Tele-Epidemiology and Public Health Applications

The successfulness of disease prevention and control requires a good understanding about health determinants and their associated risks in space and time. Environmental determinants (land use and land cover, waters quality, air quality, climate variability, etc.) were identified as key determinants of health for the emergence and re-emergence of several diseases such as vector-borne diseases. Maintaining ongoing acquisition of data related to these determinants for large regions or remote areas constitutes a significant challenge. Earth observation satellites offer a framework of spatial and temporal continuity to overcome this challenge through a set of proxy measures or indicators related to health determinants. However, Earth observation image analysis methods – like supervised classification – commonly used to estimate environmental determinants related to land use and land cover are time and resource consuming. Moreover, variations of microclimatic conditions combined with high landscape heterogeneity limit the effectiveness of climatic indicators derived from remote sensing. A better knowledge of the spatial variability of environmental determinants is needed to target areas at-risk on local scales and to establish a site specific management for disease prevention and control.

This study aims to develop indicators pertaining to environmental determinants by using Earth observation to support the assessment and the monitoring of risks of exposure to vector-borne diseases including West Nile virus. It is based on multi-sensor (Landsat, Radarsat, SPOT) and multi-temporal data as well as artificial intelligence and geostatistics in order to improve and to automate the estimation of these determinants. Temperature and humidity indicators (THI) were used to characterize the local variabilities of surface temperature, near surface air temperature, surface humidity, and air humidity that determine heat islands and the suitable habitat for the occurrence and the development of vectors (e.g. mosquitos). Land use and land cover indicators (LULCI) based on novel divergence indices were used to characterize the local variabilities and densities of five key environmental determinants: agricultural lands, impervious surfaces (urban and semi-urban areas), forests, surface waters, and wetlands. THI and LULCI indicators were used to map suitable habitats for vectors and the proximity of those habitats to human populations and used to develop a mosquito density model, which will lead to the estimation of an index of exposure to West Nile virus.

Preliminary results show that the new divergence indices significantly enhance the separability between thematic classes of land use and land cover compared to existing vegetation indices. They contribute to a better estimation of environmental determinants. Also, THI indicators allow for a better characterization of the spatial variability of microclimate conditions and provide greater accuracy.

The output of this project will contribute to improve the effectiveness of surveillance, adaptation policies, decision-making, and interventions. It will also allow for the monitoring and rapid identification of at-risk areas to improve emergency preparedness and response.

The paper will present first results of ESAs EOEP-4 Data User Element Innovators III project VecBorn (VecBorn = Earth observation products and services for Vector Borne diseases).

Health mapping targets a user community that is not yet well acquainted with the use of Earth Observation Applications. ESA supported first activities to test and demonstrate the potential of earth observation in epidemiology with its DUE project EPIDEMIO and within RESPOND. Both projects have shown that the potential of earth observation in epidemiology and the need of support from space to fight against diseases are very high. The project VecBorn will continue ESAs activities in a close partnership with key user organisations.

The objective of the project is to develop satellite based methods for the delineation of risk areas and to support the development of disease control measures for 2 important vector borne diseases, for Tick Borne Encephalitis and for the skin disease Buruli Ulcer.

The presented results focus on the study about Buruli Ulcer. It is one of the 17 neglected tropical diseases (classified by WHO). The exact mode of transmission of M. ulcerans is still unknown and therefore preventive measures cannot be applied so far.

With an increasing number of cases and the complications currently connected with the disease in Cameroon, it is necessary to find a way to identify risk areas and in conclusion support the development of disease control measures. Although many aspects of the disease are still unclear, there are several assumptions about the environmental factors and the prevalence of BU. Especially man-made environmental changes such as construction of dams and the resulting changes of water bodies are factors hypothesised to be connected to the prevalence of Buruli Ulcer.

Health investigations are mostly based on clinical data only. Such clinical data only give very rough estimations about the real local origin of a disease and the environmental conditions relevant for vector distribution and risk of infection. The Swiss THP provided case data from long term investigations which are clearly related to the local origin of the infection.

In this study, the risk areas of Buruli ulcer in Cameroon were explored using satellite imagery  (in particular Sentinel and Landsat data) and ground data, in order to gain better knowledge of the spatio-temporal patterns of the disease transmission and a possible relationship to environmental factors.

The study aimed to connect observations on land cover and other environmental conditions, which are typically assessed with Remote Sensing, to a vector-borne disease that is typically assessed with Epidemiology. Therefore, this study was also targeting to combine methods of both disciplines. Multitemporal Landsat imagery was acquired and used in combination with high-resolution SPOT and Sentinel-1 data in order to obtain land cover classes and changes in land cover over time. Furthermore, an ASTER DEM was used to determine, how the elevation affects the prevalence of the disease. In addition, several geographic and demographic data sets were used. Finally, Buruli ulcer case data was correlated with the risk areas obtained with remote sensing techniques. Statistical correlations have great potential to evaluate and give valuable insight into the role of environmental factors for the spreading of the disease.

The results of the study provide better understanding and evidences about how environmental factors affect the distribution of BU. It also helps to recognise BU as a health and developmental problem and to advocate for support to endemic countries.

Emerging/re-emerging infectious diseases with high epidemiological potential risks, lead public health managers to adapt their policies. Adaptation includes early knowledge of risks. The latter requires new tools to prevent re-emerging risks.

Infectious diseases such as Rift valley fever or malaria are closely tied to climatic and/or natural and anthropogenic environmental factors among which some could be identified using remote-sensing. Then, they can be assimilated into bio-mathematical models. The French Spatial Agency (CNES) with its partners has developed a conceptual approach so-called Tele-epidemiology based upon studying climate-environment-health relationships with applicable products.

