SISS-Geo: Leveraging Citizen Science to Monitor Wildlife Health Risks in Brazil

Phytophagous, blood-suckers or predators? Automated identification of Chagas disease vectors and similar bugs using convolutional neural network algorithms

Correct identification of blood-sucking bugs, such as triatomines, is important because they are vectors of Chagas' disease. Identifying these insects is often difficult for non-specialists. Deep learning is emerging as a solution for automated identification. This study evaluates the performance of three convolutional neural networks (CNNs) - AlexNet, MobileNetV2 and ResNet-50 - to identify bugs categorized by their feeding habits: 'blood-suckers', 'phytophagous' and 'predators'. A dataset of 707 dorsal view pictures was divided into training, validation, and test subsets (70 %, 10 %, and 20 %, respectively). Transfer learning was used to train the models, and Grad-CAM visualizations identified the picture regions that most influenced the predictions. All models achieved an accuracy of over 94 %, with ResNet-50 slightly outperforming the other models in terms of sensitivity and specificity. ROC and AUC analyses confirmed the reliability of these algorithms, highlighting their potential for robust bug identification. This study demonstrates the applicability of CNNs in distinguishing Triatominae from other insects, paving the way for the development of affordable vector identification tools to improve Chagas disease surveillance and control.

Harmonizing Multisource Data to Inform Vector-Borne Disease Risk Management Strategies

In the last few decades, we have witnessed the emergence of new vector-borne diseases (VBDs), the globalization of endemic VBDs, and the urbanization of previously rural VBDs. Data harmonization forms the basis of robust decision-support systems designed to protect at-risk communities from VBD threats. Strong interdisciplinary partnerships, protocols, digital infrastructure, and capacity-building initiatives are essential for facilitating the coproduction of robust multisource data sets. This review provides a foundation for researchers and practitioners embarking on data harmonization efforts to (a) better understand the links among environmental degradation, climate change, socioeconomic inequalities, and VBD risk; (b) conduct risk assessments, health impact attribution, and projection studies; and (c) develop robust early warning and response systems. We draw upon best practices in harmonizing data for two well-studied VBDs, dengue and malaria, and provide recommendations for the evolution of research and digital technology to improve data harmonization for VBD risk management.

High Frequencies of kdr Mutation and Chikungunya Infection in Aedes aegypti Population from Minas Gerais, Brazil

The Chikungunya virus (CHIKV) presents global health challenges, with Brazil experiencing outbreaks since its introduction in 2014. In 2023, following a CHIKV outbreak in Minas Gerais (MG), social media was used to optimize an entomological survey aimed at identifying vectors and viral lineages and assessing insecticide resistance. Following Instagram posts, residents with suspected CHIKV infection were able to schedule mosquito aspirations. In total, 421 mosquitoes (165 Aedes aegypti and 256 Culex quinquefasciatus) were captured from 40 households in Salinas city (MG) and tested for the Dengue, Zika, and Chikungunya viruses through RT-qPCR. Twelve of 57 pools (10 Ae. aegypti and two Cx. quinquefasciatus) tested positive for CHIKV RNA. Viral RNA was also detected in the heads of nine Ae. aegypti, indicating viral dissemination but not in Cx. quinquefasciatus. Genome sequencing yielded the first near-complete genome from the 2023 outbreak, unveiling that the CHIKV strain belonged to the East/Central/South African (ECSA) genotype. Additionally, genetic analyses revealed high frequencies of kdr alleles, including in CHIKV-infected mosquitoes, suggesting resistance to pyrethroid insecticides in this Ae. aegypti population. Social media was important for guiding mosquito-capture efforts in CHIKV transmission hotspots, thus optimizing the opportunity for viral detection. These findings emphasize the urgent need for innovative vector studies and control strategies, as well as interdisciplinary approaches in public health interventions.

Integrating databases for spatial analysis of parasite-host associations and the novel Brazilian dataset

Incomplete information on parasites, their associated hosts, and their precise geographical location hampers the ability to predict disease emergence in Brazil, a continental-sized country characterised by significant regional disparities. Here, we demonstrate how the NCBI Nucleotide and GBIF databases can be used as complementary databases to study spatially georeferenced parasite-host associations. We also provide a comprehensive dataset of parasites associated with mammal species that occur in Brazil, the Brazilian Mammal Parasite Occurrence Data (BMPO). This dataset integrates wild mammal species' morphological and life-history traits, zoonotic parasite status, and zoonotic microparasite transmission modes. Through meta-networks, comprising interconnected host species linked by shared zoonotic microparasites, we elucidate patterns of zoonotic microparasite dissemination. This approach contributes to wild animal and zoonoses surveillance, identifying and targeting host species accountable for disproportionate levels of parasite sharing within distinct biomes. Moreover, our novel dataset contributes to the refinement of models concerning disease emergence and parasite distribution among host species.

The contribution of citizen science in the surveillance of wildlife and related arthropods

Environmental and anthropogenic factors may significantly affect the diffusion of wild animals, enhancing the interface of human–wildlife interactions and driving the spread of pathogens and vector-borne diseases between animals and humans. However, in the last decade, the involvement of citizens in scientific research (the so-called citizen science approach, henceforth abbreviated as CS) provided a network of large-scale and cost-effective surveillance programmes of wildlife populations and their related arthropod species. Therefore, this review aims to illustrate different methods and tools used in CS studies, by arguing the main advantages and considering the limitations of this approach. The CS approach has proven to be an effective method for establishing density and distribution of several wild animal species, in urban, peri-urban and rural environments, as well a source of information regarding vector–host associations between arthropods and wildlife. Extensive efforts are recommended to motivate citizens to be involved in scientific projects to improve both their and our knowledge of the ecology and diseases of wildlife. Following the One Health paradigm, collaborative and multidisciplinary models for the surveillance of wildlife and related arthropod species should be further developed by harnessing the potentiality of the CS approach.

