NH9.9 | Early Warning Systems for Infectious Disease Based on Climate and Environmental Variability
PICO

There are multiple environmental pathways that impact human, animal, and plant health. Increasing climatic variability, including extreme weather events, coupled with human-environmental interactions leads to increased risks of disease outbreaks including vector- (e.g. Zika, Dengue, Chikungunya, Malaria, Rift Valley Fever), water- (e.g. Cholera, Dysentery, Typhoid) and air-borne (e.g. Coronavirus, Influenza) diseases. These phenomena have a spatiotemporal distribution driven by the interactions of climate and environmental variables (e.g. precipitation, specific humidity, runoff, vegetation indices) with that of the vectors and hosts of each individual disease. This session is seeking research that advances the state-of-the-art in disease early warning. This can range from developing the system for which these disease models can reside to advancing the science behind individual routes of transmission using climatic, weather, and remote sensing data products.

Co-organized by CL3.2/ESSI4/GI4
Convener: Moiz UsmaniECSECS | Co-conveners: Anthony Nguy-Robertson, Cristiano TrevisinECSECS
PICO
| Mon, 24 Apr, 08:30–10:15 (CEST)
 
PICO spot 3b
Mon, 08:30

PICO: Mon, 24 Apr | PICO spot 3b

Chairpersons: Moiz Usmani, Cristiano Trevisin, Anthony Nguy-Robertson
08:30–08:35
08:35–08:37
|
PICO3b.1
|
EGU23-370
|
NH9.9
|
On-site presentation
|
Sophia Yacob and Roxy Mathew Koll

Dengue incidence has grown dramatically in recent decades, with about half of the world’s population now at risk. Climate plays a significant role in the incidence of dengue. However, the climate-dengue association needs to be clearly understood at regional levels due to the high spatial variability in weather conditions and the non-linear relationship between climate and dengue. The current study evaluates the impacts of weather on dengue mortality in the Pune district of India, for a 15-year period, from 2001 to 2015. To effectively resolve the complexity involved in the weather-dengue association, a new dengue metric is defined that includes temperature, relative humidity, and rainfall-dependent variables such as intraseasonal variability of monsoon (wet and dry spells), wet-week counts, flushing events, and weekly cumulative rains. We find that high dengue mortality years in Pune are comparatively dry, with fewer monsoon rains and flush events (rainfall > 150 mm), but they have more wet weeks and optimal humid days (days with relative humidity between 60–78%) than low dengue mortality years. These years also do not have heavy rains during the early monsoon days of June, and the temperatures mostly range between 27–35°C during the summer monsoon season (June–September).  Further, our analysis shows that dengue mortality over Pune occurs with a 2-5 months lag following the occurrence of favourable climatic conditions. Based on these weather-dengue associations, an early warning prediction model is built using the machine learning algorithm random forest regression. It provides a reasonable forecast accuracy with root mean square error (RMSE) = 1.01. To assess the future of dengue mortality over Pune under a global warming scenario, the dengue model is used in conjunction with climate change simulations from the Coupled Model Intercomparison Project phase 6 (CMIP6). Future projections show that dengue mortality over Pune will increase in the future by up to 86 percent (relative to the reference period 1980–2014) by the end of the 21st century under the high emission scenario SSP5-8.5, primarily due to an increase in mean temperature (3°C increase relative to the reference period). The projected increase in dengue mortality due to climate change is a serious concern that necessitates effective prevention strategies and policy-making to control the disease spread.

How to cite: Yacob, S. and Mathew Koll, R.: A climate based dengue early warning system for Pune, India, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-370, https://doi.org/10.5194/egusphere-egu23-370, 2023.

