Real Decision Support for Health Insurance Policy Selection

We report on an ongoing project to develop data-driven tools to help individuals make better choices about health insurance and to better understand the range of costs to which they are exposed under different health plans. We describe a simulation tool that we developed to evaluate the likely usage and costs for an individual and family under a wide range of health service usage outcomes, but that can be tailored to specific physicians and the needs of the user and to reflect the demographics and other special attributes of the family. The simulator can accommodate, for example, specific known physician visits or planned procedures, while also generating statistically reasonable "unexpected" events like ER visits or catastrophic diagnoses. On the other hand, if a user provides only a small amount of information (e.g., just information about the family members), the simulator makes a number of generic assumptions regarding physician usage, etc., based on the age, gender, and other features of the family. Data to parameterize all of these events is informed by a combination of the information provided by the user and a series of specialized databases that we have compiled based on publicly available government data and commercial data as well as our own analysis of this initially very coarse and rigid data. To demonstrate both the subtlety of choosing a healthcare plan and the degree to which the simulator can aid in such evaluations, we present sample results using real insurance plans and two example policy shoppers with different demographics and healthcare needs.