《保险研究》20201106-《基于机器学习模型的糖尿病带病人群医疗险风险保费测算》（张宁、陈浩、周亮、包竹青、高珊、赵颖旭）

［摘   要］本文在医疗与保险协同创新的背景下,引入机器学习方法,对糖尿病群体的住院费用进行测算,并对不同机器学习方法进行了比较。基于筛选的机器学习方法,进行了面向糖尿病群体的医疗险的保费测算,并分析了年龄、性别、地区等基础因素对保费的影响,以及投保计划相关因素对保费的影响,还重点分析了并发症和主要合并症因素对保费的影响。测算分析的结果可以助力保险公司进行商业健康险产品创新、保险欺诈识别以及优化运营管理。

［关键词］机器学习；住院费用预测；糖尿病带病人群；保费测算；商业健康险

［作者简介］张宁,中央财经大学金融学院教授,中央财经大学中国金融科技研究中心主任；陈浩,中央财经大学保险学院研究生；周亮,中央财经大学中国金融科技研究中心兼职研究员；包竹青,泰康保险集团股份有限公司运营企划主管；高珊,泰康保险集团股份有限公司运营企划专员；赵颖旭(通讯作者),流行病学博士,泰康健康产业投资控股有限公司首席医疗分析师,中央财经大学中国金融科技研究中心兼职研究员。

Calculation of the Risk Premium of Diabetes Medical Insurance Based on the Machine Learning Model

ZHANG Ning,CHEN Hao,ZHOU Liang,BAO Zhu-qing,GAO Shan,ZHAO Ying-xu

Abstract：Under the collaborative innovation of medical insurance and medical service provision,this paper introduced the machine learning method to calculate the hospitalization cost of diabetes group,and compared different machine learning methods.Then based on the screened machine learning method,the paper calculated the medical insurance premiums for the diabetes group,and analyzed the impacts on premiums by the basic factors such as age,gender,locality,and the insurance  plan.Special attention was directed to premiums impacts by the key factor of complications.The results can help insurance companies to innovate on commercial health insurance products,identify insurance frauds and optimize operation management.

Key words：machine learning;prediction of hospitalization expenses;people with diabetes;premium calculation;commercial health insurance