GIS: spatial analysis and risk mapping
CHTC: high-throughput computing for large-scale modeling
A.I. & Machine Learning: computer vision and predictive models
Macrophotography & Image Stacking: specimen imaging for identification
Thermal Drone Imagery: wildlife and roost surveys
Citizen Science: public mosquito and tick data
Current Projects:
Forecasting temperature effects (RCP8.5) on four West Nile virus mosquito vector species in North America
Wisconsin tick-borne disease forecasting: citizen-science and clinic data, with Marshfield Clinic, are used to map tick exposure and forecast Lyme and other tick-borne disease risk.
Desert mosquito ecology in Las Vegas: Aedes aegypti and Culex trapping, climate, and neighborhood data, shared through the Las Vegas Mosquito Watch dashboard.
Malaria across time and place: paleoclimate suitability for ancient malaria in Bahrain and Athens, and remote sensing with serology to map transmission in Busia, Kenya.
Mosquito larval habitats and cross-border transmission in Antananarivo, Madagascar, and Leticia, Colombia.
Past Projects:
Current Projects:
Assessing how changing environments affect the health of Flying Foxes, and in turn, the emergence and transmissibility of Nipah Virus in Bangladesh
Nipah virus and Pteropus bats in Bangladesh and India: land-cover change around roosts, deep-learning maps of water, canopy, and date-palm habitat, AI-assisted bat counts, and a citizen-science survey.
Filovirus spillover (Ebola and Marburg): the human, animal, and environment interface around documented primary introductions.
Tick-borne disease and Crimean-Congo hemorrhagic fever in the Balkans, with the Pasteur Institute Novi Sad and the University of Pécs: Borrelia seroprevalence, Hyalomma tick ecology, and where people meet ticks.
Oropouche virus and biting flies on Costa Rica's Osa Peninsula, with Osa Conservation: field surveillance, remote sensing, and AI-based detection.
Past Projects:
HLC collections by mosquito genus as a function of HLC capture rate (A) minutes from sunlight (B) by photoperiod (AM vs PM). HLC capture rates were calculated as (summation of unique species collected/summation of all species collected ) across all collections. Mosquitoes below the horizontal line were collected in the daytime (between sunrise and sunset) while mosquitoes above line were collected at nighttime (between sunset and sunrise). A total of 8 mosquitoes were collected in the AM hours, while the remainder were collected in the PM hours. Unique letters below genus/species names indicate significantly different groupings, designated by Tukey's HSD.
Box plot distributions (median, quartiles, and outliers) of mosquito by genus and species collected via human landing catch by CDC epidemiologic week. HLC collections occurred within the 55 hexagon study regions of the NWMAD over the summers of 2018 and 2019. Each dot represents an individually collected mosquito. Unique letters below genus/species names indicate significantly different groupings, designated by Tukey's HSD.
Current Projects:
Forecasting the public health burden from hurricanes and recommendations for mitigation
Florida manatee mortality: spatial risk models linking environmental conditions to mortality events.
Environmental exposure: private-well PFAS contamination and the people at risk.
Air pollution policy: a World Health Organization South-East Asia regional expert group brief on clean-air policy.
Rohingya refugee health: a health-worker knowledge, attitudes, and practices survey.
Past Projects:
West Nile virus modeling in Chicago
Model comparisons, calculated as the difference between the best-fit larger scale model and the best-fit larger-scale model, for county-local (A), county-UFS (B), and local-UFS (C).
High-resolution human WNV risk maps for county (A), local (B), and ultra-fine (B) scales. Values correspond to total infections per 10,000 people from 2005-2016. Each pixel is populated by the best-fit models for each respective scale.
Overall performance for each predictor and final model used in this study. Each of the 70 covariates used in the study, listed in alphabetic order by data availability/work load to acquire score (1-4), were evaluated by mean performance (A). Highest performing covariates are noted by enlarged label text and darker blue bar color. The overall performance for each linear and logistic model (n=14 for both) was evaluated by BIC value (B). Means for each outcome (xcovariate = 0.48; xlinear = -193406; xlogistic = 670.9) are designated by vertical dashed lines.
Mean SARS-CoV-2 age-adjusted case (a) and mortality (b) epidemic curves among United States cases between January 1, 2020 and March 30, 2021. Key periods of interests: before and after Fall 2020 semester begins, and Waves 1-3, are overlaid for visual reference. Shaded widths of each line are 95% confidence limits. Unique letters, arranged in descending order (highest mean value = A and so on), indicate statistically different groups (case and mortality rates, respectively; assessed by Student’s t-test, p-value 0.05).
Toilet parts in a residence hall during the Green Fund project installation. Photo by Nathan Jandl.