Provider Network Selection and Patient Targeting in Health Insurance Markets

Motivated by a major private health insurance payer in the U.S., this paper develops a decision support model to better control healthcare expenditure and maximize payer profit through Provider Network Selection and Patient Targeting (PNSPT). The PNSPT problem involves jointly deciding which healthcare providers are contracted by a payer to provide services at discounted prices to its insured patients, and designing personalized health insurance plans that better align with customer preferences on insurance premiums and network breadth. We develop a novel simultaneous multi-column-and-row generation algorithm that effectively solves large-size instances to optimality, and formalize several benchmark heuristics from the practice of PNSPT and related literature. We constructed a comprehensive dataset based on real data, and use it to analyze various metrics of the optimal and heuristic solutions. We find that for a payer with seven million patients, optimal PNSPT can increase payer profit by $615 million, largely by decreasing healthcare expenditure by $564 million and without lowering patient utility in health insurance plan selection