08:37–08:39
|
PICO3b.2
|
EGU23-593
|
NH9.9
|
ECS
|
Virtual presentation
|
Mayank Gangwar, Kyle Brumfield, Moiz Usmani, Yusuf Jamal, Antar Jutla, Anwar Huq, and Rita Colwell

Vibrio spp. is typically found in salty waters and is indigenous to coastal environments.  V. vulnificus and V. parahaemolyticus frequently causes food-borne and non-food-borne infections in the United States. Vibrio spp. is sensitive to changes in environmental conditions and various studies have explored their relationship with the environment and have identified water temperature as the strongest environmental predictor with salinity also affecting the abundance in some cases. It is unclear how additional environmental factors will affect intra-seasonal variance as well as the seasonal cycle. This study investigated the intra-seasonal variations in total and pathogenic V. parahaemolyticus and V. vulnificus organisms in oysters and surrounding waters from 2009 to 2012 at a few locations in the Chesapeake Bay. V. Vulnificus is always pathogenic, but it has been observed that there was greater sample-to-sample variability in pathogenic V. parahaemolyticus than in total V. parahaemolyticus. To determine the increase in the likelihood of vibrio presence when the value of a certain environmental parameter has changed, the odds ratio is examined for various values of environmental factors. The odds ratio that we employed measures the likelihood that the desired outcome would occur in samples with the vibrio in comparison to the likelihood that the desired outcome will occur in samples without the vibrio. This technique will give us the threshold value of the environmental variable above which the likelihood of vibrio spp. presence has increased drastically. With changing climate and environmental conditions, vibrio is posing increasing risks to human health. The findings of this study will demonstrate the effectiveness of the odds ratio technique in estimating the likelihood that vibrio abundance would increase when environmental conditions change, which can then be incorporated into prediction models to reduce the danger to the public's health.

How to cite: Gangwar, M., Brumfield, K., Usmani, M., Jamal, Y., Jutla, A., Huq, A., and Colwell, R.: Variability and the odds of Total and Pathogenic Vibrio abundance in Chesapeake Bay, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-593, https://doi.org/10.5194/egusphere-egu23-593, 2023.

08:39–08:41
|
PICO3b.3
|
EGU23-2467
|
NH9.9
|
Virtual presentation
Tom DeFelice, PhD

Some studies suggest atmospheric particulate matter with diameters 2.5 micron and smaller (PM2.5) may possibly play a role in the transmission of influenza and influenza-like illness (ILI) symptoms.  Those studies were predominantly conducted under moderately to highly polluted outdoor atmospheres.  We conducted our study to extend the understanding to include a less polluted atmospheric environment.  A relationship between PM2.5 and ILI activity extended to include lightly to moderately polluted atmospheres could imply a comparatively more complicated transmission mechanism.  We obtained concurrent PM2.5 mass concentration data, meteorological data and reported Influenza and influenza-like illness (ILI) activity for the light to moderately polluted atmospheres over the Tucson, AZ region. We found no relation between PM2.5 mass concentration and ILI activity. There was an expected relation between ILI, activity, temperature, and relative humidity.  There was a possible relation between PM2.5 mass concentration anomalies and ILI activity. These results might be due to the small dataset size and to the technological limitations of the PM measurements. Further study is recommended since it would improve the understanding of ILI transmission and thereby improve ILI activity/outbreak forecasts and transmission model accuracies.

How to cite: DeFelice, PhD, T.: On the Understanding of the transmission route tied to Reported Influenza/Influenza-Like Illness Activity, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2467, https://doi.org/10.5194/egusphere-egu23-2467, 2023.

08:41–08:43
|
PICO3b.4
|
EGU23-5923
|
NH9.9
|
ECS
|
On-site presentation
|
Alizée Chemison, Dimitri Defrance, Gilles Ramstein, and Cyril Caminade

Mosquitoes are climate-sensitive disease vectors. They need an aquatic environment for the development of their immature stages (egg-larva-nymph). The presence and maintenance of these egg-laying sites depends on rainfall. The development period of mosquitoes is reduced when temperature increases, up to a lethal threshold. Global warming will impact vector’s distribution and the diseases they transmit. The last deglaciation taught us that the melting of the ice sheet is highly non-linear and can include acceleration phases corresponding to sea level rise of more than 4 m per century. In addition, glacial instabilities such as iceberg break-ups (Heinrich events) had significant impacts on the North Atlantic Ocean circulation, causing major global climate changes. These melting processes and their feedbacks on climate are not considered in current climate models and their detailed impacts on health have not yet been studied.