Airborne, water-borne and vector-borne infectious diseases such as malaria, cholera, Rift valley fever or dengue are closely tied to climatic and/or natural and anthropogenic environmental factors among which some could be identified using remote-sensing. Then they can be assimilated into bio-mathematical models thus allowing re-emerging risks’ assessment. In this context, studies on climate-related infectious diseases require knowledge on evolving environmental factors favoring the disease transmission. As many information on climate and environment can be measured using earth observation satellite sensors, the French Space Agency (CNES) with its partners, has developed a conceptual approach so-called Tele-epidemiology based upon studying climate-environment-health relationships with applicable products in order to help public health decision makers to improve their surveillance and to prevent re-emerging risks.

This multidisciplinary approach is based upon the study of the key mechanisms favoring the surge and spread of those diseases. Analysis of those processes is a key step in the development of new and original risk mapping using Earth observation satellite data. The primary mission is to show how those adapted space products could contribute to diseases surveillance policy and improve Early Warning Systems (EWS). The overall objective is to attempt predicting and mitigating public health impacts from epidemics.

This approach has been applied with success for Rift valley fever, and malaria in Senegal and dengue in La Martinique. Those examples are explained and the relevant Earth observation satellite data leading to the development of environmental risk mapping are presented. The delivery of those fully adapted space products to the public health authorities and vectors control services intends to contribute to the diseases surveillance policy and to the implementation of Early Warning Systems (EWS).

Vector-borne diseases represent a major public health issue worldwide. In the last “World
Malaria Report 2015”, the WHO indicates that although malaria has declined by 37%
between 2000 and 2015, nearly half of the world’s population remains at risk of malaria
(about 3.2 billion people). Other emerging mosquito-borne diseases (e.g. dengue,
chikungunya, Zika virus) affect more and more countries and people, with lasting effects on
health which are not all identified yet. Prevent and control vector-borne disease transmission
require to better identify, characterize and monitor the environmental and socio-demographic
factors, their changes, their impacts on the emergence and on the endemic and epidemic
characters of the diseases.

Remote sensing has long been used to characterize environment in the frame of ecological studies. Yet, Newton et al. (2009) showed that among 158 articles dealing with remote sensing and ecology, only 0.5% combine high resolution and SAR data. Nowadays, the growing and easier access to spatial information, together with the increase of spatial, temporal and radiometric resolutions of satellite sensors allows to consider overriding the existing limits of the use of remote sensing for such applications in tropical regions, where cloud and vegetation cover often prevent to identify and characterize the objects and environments of interest.
However, such information are clearly underutilized, pointing out the need to develop “intelligent” systems able to help public health actors in analyzing and interpreting remote
sensing data and in integrating them in their professional practice.
In this presentation, we will first review the various scientific issues encountered, from
satellite and thematic data collection to indicator setting and their actual use by public health professionals. We will address these points through an overview of different research works using remote sensing to investigate vector-borne diseases in the Tropics.

Then, we will discuss the particular issue of SAR-optical data fusion for the mapping of land cover/use. More specifically, we will focus on the identification of wetlands, and the
monitoring of environmental changes in different habitats related to vector-borne diseases. Multispectral data allow the detection and mapping of wetlands. However, the complexity of these ecosystems require the use of SAR data and their ability to penetrate vegetation cover to detect flooding conditions beneath a canopy, independently from cloud cover. Fusions based on C-band and L-band SAR data (such as Sentinel 1 and PALSAR) combined to SPOT 5 data were performed. They lead to the detection and characterization of complex wetlands systems (vegetated or not, persistent water or not …) previously undetected using optical data alone, in various environments: French Guiana - Brazil cross-border area, where cloud cover is frequent; small tropical islands of the southwestern Indian ocean where high resolution is needed.
Our study highlights the potential of optical-SAR data fusions in the characterization of
environmental factors involved in vector-borne disease transmission. These factors, in
addition with epidemiological, socioeconomic, behavioral, demographics, and entomological ones, contribute to assess hazards/risks related to such pathologies and support disease control.

In general, but in Canada in particular, the management of key public health issues requires solid evidence-based knowledge for the prevention and control of various emerging or re-emerging vector borne diseases (e.g. Lyme disease, West Nile virus, etc) and environmentally-linked diseases (e.g. enteric infections from recreational water contamination). Earth observation (EO) images enhance knowledge and capacity to characterize risk of illness across the vast Canadian territory by deriving new and up-to-date data from population, climatic and environmental determinants of health relevant for public health actions such as risk mapping, risk communication and identification of vulnerable populations.

Modeling of infectious disease transmission has made possible the identification of risk areas and the underlying factors (human activities, ecology, environment and climate) that may explain this emergence. Recent developments in data from Earth observation satellites such as for climate, land cover and land use, and distribution and density of animal and human populations have greatly improved the resolution and the specificity of explanatory and predictive models.

This presentation will focus on the scope of tele-epidemiology activities of the Canadian public health community as well as current and potential future fields of application for Earth observation data. It will demonstrate the solidity, sustainability and innovative character of such an approach to improve scale-dependent decision making at different levels of government in Canada (federal, provincial/territorial and regional) and increase the efficiency of many preventive, preparedness and response actions.

Examples of tele-epidemiology applications will be presented such as the risk assessment of microbial contamination of recreational waters and modeling the risk of vector borne diseases. Opportunities for international collaborations in these objectives will be discussed.