健康保险数据及数据管理系统研究
|
摘要
自从上世纪80年代恢复保险业务以来,我国的保险业一直在快速地发展,2006年国务院颁布了《国务院关于保险业改革发展的若干意见》,对发展健康保险提出了殷切的希望和要求,明确指出要“统筹发展城乡商业养老保险和健康保险,完善多层次社会保障体系”。这就意味着即将到来的几年将是中国健康保险的一个高速发展期,对于每一位有志投身于健康保险事业的人来说,虽然这是千载难逢的良机,但是对我国的健康保险业务我们既需要看到美好的前景,也需要意识到可能面临的诸多挑战。
目前,我国的商业健康保险供需两旺,但是业务经营并不尽人意,其表现在以下几个方面:健康保险管理制度还很落后、市场开发程度还不够高、各保险公司所提供的健康保险产品和服务远远不能满足人民群众的需要、产品雷同程度高等。另外,由于数据是风险管理的基础,是健康保险专业化经营的依托,没有强大有效的数据库健康保险专业化难以取得实效,所以在当前情况下为了更好的经营健康保险业务、管理健康保险风险,对健康保险数据的研究意义无疑是重大的。
在国外,有关商业健康保险数据的研究已经比较成熟,保险机构已积累了比较完善的数据。我国的商业健康保险业务开展时间短,经验数据不足,无论是保险监管部门、行业标准组织或是保险公司均未能推出一套完整、权威、准确的健康保险行业经验数据。无论实务界还是理论界均未对健康保险数据中医疗服务数据作过详细的探讨,笔者借鉴国外研究成果,并结合国内实际认为医疗服务数据应包括病人基本数据、入出转数据、电子病例、诊疗数据、医学影像数据、医学管理数据和经济数据的等几个方面,作为医疗信息的来源,其质量的好坏影响极其深远;另外,医疗服务数据具有复杂性和特殊性的特点,影响医疗服务数据质量的主要因素包括数据质量问题长期以来没有得到足够的重视、数据监管不健全、没有明确的奖惩措施、缺少良好的环节监控工具和措施、医疗服务数据缺乏统一标准、信息孤岛现象严重和操作人员违规操作各终端站点录入等几个方面。
要掌握伤病风险在人群中发生的规律性、对健康风险进行数量上的刻画并估计其转移成本,不仅需要搜集足够多的数据,还要对上述数据进行分析整理。笔者认为在健康保险数据的分析整理应包括健康保险数据准备、健康保险数据维护和健康保险数据预处理三个方面,而健康保险数据准备一般要经过数据资源和需求分析、数据收集准备、数据校验三个步骤。另外,健康保险数据准备在时间、数量、质量等方面都有严格的要求,不能有所纰漏。健康保险数据的预处理应包括以下几个方面:填补遗漏的数据值、平滑有噪声的数据、识别或删除异常数据值、数据变换以及解决不一致问题。
保险公司搜集数据的目的是为了有效地发现对经营健康保险有价值的信息,以便对健康保险进行全方位的风险控制。而仅仅通过对健康保险数据整理是不能达到其目的的,必须进行相应的数据挖掘和健康保险数据管理系统的建设,笔者认为数据挖掘在健康保险中的应用应达到以下目的:(1)对理赔客户进行细分(2)找出理赔客户的理赔特征。数据挖掘在健康保险业务中的应用有以下几个步骤:(1)由事务型健康保险数据库作为源系统组成数据仓库与数据集市;(2)根据健康保险业务的需要明确数据挖掘目标,并由此采取相应的数据挖掘方法对数据仓库与数据集市中的数据分析以得到知识,并由此构成知识库;(3)将获取的知识应用于客户服务、新业务推广、市场营销和新产品定制中;(4)评价应用结果并反馈到数据挖掘过程以改进方法。要建立健康保险数据管理系统必须进行相应的软件开发,笔者认为健康保险数据管理系统软件的开发方式有四种:自行开发、委托开发、联合开发和直接购买等四种,通过比较分析各种开发方式的优缺点认为保险公司比较适合采用联合开发和直接购买的方式。
健康保险数据中一部分数据来自于杂志等公开刊物上的二手数据,这些数据可能以直方图或者表格的形式出现。由于来源渠道不一样,数据的分组情况也可能有所不同,为了满足原始调查人员的需要,这些原始数据在最初的数据搜集的时候可能已经被简化。如何有效的利用这些简化过的数据呢?本文介绍了用最大熵法处理这些数据。
目前,我国关于健康保险数据方面的问题尚无较为系统全面的探讨,笔者结合我国健康保险业务的实际情况,进行大胆的创新,不乏有一些创新点,以期在这些方面的研究中能起到抛砖引玉的作用。这些创新点主要有:(1)不仅涉及一般数据的讨论,还讨论了医疗服务数据特点和质量问题以及二手数据的处理问题;(2)笔者前瞻性地对健康保险数据管理系统提出基本的框架,对今后实务中处理健康保险数据问题具有一定的借鉴和指导意义。
限于篇幅和资料的限制,对一些具体问题没有展开深入的探讨,比如研究健康保险数据可以从多个方面着手,像如何从外部搜集数据、如何从保险公司经营环节中获取经验数据等,限于这些方面和财产保险数据及寿险数据的处理办法差不多,本文着重介绍了健康保险数据中的医疗服务数据、二手数据的处理和健康保险数据管理系统,加之时间仓促及笔者自身水平有限,文中尚存在一些的缺点和不足之处。
Since the 1980s, China's resumption of domestic insurance business, China's insurance industry has been the rapid development .In the year of 2006 ,the State Council issued the "State Council's views on the reform and development of the insurance industry "policy, raised some ardent hopes and demands on the development of health insurance. It clearly pointed out "co-ordinate the development of urban and rural commercial pension insurance and health insurance, and improve a multi-level social security system ".This means that the upcoming years will be China's health insurance a high-speed development period, for those who are aspiring health insurance this is a once-successful career opportunities.We not only need to see bright prospects of health insurance in china, but also need to be aware of the many challenges that we will facing.
The current supply and demand in China's booming commercial health insurance business and however unsatisfactory circumstances,insurance companies who provide health insurance products and services, is far from the satisfaction of people. In addition, data is the basis of risk management, and health insurancel's specialized operation is relying on data , if absence of the strong and effective database, health insurance specialization is difficult to achieve real results, so under the current situation, in order to improve health insurance business and health insurance risk management,we should study on health insurance data.