To simulate an accelerated partial melting of the Greenland ice sheet, a freshwater flux corresponding to a sea level rise of +1 and +3 m over a 50-year period is superimposed on the standard RCP8.5 radiative forcing scenario. These scenarios are then used as inputs for the IPSL-CM5A climate model to simulate global climate change for the 21st century. These simulations allow to explore the consequences of such melting on the distribution of two vector-borne diseases which affect the African continent: malaria and Rift Valley Fever (RVF).  Malaria is a parasitic disease that causes more than 200 million cases and more than 600,000 deaths annually worldwide. RVF causes deaths and high abortion rates in herds and poses health risks to humans through contact with infected blood. Former studies have already characterised the evolution of the global distribution of malaria according to standard RCPs. Using the same malaria mathematical models, we study the impact of an accelerated Greenland melting on simulated malaria transmission risk in Africa. Future malaria transmission risk decreases over the Sahel and increases over East African highlands. The decrease over the Sahel is stronger in our simulations with respect to the standard RCP8.5 scenario, while the increase over east Africa is more moderate. Malaria risk strongly increases over southern Africa due to a southern shift of the rain belt which is induced by Greenland ice sheet melting.,. For RVF, the disease model correctly simulates historical epidemics over Somalia, Kenya, Mauritania, Zambia and Senegal.  However, our results show the difficulty to validate continental scale models with available health data. It is essential to develop climate scenarios that consider climate tipping points. Assessing the impact of these tipping point scenarios and the associated uncertainties on critical sectors, such as public health, should be a future research priority.

 

How to cite: Chemison, A., Defrance, D., Ramstein, G., and Caminade, C.: Impact of global warming and Greenland ice sheet melting on malaria and Rift Valley Fever, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5923, https://doi.org/10.5194/egusphere-egu23-5923, 2023.

08:43–08:45
|
PICO3b.5
|
EGU23-6855
|
NH9.9
|
On-site presentation
|
Paola Nassisi, Alessandro D'anca, Marco Mancini, Monia Santini, Marco Milanesi, Cinzia Caroli, Giovanni Aloisio, Giovanni Chillemi, Riccardo Valentini, Riccardo Negrini, and Paolo Ajmone Marsan

New climate regimes, variability and extreme events affect the livestock sector in many aspects, ranging from animal welfare, production, reproduction, diseases and their spread, feed quality and availability. Heat stress, especially when combined with excess or low humidity, exacerbates the perceived temperature or the drought conditions, respectively, increasing hazards for animals. Also, cold extremes, extraordinary windy conditions and altered radiation regimes are detrimental to both animals and fodder.

In this context, the EU-funded SEBASTIEN project aims to provide stakeholders with a Decision Support System (DSS) for more efficient and sustainable management, and consequent valuation, of the livestock sector in Italy. SEBASTIEN DSS will integrate GIS, environmental and biological variables to generate updated risk maps for livestock diseases and zoonoses and their spread, alerting about the expected occurrence of stressing conditions for animals due to abiotic and biotic factors.

The presence of parasites, vectors, and outbreaks will be combined with environmental data, gathered by spatially distributed meteorological and satellite monitoring, to detect conditions that can potentially favor or trigger the spread of related diseases. Sensor-based monitoring data will be integrated with the above information to determine ranges in animal parameters potentially associated with a higher risk of critical pathogen load or density of vectors potential carriers of diseases. Medium to long-term climate forecasts will support predicting possible shifts of favorable conditions that will open up new areas for parasites and pathogens. The vast amounts of data will be integrated and summarized into user-tailored information through a range of techniques, from empirical/statistical indicators to Machine Learning algorithms.

How to cite: Nassisi, P., D'anca, A., Mancini, M., Santini, M., Milanesi, M., Caroli, C., Aloisio, G., Chillemi, G., Valentini, R., Negrini, R., and Ajmone Marsan, P.: An early warning decision support system for disease outbreaks in the livestock sector, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6855, https://doi.org/10.5194/egusphere-egu23-6855, 2023.

