Choice Defines Value: A Predictive Modeling Competition in Health Preference Research

Area of Science:

• Health Economics
• Decision Science
• Biostatistics

Background:

• The International Academy of Health Preference Research (IAHPR) organized a predictive modeling competition to evaluate health preference modeling approaches.
• 18 international teams participated, utilizing survey data to predict health preferences.

Purpose of the Study:

• To determine which model selection specifications and approaches most accurately predict health preferences.
• To assess the predictive validity of various analytical methods in health valuation.

Main Methods:

• An exploratory survey (n=4074) generated data on 1560 paired comparisons of health descriptions.
• Eight teams submitted predictions for 1600 additional pairs, detailing their analytical strategies.
• A confirmatory survey (n=4148) validated the predictive performance of the submitted models.

Main Results:

• The winning team, "Discreetly Charming Econometricians," achieved the lowest chi-squared value (4391.54).
• Key findings indicated that model performance varied across different health preference pairs.
• The study revealed that the value of life span is not constant and logit models exhibit limited predictive validity in health valuation.

Conclusions:

• The competition showcased the innovation and potential of novel analytical techniques in health preference research.
• Results underscore the critical role of predictive validity in accurately valuing health states and outcomes.