Keywords
Time series models; ARIMA; GARCH; Random walk; Malaria prevalence; Prediction
Abstract
Vector-borne diseases, such as Malaria, are major causes of human mortality in many areas of world, especially in the developing countries. Statistical and data-based models can provide an explicit framework to develop an understanding of infectious disease transmission dynamics. Application of different time series models to analyse and predict financial data as well as epidemiological data is of long interest to researchers. It is always interesting to see how the time series models that are extensively used in the analysis of financial data can be applied and extended to explain epidemiological data. In this paper, we have studied epidemiological data (malaria prevalence) related to Slide Positivity Rates and deaths due to Plasmodium vivax, using three major classes of time series models, namely Auto-Regressive Integrated Moving Average (ARIMA), Generalised Auto-Regressive Conditional Heteroskedastic (GARCH) and Random Walk. Our results show that as expected the chosen models fit excellently with the financial data but also show good potentiality to fit epidemiological data and provide excellent predictions. The results demonstrate the applicability of such time series models in epidemiology, specifically for malaria prevalence, where these models with appropriate choice of parameters have not been used extensively. As far as future prevalence pattern is concerned, the prediction of these models may help researchers and public health professionals to design control programmes for malaria.
Citation
: Sarkar RR and Chatterjee C. Application of Different Time Series Models on Epidemiological Data - Comparison and Predictions for Malaria Prevalence. SM J Biometrics Biostat. 2017; 2(4): 1022