Managing the two-tier healthcare system: referral discounts and artificial intelligence-based diagnosis
通过排队模型分析，发现调整高等级医院的治疗费用和转诊折扣总能实现社会福利最优，而提高AI诊断准确性也能改善福利，但效果受基层医院误诊率和高等级医院容量影响。
Without professional assistance, patients cannot accurately identify their severity and thus often make inappropriate choice among tiered care providers. Recently, artificial intelligence- based diagnosis (AI-based diagnosis) is widely used to help patients make better treatment choice in addition to traditional financial incentives. In this paper, we develop a strategic queueing model, where a low-level provider (GP) may refer patients to a high-level provider (SP) and the patients can strategically choose where to go for initial treatment after seeking AI advice. We find that the common practice of discounting the SP's treatment fees for referred patients may not be able to close the social welfare gap between market equilibrium and social optimum while this gap can always be closed by adjusting the SP's treatment fees together with such referral discounts. Moreover, social welfare can be improved by increasing the AI diagnostic accuracy. This effect becomes more significant when the GP has a higher error probability of treating a truly severe patient but may be less significant when the GP is more likely to refer a truly non-severe patient, depending on the SP's capacity. Finally, the two-tier healthcare system may be worse off using AI-based diagnosis under some circumstances.