Abstract: Model of medical cost is very important for insurance, and the cost data might include some high-dimension covariates, which may be useful to predict the future cost. However, the existing statistical methods may not be effective. Hence, in this paper, a cost model is proposed to model the data with high-dimension covariates, and the two-step method is applied to get the estimation. The first step to obtain the central dimension reduction subspace and the structural dimension, and then, local regression method is employed to get an estimate of the completely unknown regression curve. Some simulations and a real data analysis are given as well to asses the proposed model the methods in this paper.

摘要： 医疗保险费用的评估是商业医疗费用管理中的重要环节，附加信息在医疗费用的远期预测中具有重要作用。但是，如果费用数据中含有高维的附加信息，传统的方法就不再适用。因此，本文提议一个新的模型去拟合含有大范围附件信息的医疗费用，并用两步法去估计医疗费用数据。首先，我们用充分降维方法将高维协变量降为低维，得到中心降维子空间的基方向和结构维数后，再利用局部回归方法去估计医疗费用曲线，最后通过模拟和实例分析来评价我们提议的模型和方法。