目的针对医疗费用的偏峰、厚尾分布特征,探讨有限混合模型(finite mixture model,FMM)在识别肝硬化患者住院费用异质性、提高医疗费用预测精度等方面的可行性,为准确估计和预测医疗费用提供统计方法学支持。方法介绍FMM原理,并将其应用于广州市第八人民医院肝硬化住院患者医疗费用研究,识别群体异质性,分析异质性来源,并与单成分广义线性模型的预测效果进行比较。结果对2 760名肝硬化患者住院费用进行FMM分析,最优模型为包括低、中等和高费用3个成分,拟合优度与预测效果均高于广义线性模型,异质性来源分析进一步验证了FMM识别各类患者的能力。结论 FMM能够有效地识别医疗费用的异质性,解决医疗费用偏峰和厚尾分布问题,提高医疗费用预测精度。 Objective To investigate the characteristics of partial peak and thick tail distribution of medical expenses, and to explore the feasibility of finite mixture model (FMM) in identifying the heterogeneity of hospitalization costs in patients with cirrhosis and improving the prediction accuracy of medical costs. In order to accurately estimate and Forecasting medical costs provides statistical methodological support. Methods The FMM principle was introduced and used to study the medical costs of inpatients with cirrhosis in Guangzhou Eighth People ’s Hospital. The population heterogeneity was identified and the sources of heterogeneity were analyzed. The results were compared with those of generalized linear models with single components. Results The FMM analysis of 2 760 hospitalized patients with cirrhosis was carried out. The optimal model was composed of three components including low, medium and high cost. The goodness of fit and the prediction effect were higher than the generalized linear model, and the heterogeneous source analysis further verified The ability of FMM to identify various types of patients. Conclusion FMM can effectively identify the heterogeneity of medical expenses, solve the problem of partial distribution and thick tail of medical expenses, and improve the prediction accuracy of medical expenses.