核估计在医疗费用分布研究中的应用  

The Application of Kernel Estimation to the Hospital Expense Distribution

作　　者：汤国权[1] 李荣敏[1] 张群贵 陈尉华 

机构地区：[1]复旦大学数学所,上海200433 [2]中国太平洋人寿保险股份有限公司,上海200120

出　　处：《应用数学与计算数学学报》2005年第1期25-29,共5页Communication on Applied Mathematics and Computation

基　　金：本文受复旦-瑞士再保险研究基金资助.

摘　　要：医疗费用的分布是健康保险的精算基础之一.本文研究了某团体一年的住院医疗费用记录,运用非参数模型中的核密度估计方法,对每次住院费用对数的分布密度进行了估计,并用它进一步估计了每次住院费用的均值和方差,取得了较好的效果.在此基础上,本文对拟合结果及其对健康保险的启示作了一些讨论.The distribution of medical expenses amount is one of the actuarial bases for health insurance. The objective of this paper is to analyze medical expenses records of a group in a year. The Kernel Density Estimation, a nonparametric method, is used to estimate the density function of natural logarithm of medical expenses amount. Then the estimated density function is used to estimate the means and variances of medical expenses amount spent by the inpatients per time. The satisfactory results are got. Finally, this paper gives some further discussion about the results and its influence on health insurance.

关 键 词：核估计 医疗费用 密度函数 概率分布 

分 类 号：R197.3[医药卫生—卫生事业管理]