Electronic Health Markets Used to Predict the Spread of Dengue IMED - PowerPoint PPT Presentation

Electronic Health Markets Used to Predict the Spread of Dengue IMED February 6, 2011

Introduction: Dengue In the Americas Source: Data compiled from Pan American Health Organization (PAHO) Case Reports

What are Prediction Markets? • Used to predict event outcomes (commodity prices, elections, corporate earnings, movie box office returns, disease outbreaks) • Useful in quantifying professional opinion, knowledge, and experience • Participants purchase or sell shares based on their degree of certainty of a particular outcome • Prices reflect the strength of traders’ beliefs in an outcome-higher prices mean greater probability of that outcome • Prior work with prediction markets has shown that they can provide accurate predictions weeks before an event occurs

Dengue Market Methods • Participants were given $100 of a valueless currency (Dengue Dollars) • Market questions addressed dengue-related events • How many dengue cases will be reported in the 50 United States in 2010? • How many states will report locally acquired cases in 2010? • By what percent will dengue cases reported to PAHO increase in 2010 (above 2009 reported cases)? • By what percent will severe dengue cases reported to PAHO increase in 2010 (above 2009 reported cases)? • Multiple choices for trading were provided for each question • Traders bought and sold shares based on their knowledge and certainty of dengue-related outcomes • Contract (share) prices can be interpreted as the consensus probabilities of event occurrence • Final contract values were determined using traditional surveillance data

Market Participant Profile Participant Number of Occupations Participants Countries Represented Epidemiologists: 13 • Austria: 1 Students 7 • Brazil: 1 Public Health 6 • Canada: 3 Officials • Germany: 3 Physicians 6 • Jamaica: 1 • Netherlands: 1 Microbiologists 5 • Peru: 1 Economists 5 • Spain: 1 Nurses 2 • United Kingdom: 1 Journalists 1 • USA: 69 Other 37 Total 82 Active Participants

Dengue Market Results: US Case Counts (Ed. 1) U.S. Cumulative Numbers of Dengue Cases 142 165 213 232 270 286 294 315

Dengue Market Results: US Case Counts (Ed. 2) U.S. Cumulative Numbers of Dengue Cases 315 333 346 355 368 373 390 394 421 459 462 469 471

Dengue Market Results: States with Dengue 89% Probable

Dengue Market Results: PAHO Rise in Dengue Cases 68% Probable

Dengue Market Results: PAHO Rise in Severe Dengue 43% Probable

Dengue Market Limitations, Conclusions, and Policy Implications Market Limitations Future Directions • • Number of participants New markets are open for Dengue and Chikungunya • Expertise of participating traders • New markets will focus on more policy • Frequency of trading oriented questions • Reliance on traditional surveillance • We will increase the number, geographic diversity and range of expertise of traders Conclusions • The markets provided some predictions that could be useful to policymakers. • While some predictions may seem logical to experts in infectious diseases, synthesized opinion and prediction is a powerful tool to inform policymakers of likely outcomes. • Markets have the potential as a predictive tool to help guide public health policies for infectious disease outbreaks.

THANK YOU! Center for Biosecurity of UPMC Team Tara Kirk Sell, MA Crystal Boddie, MPH Associate Associate tksell@upmc.edu boddiecr@upmc.edu University of Iowa Team Anson Ho Philip Polgreen, MD, MPH PhD Graduate Student Assistant Professor ansontaiyat-ho@uiowa.edu philip-polgreen@uiowa.edu To Participate for free in the Iowa Electronic Health Markets IEhM Please Visit the Web Site: http://iehm.uiowa.edu/