Modeling Dengue vector population using remotely sensed data and machine learning

Abstract

Mosquitoes are vectors of many human diseases. In particular, Aedes ægypti (Linnaeus) is the main vector for Chikungunya, Dengue, and Zika viruses in Latin America and it represents a global threat. Public health policies that aim at combating this vector require dependable and timely information, which is usually expensive to obtain with field campaigns. For this reason, several efforts have been done to use remote sensing due to its reduced cost. The present work includes the temporal modeling of the oviposition activity (measured weekly on 50 ovitraps in a north Argentinean city) of Aedes ægypti (Linnaeus), based on time series of data extracted from operational earth observation satellite images. We use are NDVI, NDWI, LST night, LST day and TRMM-GPM rain from 2012 to 2016 as predictive variables. In contrast to previous works which use linear models, we employ Machine Learning techniques using completely accessible open source toolkits. These models have the advantages of being non-parametric and capable of describing nonlinear relationships between variables. Specifically, in addition to two linear approaches, we assess a support vector machine, an artificial neural networks, a K-nearest neighbors and a decision tree regressor. Considerations are made on parameter tuning and the validation and training approach. The results are compared to linear models used in previous works with similar data sets for generating temporal predictive models. These new tools perform better than linear approaches, in particular nearest neighbor regression (KNNR) performs the best. These results provide better alternatives to be implemented operatively on the Argentine geospatial risk system that is running since 2012.

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

“…In several studies, the importance of variables has been demonstrated, being the Land Surface Temperature (LST) and the Normalized Difference Vegetation Index (NDVI) both of the most used variables in Ae. aegypti models developed [6–8].…”

Modeling the Temporal Population Distribution of Ae. aegypti Mosquito using Big Earth Observation Data.

“…The Ae. aegypti species is known to be adaptable to urban environments due to its inclination to being bred in artificial containers [19,21,27,28]. Spread of the causal diseases by this vector is known to correlate with the local adult vector population [29].…”

“…Environmental conditions including precipitation, vegetation conditions, temperature, and humidity have been shown in previous works to significantly influence Ae. aegypti mosquito development [21,27,30]. After obtaining these environmental variables from satellite data, statistical and machine learning (ML) models of diseases outbreak can be used for vector population prediction [31,32].ML is a subdivision of artificial intelligence which deals with the implementation of algorithms to learn complex patterns from machine readable input data for classification and regression purposes [33,34].…”

“…It is a collection of methods that provide multivariate, nonlinear and non-parametric regression or classification of data. ML has proven useful for a very large number of EO system applications such as crop disease detection [35], urban area classification [23], landscape epidemiology [19,21,27], and various other bio-geographical information extraction and regression analysis procedures [33,34].In landscape epidemiology, ML algorithms can be used to map relationships between proxy environmental conditions and ground collected vector population information for the purpose of prediction, classification and/or explanation of environmental condition effects on vector spread [21,27]. Previous attempts in modeling the population of female Ae.…”

“…Previous attempts in modeling the population of female Ae. aegypti and other species of mosquitoes have employed ML algorithms including maximum entropy methods, support vector regression (SVR), decision trees regression (DTR), k-nearest neighbor regression (KNN), artificial neural networks (ANN), multilayer perceptron (MLP), among others [19][20][21]27].In spite of the works in this study domain, there are still some gaps. One of such is in the selecting the most informative environmental features subset to obtain the model with best fit and least redundancy.…”