08:45–08:47
|
PICO3b.6
|
EGU23-7652
|
NH9.9
|
On-site presentation
Cyril Caminade, Andrew P. Morse, Eric M. Fevre, Siobhan Mor, Mathew Baylis, and Louise Kelly-Hope

Vector-borne diseases are transmitted by a range of arthropod insects that are climate sensitive. Arthropods are ectothermic; hence air temperature has a significant impact on their biting and development rates. In addition, higher temperatures shorten the extrinsic incubation period of pathogens, namely the time required for an insect vector to become infectious once it has been infected. Rainfall also creates suitable conditions for breeding sites. The latest IPCC-AR6 report unequivocally concluded that recent climate change already had an impact on the distribution of important human and animal diseases and their vectors. For example, dengue is now transmitted in temperate regions of Europe, and malaria vectors are now found at higher altitudes and latitudes in the Tropics. Different streams of climate forecasts, ranging from short range numerical weather prediction (NWP) models to seasonal forecasting systems, to future climate change ensembles can be used to forecast the risk posed by key vector-borne diseases at different time scales.  

This work will first introduce vector-borne disease forecasting system prototypes developed for different time scales and applications. Three examples will be presented; first a NWP driven model to forecast the risk of the animal disease Bluetongue in the UK, second the skill of the Liverpool malaria model simulations driven by seasonal forecasts in Botswana, and third the impact of RCP-SSP climate change scenarios on the risk posed by dengue and malaria at global scale. In addition, the use of mathematical disease models in anticipating disease risk will be presented, highlighting the limited uptake by policy makers. To bridge the academic/policy making gap, novel participatory approaches which include all actors need to be developed.

The CLIMate SEnsitive DISease Forecasting Tool (CLIMSEDIS) project aims to bridge that gap. The overall aim of CLIMSEDIS is to develop and build capacity in the use of an innovative user-friendly digital tool. CLIMSEDIS will allow end-user stakeholders to utilise forecasts and delineate sub-national risk of multiple climate sensitive diseases to inform timely and targeted intervention strategies in eight countries across the Horn of Africa. Disease prioritization exercise, scoping reviews and interactive workshops with stakeholders will be carried out. The final deliverable will consist in a web-based portal and a phone application that will be used, maintained, and developed further by key African regional partners. A presentation of the CLIMSEDIS project phases and its overall strategy will be presented. 

How to cite: Caminade, C., Morse, A. P., Fevre, E. M., Mor, S., Baylis, M., and Kelly-Hope, L.: Forecasting the risk of vector-borne diseases at different time scales: an overview of the CLIMate SEnsitive DISease (CLIMSEDIS) Forecasting Tool project for the Horn of Africa, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7652, https://doi.org/10.5194/egusphere-egu23-7652, 2023.

08:47–08:49
|
PICO3b.7
|
EGU23-9509
|
NH9.9
|
Highlight
|
On-site presentation
|
Kamil Erguler, Cedric Marsboom, George Zittis, Yiannis Proestos, George Christophides, Jos Lelieveld, and William Wint

The Asian tiger mosquito, Aedes albopictus, is an invasive vector species. It is capable of transmitting more than 20 arboviruses, and is responsible for chikungunya, dengue, and zika transmission. Urbanisation, globalisation, and climate change are expected to expand its habitable range and increase the global vector-borne disease burden in the coming decades. To plan effective control strategies, early-warning and decision support systems are urgently needed.

We developed a climate- and environment-driven population dynamics model of Aedes albopictus with extensive geospatial applicability. The foundation of the model is the age- and stage-structured population dynamics model of Erguler et al. (2016)1. We replaced its rainfall- and human population density-dependent breeding site component with a large-scale mechanistic ecological model. The extension effectively created an ecological-dynamic model hybrid capable of representing niche dependence and response to changing environmental and meteorological conditions over time and under various land characteristics. To the best of our knowledge, this is the first spatiotemporal mechanistic model developed with a capacity to learn from both vector presence and longitudinal abundance data.

We calibrated the model with an extensive field surveillance dataset by combining the data collected through the AIMSurv project, the first pan-European harmonized surveillance of Aedes invasive mosquito species of relevance for human vector-borne diseases, and the global surveillance records available from VectorBase MapVEu. By deriving the model structure and environmental dependencies from the literature and allowing a complete re-configuration of the entire parameter set, we asserted the biological relevance and geospatial applicability, which extends over Europe and North America.

