Introduction

Agenda

• The BEYOND EO Center/EuroGEO Action Group on Vector-borne Diseases


• State-of-the-art in the field of epidemics


• Development of an Early Warning Decision Support System in Europe


• Human infectious disease modeling related to climatic change conditions


• Exploitation of satellite-based EO for Epidemics of mosquito-borne diseases


• The Modeling Framework and validation of an early warning system at municipal (Greece) and regional (Venneto, Italy) level

Learning outcomes

• Know about the development of an early warning system


• Understand how EO data used to forecast human cases


• Explain the application of an early warning system and its benefits on disease outbreaks

The Space/Geospatial Sector Skills Alliance

Towards an innovative strategy for skills development and capacity building in the space geo-information sector supporting Copernicus User Uptake

Copernicus Program:

Europe's eyes on Earth

Copernicus is the EU Earth observation program, looking at our planet and its environment to benefit all European citizens. It offers information services that draw from satellite Earth Observation and in-situ (non-space) data. The information services provided are free and openly accessible to users.

The EO4GEO project

• Duration: 4 years from January the 1st, 2018
• Budget: 3,87 million €
• Partnership: 25 Partners + 30+ Associated Partners (from 16 EU Countries) from Academia, Companies and networks
• Coordinators: GISIG (General), KU Leuven (Scientific & Technological), PLUS (Education & Training), Climate-KIC (Exploitation)

The EO4GEO project

• EO4GEO aims to bridge the skills gap between the supply and demand of education and training in the space/geospatial sectors, fostering the uptake and integration of space/geospatial data and services in a broad range of application domains.



• EO4GEO works in a multi and interdisciplinary way and apply innovative solutions for its education and training actions: case-based and collaborative learning scenarios; learning-while-doing in a living lab environment; on-the-job training and co-creation of knowledge, skills and competencies.

EO4GEO IS MUCH MORE

EO4GEO offers a series of tangible results that are usable for the broader EO/GI education/training community

Early Warning for Mosquito-borne epidemics at regional level

Let's start from the basics...

Mosquito-Borne Diseases in Europe

An emerging threat

Mosquitos & Diseases Culex - West Nile Virus Anopheles - Malaria Aedes albopictus - Dengue, Zika, Chikungunya

EUROGEO

Action Group

EYWA is a vision, a network, a European and even global standard.

EYWA offers a scalable, reliable and sustainable early warning system, relying on Earth observation big data combined with entomological, epidemiological and socioeconomic data, to forecast and monitor Mosquito-Borne Diseases.

EYWA

EYWA is built on the triptych: ADVOCATE ENGAGE DELIVER

EYWA

EYWA TEAM

"Together Everyone Achieves More"

How EYWA competes?

EYWA is a robust and scalable Early Warning & Decision Support System that welcomes new partners from around the world to share data and transform scientific knowledge into decision-making & mosquito control actions

EYWA system architecture

EYWA incorporates 10-years time-series of Copernicus (Sentinel-2) and other space-based data (Landsat-7 & -8, MODIS and ERA-5) in addition to in-situ entomological, epidemiological, socioeconomic and crowdsourcing data.

EYWA system architecture

A suite of APIs is developed and publicly available through BEYOND-NOA’s GitHub profile for automatic: Data harvesting Data Pre-processing EO-based indices derivation

Satellite data harvesting and processing, exploiting European and non-European services:
• Umbrella Sentinel Access Point of the Hellenic Mirror Site (an API that constitutes 100% EU innovation and has been developed by BEYOND-NOA in the framework of the NextGEOSS and EOPEN EU projects)
• CreoDIAS and Google Earth Engine

EYWA system architecture

Big Data management (278 TB and counting) Open Data Cube (ODC) technology,state-of-the-art tool for Earth Observation and other data fusion, feature engineering and data analytics All these processing steps are available through the dedicated Python API "EYWAdcAPI" at BEYOND-NOA’s GitHub profile in the epidemics repository

EYWA system architecture

	A “mammoth” feature space of at least 10-years time-series of data for every mosquito-traps network in nine regions in Europe.

How is this plethora of independent data transformed into meaningful scientific knowledge?

EYWA has a factory of dynamic and data-driven models, learning about the dynamics of mosquitoes’ abundance and mosquito-borne diseases transmission, and providing monthly, weekly, daily predictions.

