Umeå University's logo

Ours is an era of global change—including climate change, land-use change, urbanization, increased mobility of humans, species and goods, and environmental shifts. Concurrently, we are witnessing a tangible increase in the rate of (re)emerging infectious diseases, mostly driven by global change factors. This complex landscape of infectious diseases necessitates strategies underpinned by computational tools such as data-driven models to enhance our understanding, response, and predictions of potential epidemics.

In this thesis, I leveraged data science algorithms and developed data-driven models that extend beyond specific pathogens, providing insights to prepare for future epidemics, with a focus on Europe. I delved into three temporal contexts: 1) retrospective analyses to understand the contribution of global change factors—specifically climate change and human mobility—fuelling the disease outbreaks and expansion (papers I & IV), 2) develop model to improve disease severity estimation during an outbreak for immediate response (paper III), and 3) future disease transmission risk trajectories under various projected scenarios of global change (paper II)—each playing a crucial role in proactive public health planning and response.

In paper I, we assessed the predictive ability and the influence of eco-climatic factors on West Nile virus (WNV)—a pathogen with multiple hosts and mosqutio-vectors, and of public health concern in Europe. Utilizing an advanced machine learning classifier XGBoost, trained on a diverse dataset encompassing eco-climatic, sociodemographic predictors to the WNV presence/absence data, the model accurately predicted the WNV risk a season ahead. Furthermore, by employing an explainable AI algorithm, we uncovered both local and European-level drivers of WNV transmission. Higher temperatures in summer and spring, along with drier winters, were pivotal in the escalated frequency of WNV outbreaks in Europe from 2010 to 2019.

In paper II, we projected the WNV risk under climate change and socioeconomics scenarios by integrating augmenting the outputs of climate ensemble into machine learning algorithms. We projected transmission risk trends and maps at local, national, regional and European scale. We predicted a three to five fold increase in WNV transmission risk during the next few decades (2040-60) compared 2000-2020 under extreme climate change scenarios. The proportion of diseasereported European land areas could increase from 15% to 23-30%, putting 161 to 244 million people at risk. Western Europe remains at largest relative risk of WNV increase under all scenarios, and Northern Europe under extreme scenarios. With the current rate of spread and in the absence of intervention or vaccines the virus will have sustained suitability even under low carbon emission scenarios in currently endemic European regions.

In paper III, we developed a method to quantify an important epidemiological parameter-case fatality ratio (CFR)— commonly used measure to assess the disease severity during novel outbreaks. In our model, we accounted for the time lags between the reporting of a cases and that of the case fatalities and the probability distribution of time lags and derived the CFR and distribution parameters using an optimization algorithm. The method provided more accurate CFR estimations earlier than the widely used estimators under various simulation scenarios. The method also performed well on empirical COVID-19 data from 34 countries.  

In paper IV, we modelled annual dengue importations in Europe and the United States driven by human mobility and climate. Travel rates were modelled using a radiation model based on population density, geographic distance, and travel volumes. Dengue viraemic travellers were computed considering local mosquito bite risk, travel-associated bite probability, and visit duration. A dynamic vector life-stage model quantified the climatic suitability of transmissionpermissive local areas. Dengue importations linearly increased in Europe and the U.S. from 2015-2019, rising by 588% and 390%, respectively, compared to 1996-2000 estimates, driven by increased travel volumes (373%) and dengue incidence rates (30%) from endemic countries. Transmission seasons lengthened by 53% and 15% in Europe and the U.S., respectively, indicating increasingly permissive climates for local outbreaks. These findings apply to other diseases such as chikungunya, Zika, and yellow fever, sharing common intermediate host vectors, namely Aedes mosquitoes.

This thesis highlights Europe's increasing vulnerability to infectious diseases due to global change factors, putting millions at risk. It emphasizes the significance of advanced modelling and innovative data streams in anticipating epidemic risks. Developing digital early warning systems to track disease drivers and taking urgent climate change mitigation and adaptation measures are crucial to anticipate and reduce future epidemic risks. The outcomes of this research can be used to develop technology-driven decision support tools to aid public health authorities and policymakers in making evidence-based decisions during and inter-epidemic periods. 

