Added-value of mosquito vector breeding sites from street view images in the risk mapping of dengue incidence in Thailand

Abstract

Dengue is an emerging vector-borne viral disease across the world. The primary dengue mosquito vectors breed in containers with sufficient water and nutrition. Outdoor containers can be detected from geotagged images using state-of-the-art deep learning methods. In this study, we utilize such container information from street view images in developing a risk mapping model and determine the added value of including container information in predicting dengue risk. We developed seasonal-spatial models in which the target variable dengue incidence was explained using weather and container variable predictors. Linear mixed models with fixed and random effects are employed in our models to account for different characteristics of containers and weather variables. Using data from three provinces of Thailand between 2015 and 2018, the models are developed at the sub-district level resolution to facilitate the development of effective targeted intervention strategies. The performance of the models is evaluated with two baseline models: a classic linear model and a linear mixed model without container information. The performance evaluated with the correlation coefficients, R-squared, and AIC shows the proposed model with the container information outperforms both baseline models in all three provinces. Through sensitivity analysis, we investigate the containers that have a high impact on dengue risk. Our findings indicate that outdoor containers identified from street view images can be a useful data source in building effective dengue risk models and that the resulting models have potential in helping to target container elimination interventions.

A systematic review of the data, methods and environmental covariates used to mapAedes-borne arbovirus transmission risk

“…Breteau index (BI), which is defined as the number of positive containers per 100 houses, was used as a predictor in three studies [53,81,82]. Six studies included location of Aedes breeding sites in their models [74,[83][84][85][86][87]. The number of catches of female adult mosquitoes was included in two studies [58,88].…”

“…For the 33 studies that both checked the multicollinearity of covariates and performed variable selection, we summarised the retention rate of different groups of covariates in the final models (Fig 6A) [27,32,[38][39][40][41][42][43][44][46][47][48]54,59,60,72,78,79,86,93,[99][100][101][102][103][104][105][106][107][108][109][110][111][112]. Of 33 studies, 25 studies (96.2%) retained climatic variables when tested.…”

Potential Development of Digital Environmental Surveillance System in Dengue Control: A Qualitative Study

“…This complicates the setting up of the dengue infection surveillance system, which is done directly through home visits. In addition, digital technology is needed to observe the density of larvae and the dengue infection in the region [17].…”

Predicting the fine‐scale spatial distribution of zoonotic reservoirs using computer vision

“…Because pathogen spillover into the human population is facilitated by contact between humans and animals, anthropogenic activities such as agriculture, land clearing and urban expansion may increase disease risk in some systems (e.g. rodent hosts of Lassa virus (Fichet‐Calvet et al, 2007), bat hosts of Nipah virus (Chua et al, 2002) and SARS‐related coronaviruses (Rulli et al, 2021), mosquito vectors of dengue virus (Su Yin et al, 2021)). Indeed, the apparent increase in recent disease emergence events may be attributable in part to increased human expansion into natural areas (Allen et al, 2017; Rohr et al, 2019).…”

“…Additionally, because our pipeline is modular, it allows users to easily substitute their preferred models for the computer vision or regression components. Although similar approaches have been developed, for example to improve the performance of dengue risk models by automatically extracting features from GoogleStreetMaps imagery (Su Yin et al, 2021), these efforts do not allow customized features of known or hypothesized ecological relevance to be directly extracted from aerial imagery.…”