Umeå universitets logga

Aedes albopictus, a key vector for Dengue, Chikungunya, Zika, and Yellow Fever, is expanding its range beyond its tropical and subtropical origins, driven by suitable climate, population mobility, trade, and urbanization. Since its introduction to Europe, Ae. albopictus has rapidly spread and triggered recurrent outbreaks. Past model attempts have handled vector suitability and vector introduction as independent drivers. Here we develop a highly predictive spatio-temporal vector diffusion model based on climate suitability and human population predictors, integrated in one simultaneous framework. The model explains how short- and long-range spread of Ae. albopictus interacts with vector suitability, predicting areas of presence or absence with high accuracy (99% and 79%). These results show that the expansion of Ae. albopictus in Europe is predictable and provide a basis for anticipating future outbreaks in situations of dependent interacting co-drivers.

Citizen science has emerged as an effective approach for infectious disease surveillance. With advancements in machine learning, entomologists can now be relieved from the labor-intensive task of species identification. However, deploying machine learning models on mobile devices presents challenges due to constraints on battery life and memory capacity. In this study, we explore the potential of various model compression techniques for deploying machine learning models on resource-limited devices, enabling low-energy consumption and on-device processing for disease surveillance in remote or low-resource settings. We compared two main-stream model compression techniques, pruning and quantization on various mobile devices. Our findings indicate that quantization methods outperform pruning methods in terms of efficiency. Furthermore, we propose to integrate structured and unstructured pruning to enhance model performance while addressing key constraints of mobile deployment.

Infectious diseases pose a substantial threat to global health security. Key wildlife species, potentially harbouring numerous zoonotic pathogens, are increasingly being forced to adapt to disturbances from land-use change, human encroachment and climate change. Although the evidence is rather convincing pertaining to the increased risks of zoonotic diseases with degradation and disturbances, the scientific literature on the mitigating effects of ecosystem restoration on zoonotic spillover is scattered, inconclusive and challenged by the lack of a conceptual framework and practical guidance. In light of rising restoration needs and activities, we outline six critical considerations when examining impacts of zoonotic diseases from ecosystem restoration: (1) assessment of zoonotic disease targets; (2) time lag between restoration and recovery; (3) integration of trophic rewilding; (4) robust study designs; (5) controlling for confounding and modifying drivers; and (6) stakeholder engagement and co-creation with communities. Failure to account for these considerations makes the scientific contribution of restoration less valuable and may even jeopardize global efforts to reverse the global biodiversity decline.

A majority of all infectious diseases manifest some climate-sensitivity. However, many of those sensitivities are not well understood as meteorological drivers of infectious diseases co-occur with other drivers exhibiting complex non-linear influences and feedback. This makes it hard to dissect their individual contributions. Here we apply a novel deep learning Explainable AI (XAI) compartment model with covariate drivers and dynamic feedback to predict and explain the dengue incidence across Sri Lanka. We compare the compartmental Susceptible-Exposed-Infected-Recovered (SEIR) model to a deep learning model without a compartmental structure. We find that the covariate compartmental hybrid model performs better and can describe drivers of the dengue spatiotemporal incidence over time. The strongest drivers in our model in order of importance are precipitation, socio-demographics, and normalized vegetation index. The novel method demonstrated can be used to leverage known infectious disease dynamics while accounting for the influence of other drivers and different population immunity contexts. While allowing for interpretation of the covariate driver influences, the approach bridges the gap between dynamical compartmental and data driven dynamical models.

Lyme borreliosis and tick-borne encephalitis significantly impact public health in Europe, transmitted primarily by endemic tick species. The recent introduction of exotic tick species into northern Europe via migratory birds, imported animals, and travelers highlights the urgent need for rapid detection and accurate species identification. To address this, the Swedish Veterinary Agency launched a citizen science initiative, resulting in the submission of over 15,000 tick images spanning seven species. We developed, trained, and evaluated deep learning models incorporating image analysis, object detection, and transfer learning to support automated tick classification. The EfficientNetV2M model achieved a macro recall of 0.60 and a Matthews Correlation Coefficient (MCC) of 0.55 on out-of-distribution, citizen-submitted data. These results demonstrate the feasibility of integrating AI with citizen science for large-scale tick monitoring while also highlighting challenges related to class imbalance, species similarity, and morphological variability. Rather than robust species-level classification, our framework serves as a proof of concept for infrastructure that supports scalable and adaptive tick surveillance. This work lays the groundwork for future AI-driven systems in One Health contexts, extendable to other arthropod vectors and emerging public health threats.