We corroborate that temperate northern territories are becoming increasingly suitable for Aedes albopictus establishment, while neighbouring southern territories become less suitable, as climate continues to change. We identify potential hotspots over Europe and North America by employing the combination of vector abundance and activity as a proxy to pathogen transmission risk.  By investigating routes of introduction to new territories, we demonstrate the significant role of dynamic environmental suitability in the highly efficient spread of this invasive mosquito.

The model is scheduled for integration into the "Climate-driven vector-borne disease risk assessment platform", to predict habitat suitability and dynamic abundance of important disease vectors and the risk of diseases transmitted by them at any location and time up to the end of the century. With the continental model of Aedes albopictus, the platform will reliably inform public health professionals and policy makers and contribute to the global strategies of integrated vector management.

1 Erguler K, Smith-Unna SE, Waldock J, Proestos Y, Christophides GK, Lelieveld J, Parham PE. Large-scale modelling of the environmentally-driven population dynamics of temperate Aedes albopictus (Skuse). PloS one. 2016 Feb 12;11(2):e0149282.

How to cite: Erguler, K., Marsboom, C., Zittis, G., Proestos, Y., Christophides, G., Lelieveld, J., and Wint, W.: The first continental population dynamics model of the Asian tiger mosquito driven by climate and environment, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9509, https://doi.org/10.5194/egusphere-egu23-9509, 2023.

08:49–08:51
|
PICO3b.8
|
EGU23-570
|
NH9.9
|
ECS
|
Virtual presentation
Moiz Usmani, Kyle Brumfield, Yusuf Jamal, Mayank Gangwar, Rita Colwell, and Antarpreet Jutla

The association of climatic conditions with human health outcomes has been known for ages; however, the impact of climate on infectious agents in disease transmission is still evolving. Climate change alters the regional weather impacting the emergence, distribution, and prevalence of infectious (vector-, water- or air-borne) diseases. Since the last few decades, the world has experienced an apparent increase in the emergence and re-emergence of infectious diseases, such as Middle East respiratory syndrome coronavirus (MERS-CoV); severe acute respiratory syndrome coronavirus (SARS-CoV); Ebola virus; Zika virus; and recently SARS-CoV-2. With many health agencies recommending handwashing, clean water access, and household cleaning as prevention measures, the threat to water security looms over the world population resulting in a significant public health burden under the lens of the emergence of infectious diseases. Under-resourced regions that lack adequate water supplies are on the verge of an enormous additional burden from such outbreaks. Thus, studying anthropogenic and naturogenic factors involved in the emergence of infectious diseases is crucial to managing and mitigating inequalities. This study aims to determine the impacts of climate variability on infectious diseases, namely water-, air-, and vector-borne diseases, and their association with the distribution and transmission of infectious agents. We also discuss the advancement of built infrastructure globally and its role as a mitigation or adaptation tool when coupled with an early warning system. Our study, therefore, will provide a climate-based platform to adapt and mitigate the impact of climatic variability on the transmission of infectious diseases and water insecurity.

How to cite: Usmani, M., Brumfield, K., Jamal, Y., Gangwar, M., Colwell, R., and Jutla, A.: Human health as an indicator of climate change., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-570, https://doi.org/10.5194/egusphere-egu23-570, 2023.

08:51–08:53
|
PICO3b.9
|
EGU23-10475
|
NH9.9
|
ECS
|
On-site presentation
|
Melanie Veron

Vesicular stomatitis (VS) is a multi-vector arboviral disease that affects livestock and has a significant impact on agriculture in both the US and Mexico. Biting midges (Culicoides species) are known vectors of VS. Presence-only species distribution models (SDMs) provide a powerful and versatile tool for estimating both the habitat suitability of biting midges and the distribution of VS, the disease they spread. Such models can improve our understanding of Culicoides ecology, provide opportunities for more efficient VS surveillance and mitigation, and help determine geographical areas where VS is endemic or vulnerable to potential future transmission.

