Modeling and Predicting the Impact of Climate Variability on Influenza Virus Spread in Cuba

Abstract

Influenza viruses are among the main causes of acute respiratory infections in Cuba, associated to pneumonia, and among the five first causes of dead in the last decades. Tropical climate has a great influence on influenza variations and seasonality. The significant increasing of climatic anomalies yield by anthropogenic climate change, carries modifications in the reproductive capacity of the virus and its circulation in temporal as in spatial scale. That´s why, forecast the dynamic of influenza in the country, would allow to health system decision-makers taking the appropriate actions to control the disease. Objective: To obtain seasonal models for prediction and early warning of the climate variability impact on influenza viruses in temporal as in spatial scale in Cuba. Methods: Ecologic study with retrospective-prospective analysis of influenza series and the climate anomalies described by Bultó climate indexes in the period 2010-2020. To spatial structure, the serial data interpolated from 10km weight matrix was used by Kriging method. That allows creating a continuous grid to the country. For simulation and prediction the space structure, spatial autoregressive model were used, while for the temporary scale Autoregressive Conditionally Heteroscedastic model was implemented, both with exogenous variables. Several fit tests to predictive quality were performed. Results: It was confirmed that heteroscedastic autoregressive and spatial autoregressive models based on complex climatic indexes to simulate the climate seasonal variability as exogenous variables, are adequate to forecast the climate variability impact on influenza viruses’ circulation. The fitness of temporal model with significant concordance (0.95% and 0.96% Skill factor respectively), and 0.91% and 0.90% to spatial model was determined. A monthly and quarterly early warning system to virus circulation in the country was obtained and provinces with higher viral activity in different year months were identified. Conclusions: Temporal and spatial models to forecast and early warning of the influenza viruses’ circulation were obtained, conditioned by the impact of seasonal climate variability. This modelling methodology could be used in other respiratory viruses.