Change Mappers

As precipitation patterns and other global environmental conditions change, epidemiologists have already noticed new patterns in the incidence and severity of certain diseases, including skin cancer, asthma and food-borne illnesses. Identifying the location and severity of such climate-related health changes is key to minimizing future risks. But how?

Sudipto Banerjee, associate professor of biostatistics at the University of Minnesota’s School of Public Health, is working to improve our ability to understand and predict impacts of climate change. Funded by a two-year American Recovery and Reinvestment Act grant, Banerjee and colleagues are developing new statistical models to study the effects of climate change on health. They also are building new software that will allow health and environmental scientists to analyze large amounts of data from widely varying locations.

Banerjee uses spatial, or location-based, statistics to explore relationships among three data categories: weather, such as temperature and precipitation; pollutants, such as particulate matter and ozone; and health, including rates of asthma hospitalizations, nonmelanoma skin cancer and the food-borne disease salmonellosis.

An essential component of Banerjee’s research is creating mathematical models to connect data with the locations where they were observed—indicating, for example, which parts of a state or county show highest risk for a particular illness. Studying data without tracking their recorded locations can skew results. “If you don’t account for space,” Banerjee says, “then sometimes conclusions that you draw can be wrong. Space can act as a confounder, in terms of confusing how the variables interact with each other.”

Banerjee’s research will help epidemiologists make statistical health predictions on a larger scale. “The challenge is to relate this variable, which has been recorded at a certain point, with data that has been recorded over a region,” he says. “So we build models to help us do that.”

ASHLEY KUEHL is a freelance writer from Minneapolis. She has written about sustainability and the environment for Twin Cities Daily Planet.

Common Language

Click for larger image

What’s the connection – if any – between a particular health problem and environmental conditions? Finding the answer requires being able to look at how the problem and conditions vary geographically. That’s far more easily said than done when the different trends are measured or reported at different scales and in different ways. For instance, patients hospitalized for asthma might be represented as number per region (upper left), while data on the percent of people who smoke might be expressed as shades of gray (upper right), air pollution sources as dots of different magnitude (lower left) and weather data as pixels (lower right).

Sudipto Banerjee is developing statistical tools to create a common “language” for these various ways of reporting data.

The goal: to be able to identify which factors are most closely associated with a particular health problem and so could hold a key to solving it.