In the abroad, the research on commercial health insurance data is relatively mature, insurance companys accumulated data more perfect. China's commercial health insurance business developed in a short time, it lack of empirical data,whether insurance supervision departments,the industry standards organizations or insurance companies have failed to launch a comprehensive,authoritative and accurate health insurance industry experience data. Neither practical nor theoretical circles have studied medical services data for a detail, I think that medical data including patient basic data,the transfer of data and electronic case, clinics data, medical imaging data, medical management and economic data.As the source of medical information,it's quality is good or bad will influence extremely far-reaching. Medical data has the characteristics of complexity and uniqueness, issues of data quality a long time has not been given sufficient attention, which is the main factors of medical data quality, data monitoring is not perfect, there is no clear incentives, the lack of good monitoring tools and measures links, the lack of uniform standards for medical data, and so on .
Master risk injury which occurred in population, the health risks of quantitative characterization of the transfer and its estimated cost, not only to gather enough data, but also to collate the data for analysis. I consider that data preparation is a complex and tedious work, generally it has to go through three steps: the data resources and demand analysis, data collection preparation and data calibration. In addition, the data prepared in time,quantity,quality and other aspects of the stringent requirements, can't be flawed. Data pretreatment include the followings: fill the missing data values,smooth noisy data,identification or delete anomaly data values,data transformation and resolve inconsistencies.
The purpose of insurance to collect data is effectively operate the health insurance, comprehensive control of health insurance risk. Only through data processing can not achieve their aims ,it need data mining and data management system .I consider that data mining in the application of health insurance should achieving the following objectives (1) Customer Segmentation (2) to identify characteristics of claims. Data mining in the health insurance business should follows the following steps: (1) by the service database system as a source of data warehouse and data mart; (2) in accordance with the needs of health insurance data mining targets identified by the this response to the data mining method of data warehouse and data mart data analysis in order to obtain knowledge, and thus constitute a knowledge base; (3) will be acquired knowledge to customer service, new business promotion, marketing and new product set system; (4) evaluation of the results and feedback to the application of data mining methods to improve the process.
