删失分位数回归在医疗费用中的应用

Application of Censored Quantile Regression in Medical Cost

医疗费用数据的不完全性和右偏性是医疗费用分析时两个常见的问题。针对医疗费用的数据特点,提出了医疗费用在删失时的分位数回归模型,并给出简单加权估计和模拟退火算法这两种估计方法的比较,证明模拟退火算法的估计拟合效果更好,最后给植入自动除颤器实验（MADIT）数据的实例分析,进一步证明模型的稳健性。

The incomplete and the right bias of medical cost data are two common problems in the analysis of medical cost. This paper proposes censored quantile regression model according to the char- acteristics of medical cost data and gives a simple weighted estimation and simulated annealing algorithm these two methods, simulated annealing algorithm has better fitting effect than simple weighted estima- tion after comparison. Finally, an example analysis about MADIT data is given, this example further demonstrates the robustness of the quantile regression model.