Exploring popularity bias alleviation for online doctor-patient two-sided matching

The rise of online healthcare platforms has significantly increased the accessibility and utilization of online medical services among both doctors and patients. Two-sided matching is a core but challenging task for platforms to design, and it determines the allocation efficiency of medical resources and the sustainable operation of platforms. However, the large participation of doctors introduces the popularity bias in this matching process. In this case, we propose an alleviation method for popularity bias and construct a multi-objective doctor-patient two-sided matching based on adjusted popularity in online platforms. Firstly, to smooth the popularity distribution and make it more adaptable, we alleviate the popularity bias by incorporating positive feedback to reflect service efficacy. Then, based on adjusted popularity, a multi-objective two-matching optimization model is constructed to jointly maximize both the consultation satisfaction of patients and the revenue of doctors. Finally, we apply the proposed decision-making process of matching to analyze its feasibility and effectiveness through a case study collected from “haodf.com”. By evaluating the results of the matching process, we can demonstrate the impact of adjusted popularity on reducing bias, and facilitate the design of intelligent matching systems for online medical platforms.