目前,我国的商业健康保险供需两旺,但是业务经营并不尽人意,其表现在以下几个方面:健康保险管理制度还很落后、市场开发程度还不够高、各保险公司所提供的健康保险产品和服务远远不能满足人民群众的需要、产品雷同程度高等。另外,由于数据是风险管理的基础,是健康保险专业化经营的依托,没有强大有效的数据库健康保险专业化难以取得实效,所以在当前情况下为了更好的经营健康保险业务、管理健康保险风险,对健康保险数据的研究意义无疑是重大的。
在国外,有关商业健康保险数据的研究已经比较成熟,保险机构已积累了比较完善的数据。我国的商业健康保险业务开展时间短,经验数据不足,无论是保险监管部门、行业标准组织或是保险公司均未能推出一套完整、权威、准确的健康保险行业经验数据。无论实务界还是理论界均未对健康保险数据中医疗服务数据作过详细的探讨,笔者借鉴国外研究成果,并结合国内实际认为医疗服务数据应包括病人基本数据、入出转数据、电子病例、诊疗数据、医学影像数据、医学管理数据和经济数据的等几个方面,作为医疗信息的来源,其质量的好坏影响极其深远;另外,医疗服务数据具有复杂性和特殊性的特点,影响医疗服务数据质量的主要因素包括数据质量问题长期以来没有得到足够的重视、数据监管不健全、没有明确的奖惩措施、缺少良好的环节监控工具和措施、医疗服务数据缺乏统一标准、信息孤岛现象严重和操作人员违规操作各终端站点录入等几个方面。
要掌握伤病风险在人群中发生的规律性、对健康风险进行数量上的刻画并估计其转移成本,不仅需要搜集足够多的数据,还要对上述数据进行分析整理。笔者认为在健康保险数据的分析整理应包括健康保险数据准备、健康保险数据维护和健康保险数据预处理三个方面,而健康保险数据准备一般要经过数据资源和需求分析、数据收集准备、数据校验三个步骤。另外,健康保险数据准备在时间、数量、质量等方面都有严格的要求,不能有所纰漏。健康保险数据的预处理应包括以下几个方面:填补遗漏的数据值、平滑有噪声的数据、识别或删除异常数据值、数据变换以及解决不一致问题。
保险公司搜集数据的目的是为了有效地发现对经营健康保险有价值的信息,以便对健康保险进行全方位的风险控制。而仅仅通过对健康保险数据整理是不能达到其目的的,必须进行相应的数据挖掘和健康保险数据管理系统的建设,笔者认为数据挖掘在健康保险中的应用应达到以下目的:(1)对理赔客户进行细分(2)找出理赔客户的理赔特征。数据挖掘在健康保险业务中的应用有以下几个步骤:(1)由事务型健康保险数据库作为源系统组成数据仓库与数据集市;(2)根据健康保险业务的需要明确数据挖掘目标,并由此采取相应的数据挖掘方法对数据仓库与数据集市中的数据分析以得到知识,并由此构成知识库;(3)将获取的知识应用于客户服务、新业务推广、市场营销和新产品定制中;(4)评价应用结果并反馈到数据挖掘过程以改进方法。要建立健康保险数据管理系统必须进行相应的软件开发,笔者认为健康保险数据管理系统软件的开发方式有四种:自行开发、委托开发、联合开发和直接购买等四种,通过比较分析各种开发方式的优缺点认为保险公司比较适合采用联合开发和直接购买的方式。
健康保险数据中一部分数据来自于杂志等公开刊物上的二手数据,这些数据可能以直方图或者表格的形式出现。由于来源渠道不一样,数据的分组情况也可能有所不同,为了满足原始调查人员的需要,这些原始数据在最初的数据搜集的时候可能已经被简化。如何有效的利用这些简化过的数据呢?本文介绍了用最大熵法处理这些数据。
目前,我国关于健康保险数据方面的问题尚无较为系统全面的探讨,笔者结合我国健康保险业务的实际情况,进行大胆的创新,不乏有一些创新点,以期在这些方面的研究中能起到抛砖引玉的作用。这些创新点主要有:(1)不仅涉及一般数据的讨论,还讨论了医疗服务数据特点和质量问题以及二手数据的处理问题;(2)笔者前瞻性地对健康保险数据管理系统提出基本的框架,对今后实务中处理健康保险数据问题具有一定的借鉴和指导意义。
限于篇幅和资料的限制,对一些具体问题没有展开深入的探讨,比如研究健康保险数据可以从多个方面着手,像如何从外部搜集数据、如何从保险公司经营环节中获取经验数据等,限于这些方面和财产保险数据及寿险数据的处理办法差不多,本文着重介绍了健康保险数据中的医疗服务数据、二手数据的处理和健康保险数据管理系统,加之时间仓促及笔者自身水平有限,文中尚存在一些的缺点和不足之处。
Since the 1980s, China's resumption of domestic insurance business, China's insurance industry has been the rapid development .In the year of 2006 ,the State Council issued the "State Council's views on the reform and development of the insurance industry "policy, raised some ardent hopes and demands on the development of health insurance. It clearly pointed out "co-ordinate the development of urban and rural commercial pension insurance and health insurance, and improve a multi-level social security system ".This means that the upcoming years will be China's health insurance a high-speed development period, for those who are aspiring health insurance this is a once-successful career opportunities.We not only need to see bright prospects of health insurance in china, but also need to be aware of the many challenges that we will facing.