Here, we discuss two case studies related to modeling the distribution of VS and its insect vector. The first focused on predicting the habitat suitability of biting midges, including C. sonorensis and its close relatives (C. variipennis, C. albertensis, and C. occidentalis), based on species presence records collected in the past hundred years from various sources. The second study involved directly estimating the distribution of VS in Mexico, where we used occurrence data in the form of confirmed VS cases in livestock from 2005-2020 in historically endemic regions of Mexico.

SDMs are typically generated using temporally static input data. However, we improved the accuracy of our predictions by applying the Maxent algorithm to time-specific input data, creating dynamic species distribution models and habitat suitability maps. For both case studies, a robust dynamic Maxent distribution modeling workflow was implemented using temporally matched occurrence and environmental data that were carefully selected in collaboration with domain experts.

How to cite: Veron, M.: Dynamic distribution modeling of arboviral vesicular stomatitis and its vector, the biting midge (Culicoides spp.): two case studies, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10475, https://doi.org/10.5194/egusphere-egu23-10475, 2023.

08:53–08:55
|
PICO3b.10
|
EGU23-12513
|
NH9.9
|
Highlight
|
On-site presentation
Jan Kyselý, Hana Hanzlíková, Aleš Urban, Eva Plavcová, and Jan Kynčl

Links between weather variability, influenza/acute respiratory infections (ARI), and human health are extremely complex in the cold season, and their explanation remains uncertain. It is not clear whether the winter mortality peak is related rather to low ambient temperatures or ARI, and how weather variability may modify transmission patterns of ARI and related mortality. This study investigates links between weather characteristics, influenza/ARI epidemics and all-cause mortality in the population of the Czech Republic (Central Europe), by employing long-term epidemiological and meteorological datasets over the 1982/83 to 2019/20 epidemics seasons. The links are analysed with respect to the predominant type of influenza virus in each season (A/H3N2 and A/H1N1 subtypes, and B lineages). We focus on i) identification of meteorological conditions associated with epidemics, ii) how timing of the epidemics and their magnitude are linked to weather characteristics, and iii) whether there are synergetic effects of cold weather and epidemics on the mortality impacts. Preliminary results suggest that high excess mortality during influenza epidemics was associated with low temperatures while above-average temperatures were linked to lower morbidity and mortality impacts. The role of other meteorological characteristics is less clear. Understanding weather conditions that increase the transmission and survival of influenza and respiratory viruses could help to better inform at-risk populations, implement preventive measures, and mitigate the negative impacts of influenza and ARI.

How to cite: Kyselý, J., Hanzlíková, H., Urban, A., Plavcová, E., and Kynčl, J.: Links between weather and seasonal influenza epidemics, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12513, https://doi.org/10.5194/egusphere-egu23-12513, 2023.

08:55–08:57
|
PICO3b.11
|
EGU23-13298
|
NH9.9
|
ECS
|
On-site presentation
Cristiano Trevisin and Andrea Rinaldo

Prognostic indices, such as the reproduction number or the epidemicity index, help assess the fate of ongoing infectious disease epidemics. While the first is of established importance, the latter focuses on the instantaneous reactivity of the infective compartment to new flare-ups. When subthreshold values of such indices apply (respectively, below the unity for the first and below zero for the latter), they warrant long-term disease-free and unreactive epidemiological conditions. 

These prognostic indicators benefit policymakers during the assessment and implementation of containment measures to reduce the disease burden. They may depend on an array of factors, including environmental forcings and the effect of containment measures on disease transmission.

We showcase a possible implementation of such prognostic indices with reference to the disastrous 2010-2019 Haiti cholera outbreak. To this end, we use a compartmental model that considers rainfalls as an environmental forcing and societal actions tackling the disease's spread. We thus test several scenarios considering a different deployment of intervention measures and we evaluate the outcome of the evolution of the prognostic indices and the epidemiological trajectory in the Haitian regions. We find that subthreshold values of these indices lead to faster waning-disease conditions.
As these indices can recap diverse epidemiological signatures induced by the spatial and temporal deployment of containment measures and potentially by evolving environmental forcings, their implementation could enable policymakers to strategically adopt containment measures in response to both evolving epidemiological and climate forcings.

How to cite: Trevisin, C. and Rinaldo, A.: Prognostic epidemiological indices and the fate of ongoing infectious disease outbreaks, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13298, https://doi.org/10.5194/egusphere-egu23-13298, 2023.

