Methods for modellind and forecasting dynamics of infectious diseases

Abstract

Introduction.Despite dramatic progress in public health, infectious diseases are common issue leading to significant burden in terms of morbidity and mortality, and emergence and re-emergence of infections and its dynamic are often unpredictable. Infectious diseases modelling and forecasting is effective instrument for policy making in epidemiology.The aim of the reviewis to systematize current literature on modelling and forecasting in infectious disease epidemiology.Materials and methodsLiterature review in field of modelling and forecasting of infectious diseases without restrictions by publication date was conducted. Publication activity was estimated using text mining software.Results and discussion.The following most common classes of modelling methods were marked: regression models, time-series models, compartmental models, agent-based models and artificial neural networks. It was noted that a number of methods (regression analysis, time-series models and artificial neural networks) are relatively simple to implement, but a considerable volume of history data is required for teaching these models. Compartmental models are partially free from this restriction, and they can be rapidly developed for assessment of emerging and reemerging infections, but their implementation presents issues caused by host population heterogeneity. Agent-based models that present most complete descriptions of host population heterogeneity and social interactions within it are extremely complex from the technical point of view.Conclusion.Despite the presence of various mathematical algorithms for disease modelling, the demand for user-friendly statistical software for disease forecasting in field practice is persist.