A prediction model of patient satisfaction: policy evaluation and sensitivity analysis

China's healthcare system has been challenged by patient dissatisfaction with primary care and an increased tendency toward visiting high-level general hospitals, which undermine the vision of a hierarchical diagnosis and treatment system. Using the game theory, we built a model that incorporated patients' preferences and use of healthcare facilities to predict the utilization of primary care at the population level. We modeled patient behavior as an incomplete information game, whose equilibrium represents patient choice. A discrete choice model was built to describe patient satisfaction to compare the expected and actual utility. We proposed the quick fictitious play algorithm for the game model that could improve computation efficiency, using survey data from Jilin Province in the year of 2008 and 2013 in estimation, and data from 2018 to test the model's prediction accuracy, with a prediction error of approximately 5%. We subsequently used the prediction model to simulate various scenarios, to shed light on policy recommendations, to make a theoretical contribution that estimates patient utility under ordered multi-classification choice sets, and provided policy recommendations for proceeding toward a hierarchical diagnosis and treatment system.