MODIS Environmental Data to Assess Chikungunya, Dengue, and Zika Diseases Through Aedes (Stegomia) aegypti Oviposition Activity Estimation

Abstract

Aedes aegypti is the main vector for Chikungunya, Dengue, and Zika viruses in Latin America and it represents a main threat for our region. Taking into account this situation, several efforts have been done to use remote sensing to support public health decision making. Moderate resolution imaging spectroradiometer (MODIS) sensor provides moderate-resolution remote sensing products; therefore, we explore the application of MODIS products to vector-borne disease problems in Argentina. We develop temporal forecasting models of Ae. aegypti oviposition, and we include its validation and its application to the 2016 Dengue outbreak. Temporal series (10/2005 to 09/2007) from MODIS products of normalized difference vegetation index and diurnal land surface temperature were built. Two linear regression models were developed: model 1 which uses environmental variables with time lag and model 2 uses environmental variables without time lags. Model 2 was the best model (AIC = 112) with high correlation (r = 0.88, p < 0.05) between observed and predicted data. We can suggest that MODIS products could be a good tool for estimating both Ae. aegypti oviposition activity and risks for Ae. aegypti-borne diseases. That statement is also supported by model results for 2016 when a dengue outbreak that started unusually earlier this season. If such activity could be forecast by a model based on remote sensing data, then a potential outbreak could be predicted.

Temporal models using environmental variables to predict Aedes aegypti oviposition activity in a temperate region of Argentina

“…Vegetation index is one of the most important variables, showing a positive relationship and fluctuating in a similar way as the oviposition activity. This relationship has also been observed between the NDVI and the egg number fluctuation of this species in the north of the country [7]. In addition, studies in La Habana, Cuba, showed a correlation between the number of larva infested containers and the presence of trees and vegetation, which demonstrates the importance of the shade that vegetation provides to containers in which the Ae.…”

“…Considering the biological characteristics of Ae. aegypti , time lags were applied to the explanatory variables [7, 27], meaning the length of time that has elapsed since the change in weather conditions that is reflected in the response variable in order to obtain a better fitted model.…”

Modeling Dengue vector population using remotely sensed data and machine learning

“…This study improved previous epidemiology studies, which consider statistical models with linear relationships [47,36,37]. Such improvement is obtained by the use of Machine Learning tools that impose the user no significant additional effort.…”

“…Previous experiences on the modeling of epidemiological applications using remotely sensed environmental variables report good results with this approach [51,37,52]. We used simple linear and ridge regressions, the latter with Tikhonov regularization with cross-validation.…”

“…At a global scope we can find interesting contributions [32,33,34] with also some operatives experiences [35] Specifically, in a interinstitutional framework between the Argentinean National Space Agency (CONAE) and the Health Ministry of Argentina, there have been initiatives to model the temporal evolution of mosquito populations using environmental variables obtained from remote sensors. These works used series of a few years and are based on a small number of satellite variables [36,37]. In an effort to improve this, [38] constructed models based on a large number of variables from various sensors for four years.…”