Aedes and Culex mosquitoes, known for spreading arboviruses like dengue and West Nile, thrive in cities, posing health risks to urban populations. Climate change can create suitable climatic conditions for these vectors to spread further in Europe. Cities contain numerous landscape and infrastructure elements, such as storm drains, that allow stagnant water build-up facilitating mosquito breeding. Modifying urban infrastructure to prevent water accumulation can reduce mosquito populations, but evidence is limited. The Public Health Agency of Barcelona, Spain, introduced a structural modification of storm drains to prevent water accumulation. Together with the Agency, we designed a randomised controlled trial (RCT) to experimentally assess the effectiveness of these modifications on adult Aedes albopictus and Culex pipiens populations. It is a parallel-arm RCT with equal randomization of 44 drains to receive mosquito-proofing modifications (intervention) or not (control). Primary outcomes are adult mosquito counts and secondary outcomes are larvae and mosquito presence, assessed weekly at each drain until no mosquitoes are detected. Data analyses include generalised linear mixed models to estimate the time-averaged and highest intervention effects, subgroup and sensitivity analyses. The trial results will guide a city-wide expansion of the storm drain modifications and provide valuable evidence to enhance existing vector control measures.

Storm Daniel, the deadliest Mediterranean tropical-like cyclone, caused significant flooding in Thessaly Region, Greece, from September 4 to 7, 2023. This study examines the potential impact of such extreme weather events on vector-borne disease transmission by assessing changes in mosquito populations and West Nile virus (WNV) circulation before and after the flood in two regional units of Thessaly. Systematic monitoring data on mosquito larvae and adults, along with WNV circulation in mosquitoes and humans from 2021 to 2023, were analyzed using a weekly interrupted time series regression design controlling for confounding drivers and temporal trends. Results indicate a significant post-flood increase in Culex mosquito populations over the 7 weeks following the event. However, despite this increase—alongside optimal temperature conditions and pre-flood amplification of WNV—no corresponding rise in WNV circulation was observed in mosquitoes or human cases. This unexpected outcome may be influenced by multiple ecological factors, including disruptions of avian host communities, human displacement, and the timing of the flood during the autumn bird migration period. These findings underscore the complexity of vector-virus-host interactions and highlight the importance of continued systematic entomological surveillance for targeted mosquito control practices.

Background: The rapid spread of the Asian tiger mosquito (Aedes albopictus) poses a notable public health threat in Europe due to its ability to transmit tropical diseases such as dengue and chikungunya. We aimed to quantify the underlying drivers facilitating and accelerating Europe's transition from sporadic arbovirus outbreaks to Aedes-borne disease endemicity, focusing on dengue and chikungunya outbreaks.

Methods: We conducted a time-to-event analysis to investigate the period between establishment of Ae albopictus and autochthonous dengue and chikungunya outbreaks across Nomenclature of Territorial Units for Statistics (NUTS) 3 regions in the EU. We incorporated data from the European Centre for Disease Prevention and Control, WHO, technical and surveillance reports, and other entomological data sources on regional Ae albopictus establishment and subsequent dengue and chikungunya outbreaks from 1990 (when Ae albopictus was first introduced to an EU country) to 2024. The main outcome was survival time (ie, the time from Ae albopictus establishment to an outbreak of dengue or chikungunya), accounting for land-use types, demographic and socioeconomic factors, imported cases, and climatic variables via univariable and multivariable regression. To address recurrent outbreaks, we applied the Andersen–Gill extension of the Cox proportional hazards model to analyse all events. We further stratified regions into warm and cool groups on the basis of mean summer temperatures above or below 20°C and conducted a stratified analysis with Kaplan–Meier curves and the log-rank test to evaluate differences between these groups. We also estimated projected outbreak hazards from the 2030s to the 2060s at a decadal scale under three distinct shared socioeconomic pathways (SSPs; SSP1–2·6, SSP3–7·0, and SSP5–8·5) to assess the future impact of climate change on outbreak hazard estimates.

Findings: Between 1990 and 2024, the interval from the first NUTS 3 regional establishment of Ae albopictus to the first outbreak of dengue or chikungunya decreased from 25 years to less than 5 years. Similarly, the interval from the first outbreak to the second outbreak decreased from 12 years in 1990 to less than 1 year in 2024. Our regression analyses indicate that increasingly favourable climatic conditions play a significant role in this trend. A 1°C rise in mean summer temperature was associated with a hazard ratio of 1·55 (95% CI 1·30–1·85; p<0·0001) after controlling for health-care expenditure and imported cases and land-use type. First outbreak events might have occurred more frequently and earlier in warmer regions than cooler ones (log-rank p=0·088), reflecting a lower probability of remaining outbreak-free over time. This trend is expected to intensify under extreme climate change scenarios, with projections under the SSP5–8·5 scenario suggesting an almost five-fold increase in dengue or chikungunya outbreaks by the 2060s, relative to the 1990–2024 baseline period.

Interpretation: The findings in this study underscore the pressing need for robust control measures, enhanced surveillance, and early warning systems in the EU to mitigate the impending risk of Aedes-borne disease endemicity in the region.