Using Prediction Markets to Forecast Trends in Infectious Diseases Prediction markets provide a means for collecting, analyzing expert opinions to forecast infectious disease outbreaks

P ublic health agencies conduct surveillance for a number of infectious diseases and related phenomena. Typically, the collected information is used to help public health officials, researchers, and health care professionals cope with and plan for future disease activity. In many cases, however, such information is no longer particularly relevant for clinical purposes when it becomes widely available. For example, influenza surveillance data are usually at least one to two weeks old by the time this information is available at the state and regional levels. For sexually transmitted and other infectious diseases, the information lag often is much longer because it is published on a quarterly or yearly basis. Some data may not be disseminated until after scientific manuscripts are published, typically months after manuscripts are submitted. Prediction markets provide an alternative approach for gathering and analyzing infectious disease-related information, including informed opinions, for the purpose of spotting trends. Such “markets” predict events and trends by using information contained in prices that are part of either real or simulated transactions. Operators construct prediction markets by defining specific financial instruments or contracts that are based on some particular event of interest, such as “Will there be a documented case of human-to-human transmission of H5N1 influenza in North America by January 2008?” Findings are presented in an easy-to-interpret graphical format. Although prediction markets will never replace traditional disease-surveillance systems, they can serve as a useful supplement. Importantly, these markets can be used to generate data that are useful for improving existing surveillance systems and for determining how microbiologists and medical professionals respond to new findings as they identify and then cope with specific disease outbreaks. By recording every trade and retrospectively determining who knew what and when during the early stages of outbreaks, we can use such systems to help the medical and microbiology communities build more sensitive conventional surveillance systems. Summary