Record-breaking temperatures were recorded across the globe in 2023. Without climate action, adverse climate-related health impacts are expected to worsen worldwide, affecting billions of people. Temperatures in Europe are warming at twice the rate of the global average, threatening the health of populations across the continent and leading to unnecessary loss of life. The Lancet Countdown in Europe was established in 2021, to assess the health profile of climate change aiming to stimulate European social and political will to implement rapid health-responsive climate mitigation and adaptation actions. In 2022, the collaboration published its indicator report, tracking progress on health and climate change via 33 indicators and across five domains.

This new report tracks 42 indicators highlighting the negative impacts of climate change on human health, the delayed climate action of European countries, and the missed opportunities to protect or improve health with health-responsive climate action. The methods behind indicators presented in the 2022 report have been improved, and nine new indicators have been added, covering leishmaniasis, ticks, food security, health-care emissions, production and consumption-based emissions, clean energy investment, and scientific, political, and media engagement with climate and health. Considering that negative climate-related health impacts and the responsibility for climate change are not equal at the regional and global levels, this report also endeavours to reflect on aspects of inequality and justice by highlighting at-risk groups within Europe and Europe's responsibility for the climate crisis.

Climate change is one of several drivers of recurrent outbreaks and geographical range expansion of infectious diseases in Europe. We propose a framework for the co-production of policy-relevant indicators and decision-support tools that track past, present, and future climate-induced disease risks across hazard, exposure, and vulnerability domains at the animal, human, and environmental interface. This entails the co-development of early warning and response systems and tools to assess the costs and benefits of climate change adaptation and mitigation measures across sectors, to increase health system resilience at regional and local levels and reveal novel policy entry points and opportunities. Our approach involves multi-level engagement, innovative methodologies, and novel data streams. We take advantage of intelligence generated locally and empirically to quantify effects in areas experiencing rapid urban transformation and heterogeneous climate-induced disease threats. Our goal is to reduce the knowledge-to-action gap by developing an integrated One Health—Climate Risk framework.

West Nile virus (WNV), a mosquito-borne zoonosis, has emerged as a disease of public health concern in Europe. Recent outbreaks have been attributed to suitable climatic conditions for its vectors favoring transmission. However, to date, projections of the risk for WNV expansion under climate change scenarios is lacking. Here, we estimate the WNV-outbreaks risk for a set of climate change and socioeconomic scenarios. We delineate the potential risk-areas and estimate the growth in the population at risk (PAR). We used supervised machine learning classifier, XGBoost, to estimate the WNV-outbreak risk using an ensemble climate model and multi-scenario approach. The model was trained by collating climatic, socioeconomic, and reported WNV-infections data (2010−22) and the out-of-sample results (1950–2009, 2023–99) were validated using a novel Confidence-Based Performance Estimation (CBPE) method. Projections of area specific outbreak risk trends, and corresponding population at risk were estimated and compared across scenarios. Our results show up to 5-fold increase in West Nile virus (WNV) risk for 2040-60 in Europe, depending on geographical region and climate scenario, compared to 2000-20. The proportion of disease-reported European land areas could increase from 15% to 23-30%, putting 161 to 244 million people at risk. Across scenarios, Western Europe appears to be facing the largest increase in the outbreak risk of WNV. The increase in the risk is not linear but undergoes periods of sharp changes governed by climatic thresholds associated with ideal conditions for WNV vectors. The increased risk will require a targeted public health response to manage the expansion of WNV with climate change in Europe.