08:57–08:59
|
PICO3b.12
|
EGU23-14214
|
NH9.9
|
ECS
|
On-site presentation
Argyro Tsantalidou, Konstantinos Tsaprailis, George Arvanitakis, Diletta Fornasiero, Daniel Wohlgemuth, Dusan Petric, and Charalampos Kontoes

Mosquito-borne diseases (MBDs) have been spreading across many countries including Europe over the past two decades, causing thousands of deaths annually. They are transmitted through the bites of infectious mosquitoes. Environmental, meteorological and other spatio-temporally variables affect the mosquito abundance (MA), and thus affect the circulation of the MBDs in the community. So an early warning system of MA based on these parameters could serve as a warning for the upcoming MBDs incidence. 

We propose Deep-MAMOTH, a data driven, generic and accurate early warning system for predicting MA in the upcoming period, based on earth observational (EO) environmental data and optionally in-situ entomological data. Deep-MAMOTH can be easily replicated and applied to multiple areas of interest without any special parametrization.

The Deep-MAMOTH pipeline collects EO information from various data sources (temperature, rainfall, vegetation, distance from coast, elevation, etc.) and in-situ entomological data for each area of interest. Then, there is a feature extraction phase that combines the previous collected information to more complex features, and finally this data is fed into a Deep Neural Network responsible to capture the relationship between the above mentioned features and the MA, delivering a MA risk class ordered from 0 to 9 for the upcoming period (e.g. 15 days). The pipeline provides a standardized way to predict MA without depending on the area of interest or the mosquito genus and can be modified to predict the actual MA instead of a risk class. However, risk classes help to better propagate the severity of the situation.

Two versions of Deep-MAMOTH were implemented, the first one is using recently collected entomological information in order to produce predictions (i.e. mosquitoes collected 1 week ago). The other version works when there is no recently collected entomological information for the area of interest. The latter version is expected to perform worse than the first one, but gives us the capability to produce predictions anywhere on earth without the need of recently collected entomological data. 

We applied Deep-MAMOTH in Veneto (Italy), in Upper Rhine region (Germany), and the Vojvodina region (Serbia) regarding the Culex spp. genus mosquito. The results are promising as Deep-MAMOTH in Italy achieves a mean absolute error (MAE) of 1.27 classes with the percentage of predictions that deviate at most 3 classes (e3) from the actual one reaching up to 95%. In Serbia MAE is 1.77 classes, with e3 equal to 88% and finally for Germany MAE is 0.92 classes and e3 equal to 94%.  

It’s worth mentioning that prediction performance in the version of Deep-MAMOTH without using entomological information remains promising. MAE in Italy was increased only by 0.02 and in Germany by 0.1, with e3 remaining at the same level in both cases, while in Serbia MAE increased by 0.2 with e3 decreasing by 8%. We conclude that the prediction of MA from EO data can be accurate with or without recently collected entomological data.

Acknowledgment:This research has been co-financed by the ERD Fund of the EU and Greek national funds through the Operational Program Competitiveness, Entrepreneurship and Innovation, RESEARCH-CREATE-INNOVATE(project code:T2EDK-02070)

How to cite: Tsantalidou, A., Tsaprailis, K., Arvanitakis, G., Fornasiero, D., Wohlgemuth, D., Petric, D., and Kontoes, C.: A Deep Early Warning System of Mosquito Borne Diseases using Earth Observational Data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14214, https://doi.org/10.5194/egusphere-egu23-14214, 2023.

08:59–09:01
|
PICO3b.13
|
EGU23-15398
|
NH9.9
|
Virtual presentation
Theoktisti Makridou, George Arvanitakis, Konstantinos Tsaprailis, Diletta Fornasiero, and Charalampos Kontoes

An Early Warning System for mosquito abundance is a valuable tool that can alert authorities for potential outbreaks of mosquito populations in a given area for the upcoming period. This information is used to take mitigation actions in order to avoid spread of vector borne diseases such as West-Nile Virus, Malaria, Zika etc. A promising direction of those systems today aims to predict the upcoming mosquito population by following a data driven approach and taking advantage of machine learning (ML) algorithms. The ML algorithms are trained on a limited set of point level data that include the environmental, geomorphological, climatic information and historical in-situ measurements of mosquito population for specific latitude and longitude coordinates. Goal of the ML algorithms is to learn the patents that connect the characteristics (features) of a given area (temperature, humidity, NDVI, rainfall, latitude, longitude, etc) with the upcoming mosquito population.