WildHealthNet: Supporting the development of sustainable wildlife health surveillance networks in Southeast Asia

Wildlife and wildlife interfaces with people and livestock are essential surveillance targets to monitor emergent or endemic pathogens or new threats affecting wildlife, livestock, and human health. However, limitations of previous investments in scope and duration have resulted in a neglect of wildlife health surveillance (WHS) systems at national and global scales, particularly in lower and middle income countries (LMICs). Building on decades of wildlife health activities in LMICs, we demonstrate the implementation of a locally-driven multi-pronged One Health approach to establishing WHS in Cambodia, Lao PDR and Viet Nam under the WildHealthNet initiative. WildHealthNet utilizes existing local capacity in the animal, public health, and environmental sectors for event based or targeted surveillance and disease detection. To scale up surveillance systems to the national level, WildHealthNet relies on iterative field implementation and policy development, capacity bridging, improving data collection and management systems, and implementing context specific responses to wildlife health intelligence. National WHS systems piloted in Cambodia, Lao PDR, and Viet Nam engaged protected area rangers, wildlife rescue centers, community members, and livestock and human health sector staff and laboratories. Surveillance activities detected outbreaks of H5N1 highly pathogenic avian influenza in wild birds, African swine fever in wild boar (Sus scrofa), Lumpy skin disease in banteng (Bos javanicus), and other endemic zoonotic pathogens identified as surveillance priorities by local stakeholders. In Cambodia and Lao PDR, national plans for wildlife disease surveillance are being signed into legislation. Cross-sectoral and trans-disciplinary approaches are needed to implement effective WHS systems. Long-term commitment, and paralleled implementation and policy development are key to sustainable WHS networks. WildHealthNet offers a roadmap to aid in the development of locally-relevant and locally-led WHS systems that support the global objectives of the World Organization for Animal Health's Wildlife Health Framework and other international agendas.

Fast surveillance response reveals the introduction of a new yellow fever virus sub-lineage in 2021, in Minas Gerais, Brazil

BACKGROUND

In Brazil, the yellow fever virus (YFV) is maintained in a sylvatic cycle involving wild mosquitoes and non-human primates (NHPs). The virus is endemic to the Amazon region; however, waves of epidemic expansion reaching other Brazilian states sporadically occur, eventually causing spillovers to humans.

OBJECTIVES

To report a surveillance effort that led to the first confirmation of YFV in NHPs in the state of Minas Gerais (MG), Southeast region, in 2021.

METHODS

A surveillance network was created, encompassing the technology of smartphone applications and coordinated actions of several research institutions and health services to monitor and investigate NHP epizootics.

FINDINGS

When alerts were spread through the network, samples from NHPs were collected and YFV infection confirmed by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and genome sequencing at an interval of only 10 days. Near-complete genomes were generated using the Nanopore MinION sequencer. Phylogenetic analysis indicated that viral genomes were related to the South American genotype I, clustering with a genome detected in the Amazon region (state of Pará) in 2017, named YFVPA/MG sub-lineage. Fast YFV confirmation potentialised vaccination campaigns.

MAIN CONCLUSIONS

A new YFV introduction was detected in MG 6 years after the beginning of the major outbreak reported in the state (2015-2018). The YFV strain was not related to the sub-lineages previously reported in MG. No human cases have been reported, suggesting the importance of coordinated surveillance of NHPs using available technologies and supporting laboratories to ensure a quick response and implementation of contingency measures to avoid YFV spillover to humans.

Statistically enriched geospatial datasets of Brazilian municipalities for data-driven modeling

The lack of georeferencing in geospatial datasets hinders the accomplishment of scientific studies that rely on accurate data. This is particularly concerning in the field of health sciences, where georeferenced data could lead to scientific results of great relevance to society. The Brazilian health systems, especially those for Notifiable Diseases, in practice do not register georeferenced data; instead, the records indicate merely the municipality in which the event occurred. Typically in data-driven modeling, accurate disease prediction models based on occurrence requires socioenvironmental characteristics of the exact location of each event, which is often unavailable. To enrich the expressiveness of data-driven models when the municipality of the event is the best available information, we produced datasets with statistical characterization of all 5,570 Brazilian municipalities in 642 layers of thematic data that represent the natural and artificial characteristics of the municipalities' landscapes over time. This resulted in a collection of datasets comprising a total of 11,556 descriptive statistics attributes for each municipality.

Examining the paradox of urban disease ecology by linking the perspectives of Urban One Health and Ecology with Cities

The ecology of zoonotic, including vector-borne, diseases in urban social-ecological systems is influenced by complex interactions among human and environmental factors. Several characteristics contribute to the emergence and spread of infectious diseases in urban places, such as high human population densities, favorable habitat for vectors, and humans' close proximity to animals and their pathogens. On the other hand, urban living can contribute to the improvement of public health through better access to health services and creation of ecological and technological infrastructure that reduces disease burdens. Therefore, urbanization creates a disease ecology paradox through the interplay of urban health penalties and advantages for individual and community outcomes. To address this contradiction, we advocate a holistic Urban One Health perspective for managing urban systems, especially their green spaces and animal populations, in ways that more effectively control the spread of zoonotic diseases. This view should be coupled with an Ecology with Cities approach which emphasizes actionable science needed for urban planning, management and policymaking; developing disease and vector surveillance programs using citizen and community science methods; and improving education and communication actions that help diverse stakeholders understand the complexities of urban disease ecology. Such measures will enable scholars from many disciplines to collaborate with professionals, government officials, and others to tackle challenges of the urban disease paradox and create more sustainable, health-promoting environments.