The current supply and demand in China's booming commercial health insurance business and however unsatisfactory circumstances,insurance companies who provide health insurance products and services, is far from the satisfaction of people. In addition, data is the basis of risk management, and health insurancel's specialized operation is relying on data , if absence of the strong and effective database, health insurance specialization is difficult to achieve real results, so under the current situation, in order to improve health insurance business and health insurance risk management,we should study on health insurance data.
In the abroad, the research on commercial health insurance data is relatively mature, insurance companys accumulated data more perfect. China's commercial health insurance business developed in a short time, it lack of empirical data,whether insurance supervision departments,the industry standards organizations or insurance companies have failed to launch a comprehensive,authoritative and accurate health insurance industry experience data. Neither practical nor theoretical circles have studied medical services data for a detail, I think that medical data including patient basic data,the transfer of data and electronic case, clinics data, medical imaging data, medical management and economic data.As the source of medical information,it's quality is good or bad will influence extremely far-reaching. Medical data has the characteristics of complexity and uniqueness, issues of data quality a long time has not been given sufficient attention, which is the main factors of medical data quality, data monitoring is not perfect, there is no clear incentives, the lack of good monitoring tools and measures links, the lack of uniform standards for medical data, and so on .
Master risk injury which occurred in population, the health risks of quantitative characterization of the transfer and its estimated cost, not only to gather enough data, but also to collate the data for analysis. I consider that data preparation is a complex and tedious work, generally it has to go through three steps: the data resources and demand analysis, data collection preparation and data calibration. In addition, the data prepared in time,quantity,quality and other aspects of the stringent requirements, can't be flawed. Data pretreatment include the followings: fill the missing data values,smooth noisy data,identification or delete anomaly data values,data transformation and resolve inconsistencies.
The purpose of insurance to collect data is effectively operate the health insurance, comprehensive control of health insurance risk. Only through data processing can not achieve their aims ,it need data mining and data management system .I consider that data mining in the application of health insurance should achieving the following objectives (1) Customer Segmentation (2) to identify characteristics of claims. Data mining in the health insurance business should follows the following steps: (1) by the service database system as a source of data warehouse and data mart; (2) in accordance with the needs of health insurance data mining targets identified by the this response to the data mining method of data warehouse and data mart data analysis in order to obtain knowledge, and thus constitute a knowledge base; (3) will be acquired knowledge to customer service, new business promotion, marketing and new product set system; (4) evaluation of the results and feedback to the application of data mining methods to improve the process.