 

Once the in-situ entomological data are expensive to be collected and limited, one of the key challenge of the aforementioned approach is to understand where those models can generalize with an acceptable accuracy in order to be re-used in areas that prior entomological information do not exist or in other words to understand the area of applicability of those models.

 

In this study we analyze the performance of ML algorithms that have been trained in specific areas and applied to “unseen” areas. Our analysis aims to understand the characteristics of the cases where the algorithms manage to generalize compared with the ones where the performance is poor. Our scope is to establish a systematic approach for determining the area of applicability of the models, thus, to obtain a prior knowledge regarding the areas that we expect models to generalize properly and the areas the predictions of the models are not trustworthy.

 

Our work relied on historical data of Culex pipiens mosquitoes (West Nile virus) collected in the Veneto region of Italy for the decade 2011-2021 and satellite Earth Observation data. For ML regressor we used a feedforward Neural Network with typical mean square error cost function. Initially we conclude that the typical euclidian distance between the coordinates of the trained area and the unseen data is not an informative metric about the model’s area of applicability. Instead, we propose a metric that calculates the distance between the known and the unknown points in the feature space (environmental, geomorphological etc.) and also takes into account the feature importance of trained Neural Network using the SHAP values.

 

The results showed that our proposed metric is informative regarding where the model is expected to have more accurate predictions and manage to capture the cases where the generalization will be poor. This information is useful both to judge if the predictions of a model are trustworthy and also to understand for which areas our prior information is not sufficient and to take actions in future network planning.

How to cite: Makridou, T., Arvanitakis, G., Tsaprailis, K., Fornasiero, D., and Kontoes, C.: Understanding the Area of Applicability of Data Driven Mosquito Abundance Prediction Models, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15398, https://doi.org/10.5194/egusphere-egu23-15398, 2023.

09:01–09:03
|
PICO3b.14
|
EGU23-17226
|
NH9.9
|
ECS
|
Highlight
|
Virtual presentation
|
Yusuf Jamal, Moiz Usmani, Mayank Gangwar, and Antarpreet Jutla

Vibrio spp. are pathogenic bacteria native to warm and brackish water. Vibriosis- the disease caused by these pathogens in humans accounts for around 80000 illnesses and 100 deaths annually in the United States. Of all the species, V. vulnificus has the highest mortality rate of all seafood-borne pathogens in the United States. In this context, understanding the environmental conditions that lead to increased V. vulnificus growth and spread can aid in the development of early warning systems and targeted prevention strategies. Besides sea surface temperature (SST), biotic parameters like coastal chlorophyll are also determined to affect V. vulnificus incidence in humans locally. However, the precise role of coastal chlorophyll as a potential confounding variable is understudied. Moreover, the spatial scale to which the data for environmental variables could be obtained also poses characterization constraints for researchers since the commonly employed in-situ sampling-based methods usually work with discrete locations covering a small area. The present study uses the odds ratio analysis to determine SST and chlorophyll-a threshold values critical to V. vulnificus incidence. The analysis reveals a definite positive relationship between remotely derived environmental variables and the odds of V. vulnificus incidence, where a specific statistical value of SST and chlorophyll-a marks a clear distinction between low and high odds of V. vulnificus incidence. This finding translates into a consistent pattern when checked for counties of coastal Florida. We anticipate our methodology to help distinguish between high and low-risk conditions, enabling public health workers to take proactive measures to protect the health and well-being of the public.

How to cite: Jamal, Y., Usmani, M., Gangwar, M., and Jutla, A.: Identification of thresholds on Sea surface temperature and coastal chlorophyll for understanding environmental suitability of V. vulnificus incidence, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17226, https://doi.org/10.5194/egusphere-egu23-17226, 2023.

09:03–10:15