Improving Forecasts of Human West Nile Virus (WNV) Disease Risk in the United States

West Nile virus (WNV), a zoonotic virus transmitted primarily to humans by an infected Culex spp. mosquito (after feeding on infected birds), is the leading mosquito-borne disease in the contiguous United States. As a persistent public health hazard in many U.S. states, there is a need for better knowledge of when and where WNV outbreaks will occur to help target public health responses. In particular, the Northern Great Plains are a hot spot for WNV transmission, and South Dakota has the highest long-term incidence of West Nile neuroinvasive disease in the country.

The project aimed to forecast WNV risk through a sustainable software system integrating entomological surveillance data of mosquito infection, epidemiological surveillance of human WNV cases, and remotely-sensed environmental data for the South Dakota Department of Health surveillance activities. Models are calibrated and predictions are validated with historical data. To generate real time forecasts, an automated Google Earth Engine App is used to process data into the ArboMAP (Arbovirus Mapping and Prediction) system. ArboMAP automates data ingestion, processing, and modelling and generation of reports to support public health decision making.

ArboMAP has been integrated into the South Dakota Department of Health operational WNV surveillance activities. This software tool produces weekly, county-level maps and seasonal forecasts of WNV risk that the South Dakota Department of Health can use to alert the public during high peak WNV transmission periods.

After the successful integration of ArboMAP in the South Dakota Department of Health WNV surveillance activities, the project team partnered with departments of health and collaborating universities to determine the feasibility of ArboMAP implementation in four other states of WNV burden. The states of Louisiana, Michigan, Nebraska, and Oklahoma have a substantial burden of severe WNV neuroinvasive disease, but distinctive climate, landscapes, mosquito vector species, and avian host communities compared to South Dakota. The team analyzed relationships between climate variations and WNV transmission in these states, finding that best predictors of human WNV cases were mean daily temperature and humidity. They used these results to update and improve the ArboMAP modelling approach to capture the characteristic environmental triggers of WNV outbreaks.

WNV is a nationally notifiable disease that is reported to state and territorial health departments and subsequently to the U.S. Centers for Disease Control and Prevention through ArboNET (https://www.cdc.gov/vector-borne-diseases/php/arbonet/index.html). Disease transmission is associated with summer and fall months, with peak transmission from July to September. This project fuses weather, satellite, and regional entomological and epidemiological survey data to predict WNV case prevalence, supporting decision makers to prepare, warn, and mitigate disease transmission risk and protect community health. Dr. Michael Wimberly, University of Oklahoma, said: “We came up with some new ideas for how to develop an improved system for forecasting West Nile virus, leveraging information in satellite imagery.”

The departments of health represent the key decision-maker who have incorporated these modelled outputs into the state’s operational surveillance activities. The first project partner, the South Dakota Department of Health, articulated the utility and value of the ArboMAP forecasting tool. Dr. Lon Kightlinger, South Dakota Department of Health, said: “The information that we’re getting from the project are maps of risk factors associated with things that we, at the health department, really can’t track or figure out. Things that have to do with climate, lay of the land and precipitation… I think this will be absolutely an essential tool in controlling West Nile here in South Dakota.” Furthermore, Dr. Josh Clayton, South Dakota State Epidemiologist reiterated the importance of understanding annual epidemiology trends: “It helps complete the picture in terms of what we see for the year.”

The team co-designed and co-developed the ArboMAP system with all public health partners. Regular design and evaluation workshops facilitated partner feedback, including prioritizing visualizations (vs written text), incorporating consistent colour and format, using stand-alone figures for communication, and highlighting key results on the first pages.

This project incorporated environmental parameters associated with the distinctive climate, landscapes, mosquito vector species, and avian host communities. Estimating the vector suitability indices for Culex spp. mosquitoes is crucial to understand the influence of mosquito development, survival, and potential of WNV transmission.

Dr. Michael Wimberly, formerly of South Dakota State University and currently at the University of Oklahoma, said: “What we’re able to do is strengthen the disease collection data by using the Earth observation data. You can’t track mosquitos all across this huge, sparsely-populated state [South Dakota]. That’s where the Earth observations come in. We can look everywhere across the state.”

MODIS (NASA’s Terra) and Landsat offer land cover data. The North American Land Data Assimilation System (NLDAS) integrates observation-based and model reanalysis data to derive land surface and hydrological models. GridMET represents gridded meteorological data from the Parameter-elevation Relationships on Independent Slopes Model (PRISM). The Advanced Microwave Scanning Radiometers (AMSR-E and AMSR2) and the Integrated Multi-satellitE Retrievals for GPM (Global Precipitation Measurement) provide data on global precipitation estimates.

These primary Earth observation data products for ArboMAP are entomological and epidemiological surveillance data; NASA’s Terra and Aqua atmosphere and land environmental satellites; NASA’s and NOAA’s NLDAS land-surface model datasets; NASA’s and USGS’s Landsat; and the IMERG global precipitation estimates.

This project created an impactful decision-support tool for the South Dakota, Louisiana, Michigan, Nebraska, and Oklahoma Departments of Health, and residents have benefited from more precise and timely public alerts of WNV risk. Dr. Michael Wimberly, University of Oklahoma, commented: “Part of the [project] goal was always to create something transferrable and sustainable.” Mosquito control programs now have real-time information that can guide their operational WNV surveillance activities, including vector control (e.g., insecticide spraying). Training workshops can be offered to epidemiologists and other staff members to build capacity for their knowledgeable use of ArboMAP for ongoing WNV surveillance activities. With this real-time information, public alerts and other educational outreach activities have been developed and implemented across counties during high peak WNV transmission periods.