Machine Learning to Predict Zoonotic Disease

Why do the majority of human infectious diseases originate from wildlife? What distinguishes the small fraction of species that carry and transmit zoonoses to humans? Intrinsic organismal characteristics (e.g., life history, ecological, physiological traits) recapitulate a long evolutionary history that may signal species' capacity to be future reservoirs, vectors, or microbial agents of zoonoses (human diseases with animal origins). By examining particular species groups (mammal orders), vector groups (ticks, mosquitoes), and pathogen and parasite types, this work aims to understand the biological underpinnings of zoonotic diseases.


Our work is responsive to current global health events, such as the recent outbreaks of Ebola, Zika, and Nipah viruses, to extract actionable, data-driven predictions to inform management and public health preparedness. We combine tools from data science, machine learning, and dynamical modeling to generate infectious disease intelligence and a develop a predictive strategy to confront growing threats to global health.


Return to Top

Macroecology of Zoonotic Disease


Zoonotic diseases are distributed across space, time, and species. Our lab focuses on answering outstanding 'who, what, where' questions about zoonotic diseases. Where are zoonotic reservoirs currently distributed, and where are their hotspots and coldspots? What particular species pose the greatest spillover risk to humans? When do zoonotic disease threats grow and when do they diminish? We pursue these questions with two goals: 1) to inform global health preparedness and improve ecologically comprehensive disease management, and 2) to draw out mechanistic hypotheses about the outbreak and emergence process from empirical data.


Example projects include spatiotemporal risk modeling to understand the emergence ecology of Ebola and related filoviruses; identifying wild reservoirs to prevent long-term endemicity of Zika virus in the Americas; predicting novel tick vectors and understanding why some ticks are better than others at transmitting zoonoses; an NSF EEID funded project investigating the assembly and interaction of viromes found within a widespread mouse reservoir and a widespread tick vector to explore how viromes influence pathogen transmission; a second NSF EEID funded project to understand global trait patterns of zoonotic diseases across mammal groups and the consequences of these host traits for disease dynamics in host populations; a multinational DARPA funded project to understand drivers of henipavirus shedding and to develop therapeutic interventions for spillover prevention in bats.


Return to Top

Disease Ecology of Wildlife


While the majority of our work centers on zoonotic diseases, ongoing collaborations explore the ecology of diseases found only in wildlife.  Previous work examined amphibians are important indicators of ecosystem health that are declining globally due to pathogenic threats such as Batrachochytrium dendrobatidis (Bd), a fungal pathogen. The disease, chytridiomycosis, is estimated to have impacted hundreds of frog species worldwide and is widely considered among the greatest threats to global biodiversity. In previous work, Dr. Han investigated the impact of Bd infection on host behaviors, such as aggregation, thermoregulation, and anti-predator behaviors, and the consequences of these changes for host community stability and diversity.


A recent project examined the role of salamander larvae as voracious mosquito predators that suffer heavy mortality caused by exposure to common mosquito-repellents, highlighting the potential for negative feedbacks that may drastically reduce mosquito vector control and vector-borne disease control as a crucial ecosystem service (link). 


Return to Top

Return to Top