Modelado espacio-temporal de la incidencia de enfermedades transmisibles en Ecuador mediante Spatial GARCH

Abstract

CONDITION FOR PROJECT PUBLICATION. Communicable diseases have complex spatial-temporal dynamics that are not adequately represented by traditional epidemiological models, especially when there is spatial heterogeneity and variability in incidence. The aim of this project is to analyze the behavior of dengue in the city of Guayaquil using a conditional volatility model based on spGARCH, evaluating its ability to identify areas of dynamic risk. The hypothesis is that incorporating spatial dependence and heteroscedasticity improves the early detection of critical areas, justified by the need to strengthen epidemiological surveillance with more realistic quantitative tools. The project was developed using weekly epidemiological records from the period 2021– 2025, integrated with district geospatial information. Exploratory analysis techniques were applied, spatial weight matrices were constructed, conditional volatility was estimated using log-spGARCH, and endemic channels with dynamic thresholds were generated. Processing was performed in the RStudio environment and outputs were created for geographic information systems such as QGIS. The results showed spatial-temporal heterogeneity of epidemic risk and an improvement in the identification of critical episodes when incorporating conditional volatility. Spatial clusters of high volatility and epidemic persistence were identified, concentrated mainly in the northeast corridor of the city, particularly in districts 09D02, 09D03, 09D04, 09D07, and 09D08. In addition, districts such as 09D04 and 09D01 are prone to outbreaks due to neighboring districts 09D02 and 09D03, which have high levels of volatility. It is concluded that the spGARCH model is a complementary and robust tool for the epidemiological surveillance of dengue. Keywords: Dengue, Heteroscedasticity, spGARCH, Endemic Channel.