Background: In Europe, the frequency, intensity, and geographic range of West Nile virus (WNV)-outbreaks have increased over the past decade, with a 7.2-fold increase in 2018 compared to 2017, and a markedly expanded geographic area compared to 2010. The reasons for this increase and range expansion remain largely unknown due to the complexity of the transmission pathways and underlying disease drivers. In a first, we use advanced artificial intelligence to disentangle the contribution of eco-climatic drivers to WNV-outbreaks across Europe using decade-long (2010-2019) data at high spatial resolution. Methods: We use a high-performance machine learning classifier, XGBoost (eXtreme gradient boosting) combined with state-of-the-art XAI (eXplainable artificial intelligence) methodology to describe the predictive ability and contribution of different drivers of the emergence and transmission of WNV-outbreaks in Europe, respectively. Findings: Our model, trained on 2010-2017 data achieved an AUC (area under the receiver operating characteristic curve) score of 0.97 and 0.93 when tested with 2018 and 2019 data, respectively, showing a high discriminatory power to classify a WNV-endemic area. Overall, positive summer/spring temperatures anomalies, lower water availability index (NDWI), and drier winter conditions were found to be the main determinants of WNV-outbreaks across Europe. The climate trends of the preceding year in combination with eco-climatic predictors of the first half of the year provided a robust predictive ability of the entire transmission season ahead of time. For the extraordinary 2018 outbreak year, relatively higher spring temperatures and the abundance of Culex mosquitoes were the strongest predictors, in addition to past climatic trends. Interpretation: Our AI-based framework can be deployed to trigger rapid and timely alerts for active surveillance and vector control measures in order to intercept an imminent WNV-outbreak in Europe. Funding: The work was partially funded by the Swedish Research Council FORMAS for the project ARBOPREVENT (grant agreement 2018-05973).

BACKGROUND: While the COVID-19 outbreak in China now appears suppressed, Europe and the USA have become the epicentres, both reporting many more deaths than China. Responding to the pandemic, Sweden has taken a different approach aiming to mitigate, not suppress, community transmission, by using physical distancing without lockdowns. Here we contrast the consequences of different responses to COVID-19 within Sweden, the resulting demand for care, intensive care, the death tolls and the associated direct healthcare related costs.

METHODS: We used an age-stratified health-care demand extended SEIR (susceptible, exposed, infectious, recovered) compartmental model for all municipalities in Sweden, and a radiation model for describing inter-municipality mobility. The model was calibrated against data from municipalities in the Stockholm healthcare region.

RESULTS: Our scenario with moderate to strong physical distancing describes well the observed health demand and deaths in Sweden up to the end of May 2020. In this scenario, the intensive care unit (ICU) demand reaches the pre-pandemic maximum capacity just above 500 beds. In the counterfactual scenario, the ICU demand is estimated to reach ∼20 times higher than the pre-pandemic ICU capacity. The different scenarios show quite different death tolls up to 1 September, ranging from 5000 to 41 000, excluding deaths potentially caused by ICU shortage. Additionally, our statistical analysis of all causes excess mortality indicates that the number of deaths attributable to COVID-19 could be increased by 40% (95% confidence interval: 0.24, 0.57).

CONCLUSION: The results of this study highlight the impact of different combinations of non-pharmaceutical interventions, especially moderate physical distancing in combination with more effective isolation of infectious individuals, on reducing deaths, health demands and lowering healthcare costs. In less effective mitigation scenarios, the demand on ICU beds would rapidly exceed capacity, showing the tight interconnection between the healthcare demand and physical distancing in the society. These findings have relevance for Swedish policy and response to the COVID-19 pandemic and illustrate the importance of maintaining the level of physical distancing for a longer period beyond the study period to suppress or mitigate the impacts from the pandemic.

How living cells can reliably process biochemical cues in the presence of molecular noise is not fully understood. Here we investigate the fidelity of information transfer in the expression of a single gene. We use the established model of gene expression to examine how precisely the protein levels can be controlled by two distinct mechanisms: (i) the transcription rate of the gene, or (ii) the translation rate for the corresponding mRNA. The fidelity of gene expression is quantified with the information-theoretic notion of information capacity. Derived information capacity formulae reveal that transcriptional control generally provides a tangibly higher capacity as compared to the translational control. We next introduce negative feedback regulation in gene expression, where the protein directly inhibits its own transcription. While negative feedback reduces noise in the level of the protein for a given input signal, it also decreases the input-to-output sensitivity. Our results show that the combined effect of these two opposing forces is a reduced capacity in the presence of feedback. In summary, our analysis presents analytical quantification of information transfer in simple gene expression models, which provides insight into the fidelity of basic gene expression control mechanisms.

Several Italian towns are under lockdown to contain the COVID-19 outbreak. The level of transmission reduction required for physical distancing interventions to mitigate the epidemic is a crucial question. We show that very high adherence to community quarantine (total stay-home policy) and a small household size is necessary for curbing the outbreak in a locked-down town. The larger the household size and amount of time in the public, the longer the lockdown period needed.