenza, mumps and infectious gastroenteritis were used for analysis. These data were separated into two groups: one (2000–2007) was designated the “training” set and the other (2008–2010), the “validation” set. We applied three models: an exponential smoothing method (ESM), an autoregressive integrated moving average (ARIMA) and a nearest neighbour method (NNM), to make predictions on the morbidity of the diseases. Statistical analysis for the ESM and the ARIMA model were carried out using SPSS Ver.19. The NNM was executed using a computer program made by us based on its' algorithm. We used the mean absolute percentage error (MAPE) to measure and quantify how well the data matched or “fit”. For example, a lower MAPE value would indicate a better fit of the data.
Results The best-fit model for influenza was the NNM, where the MAPE was 70%. For mumps and infectious gastroenteritis, the ARIMA revealed the best fit, where the MAPE for these were 7% and 16%, respectively.
Conclusion We found that the ARIMA and NNM provide a useful way of making predictions of disease. The models could well be used in planning for risk management against infectious diseases.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.