EYWA system architecture

the MAMOTH model

• MAMOTH (Mosquitoes Abundance prediction Model autO-calibrated from features pleTHora)
• Generic data-driven approach relying on open EO data
• Automated Feature Selection:
• No human bias on feature selection is injected in the model
• Transferable and replicable
• Knowledge expansion: Operating under the same architecture and the same mathematical principles to different cases, offers extensive capability of comparative studies, responding to: "which characteristics seem important in one case and which on the other?"
• MAMOTH is composed of 32 (up to now)in-housed developed Python functions that are organized in 5 operational Modules

EYWA system architecture

Predictions results dissemination to the relevant Public Health Authorities through monthly reports and the EYWA Web Platform

EYWA & open data sharing

EYWA is an autonomous European Initiative, building upon the Open Innovation, Open Science and Open to the World vision for Europe.

Analysis ready 10-years’ time-series of environmental, meteorological and geomorphological data for every mosquito-traps network in 10 European regions. Accessible through:
• The EYWAopenAPI
• NextGEOSS DataHub

EYWA in action

EYWA’s operational implementation in 2020 (TRL>7 ) with a demonstrated impact in:
• Greece (Regions of Central Macedonia, Thessaly, Western Greece and Crete)
• Italy (Veneto region)

forecasting Culex mosquito populations and West Nile Virus outbreaks in 2020.

EYWA’s pre-operational test in 2020 for:
• Culex (WNV) abundance prediction in Serbia (Vojvodina region) and Germany (Baden-Württemberg region)
• Anopheles (Malaria) in Italy (Veneto region)
• Aedes albopictus (Chikungunya, Dengue, Zika) in France (Grand Est and Corsica regions)

EYWA in action

EYWA is now fully operational in all these areas and started delivering results again for the 2021 mosquito season in the end of April.

EYWA in action

Reports with operational results

EYWA produces knowledge in the form of reports, statistics, validated assessments and web GIS information layers, all available to the end-users through the EYWA Web Platform.

The reports indicate: Up-to-date epidemiological status of the Region The state-of-the-art models used The mosquito abundance predictions for the month The estimated human risk

The EYWA Reports:
• are delivered operationally from April to October every year to the relevant Public Health Authorities and decision makers.
• assist the authorities in organizing their mosquito control strategy and actions. Measurable performance indicators are used to evaluate the level of EYWA’s effectiveness toward the protection of the engaged communities against the disease outbreak.

EYWA in a nutshell

• Plethora of satellite Earth Observation data
• Entomological, epidemiological, crowdsourced, socioeconomic and auxiliary data
• State-of-the-art technological tools

Partners

West Nile Virus (WNV)

Europe's eyes on Earth

West Nile Virus (WNV)

continental/regional scale

EFSA and ECDC, 2021 2018: All seasons were warmer than usual, with late spring, summer and autumn all seeing temperatures more than 1 ℃ above average

West Nile Virus (WNV)

prefecture/province/municipality scale

2014-16: A minimum occurs after ~5y cycle.

West Nile Virus (WNV) dynamics

the EYWA approach

Mesoscale monthly to seasonal Causality Driven (forecasting) Seek predictable signals in Input Data: Environmental Epidemiological Entomological Geographical Demographical Key elements of the prediction process: Meteorological Uncertainty Epidemiological Uncertainty Microscale weekly to monthly Data Driven (nowcasting)

Dynamical core of the EYWA system: the MIMESIS model

MIMESIS: spatial dynaMIcal Model for wESt nIle viruS

Conclusions

Quantitative early warning assessments (t~months) of the spatial and temporal variability of WNV circulation in mosquitoes, birds and humans is a major challenge and requires a causative attribution.

The MIMESIS model, developed at the LapUP, operates spatially at the mesoscale and temporally at the monthly to seasonal scale. It is a dynamic model nested in an ensemble framework to account for the meteorological and epidemiological uncertainty.


• Meteorological uncertainty: ensemble seasonal forecasts.
• Epidemiological uncertainty: relies in an ensemble model developed for this purpose.

Average detection probability is 72% in GR and 94% in VE, in optimal sense. In forecast mode for 2020, it reached 66-83%.

The May temperatures are found most critical in shaping the WNV dynamics.