引文
①陈滔.健康保险精算:模型、方法和应用.北京:中国统计出版社,2006,P2
③张晓.商业健康保险.北京:中国劳动社保障出版社,2005,P114
1. Kevin Pledge . Controlling Your Data 1—Strategies .Society of Actuaries Annual Meeting and Exhibit,New York, 14 November 2005
2. Ian G. Duncan ,Henry Dove .Useful Sources of Healthcare Data, Health Services Research and Health Policy Information for Actuaries .Duncan/soa/Datasets Project ,2005(August 5)
3. Thomas Herzog, Leonid Gavrilov, Jean-Marie Robine,Bertram Kestenbaum , Ching-Syang Yue . Data Sources , Quality and Analyses .Living to 100 Symposium ,2005(1)
4. Karl. G . Volknar , James . E . Drennan , David .V. Willams, J. Michael . Crooks . Data Quality Concerns for Health Actuaries . Dallas Meeting,2001(May 30–June 1)
5. Patrick . L . Brocket, Samuel . H . Cox, Boaz Golany , Fredy Phillips, Yun Song .Actuarial Usage of Grouped Data : An Approach to Incorporating Secondary Data .Transations of Society of Actuaries,1995 (VOL. 47)
6.彭传薇、李小华.医院医疗服务数据质量现状和影响因素分析.中国医院管理, 2005(9)
7.彭传薇、刘琛玺.浅谈医疗服务数据质量主要性及其影响.解放军医院管理杂志,2005(5)
8.张劲松.保险公司数据挖掘技术应用探索.商场现代化, 2004(12)
9.李琼、文珊.商业健康保险破局之道.中国保险,2004(11)
10.刘鹏、雷蕾.缺失数据处理方法的比较研究.计算机科学,2004(10)
11.陈滔.国外健康保险精算现状分析及启示研究.上海保险,2004(5)
12.王爱荣.浅析医疗保险数据信息的采集、开发和利用.中国卫生统计, 2004(1)
13.陈滔、任仕泉.住院医疗保险精算中预期住院费用和天数估计.中国卫生统计,2001(6)
14.陈滔.健康保险精算:模型、方法和应用.北京:中国统计出版社,2006
15.卓志等.保险精算通论.成都:西南财经大学出版社,2006
16.陈文辉.健康保险管理办法指引.北京:中国劳动社会保障出版社,2006
17.任仕泉.医疗保险风险管理办法.成都:四川出版集体·四川科学技术出版社,2006
18.吴岚、王燕.风险理论.北京:中国财政经济出版社,2006
19.刘畅.中资保险公司核心竞争力培育研究.哈尔滨:东北林业大学出版社,2006
20.潘国臣.保险企业创新能力问题研究.武汉:武汉大学出版社,2006
21.张晓.商业健康保险.北京:中国劳动社保障出版社,2005
22.北京保监局《寿险公司内部控制研究》课题组.寿险公司内部控制研究.北京:经济科学出版社,2005
23.孙荣、兰虹.保险学原理.成都:西南财经大学出版社,2004
24.王绪瑾.保险学.北京:经济管理出版社,2004
25.程晓明.医疗保险学.上海:复旦大学出版社,2003
26.陈滔.医疗保险精算和风险控制方法.成都:西南财经大学出版社,2002
27.陈滔.健康保险.成都:西南财经大学出版社,2002
28.刘子操、陶阳.健康保险.北京:中国金融出版社,2001
29.陈朝先.人身保险.北京:中国金融出版社,2000
30.李秀芳.寿险精算实务.天津:南开大学出版社,2000
31.王虎峰.社会保险管理信息系统.北京:改革出版社,1999
32.韦生琼等.人身保险.成都:西南财经大学出版社,1996
33.徐文虎.保险学.上海:上海人民出版社,1994
34. http://www.circ.gov.cn
35. http://www.soa.org
③张晓.商业健康保险.北京:中国劳动社保障出版社,2005,P114
1. Kevin Pledge . Controlling Your Data 1—Strategies .Society of Actuaries Annual Meeting and Exhibit,New York, 14 November 2005
2. Ian G. Duncan ,Henry Dove .Useful Sources of Healthcare Data, Health Services Research and Health Policy Information for Actuaries .Duncan/soa/Datasets Project ,2005(August 5)
3. Thomas Herzog, Leonid Gavrilov, Jean-Marie Robine,Bertram Kestenbaum , Ching-Syang Yue . Data Sources , Quality and Analyses .Living to 100 Symposium ,2005(1)
4. Karl. G . Volknar , James . E . Drennan , David .V. Willams, J. Michael . Crooks . Data Quality Concerns for Health Actuaries . Dallas Meeting,2001(May 30–June 1)
5. Patrick . L . Brocket, Samuel . H . Cox, Boaz Golany , Fredy Phillips, Yun Song .Actuarial Usage of Grouped Data : An Approach to Incorporating Secondary Data .Transations of Society of Actuaries,1995 (VOL. 47)
6.彭传薇、李小华.医院医疗服务数据质量现状和影响因素分析.中国医院管理, 2005(9)
7.彭传薇、刘琛玺.浅谈医疗服务数据质量主要性及其影响.解放军医院管理杂志,2005(5)
8.张劲松.保险公司数据挖掘技术应用探索.商场现代化, 2004(12)
9.李琼、文珊.商业健康保险破局之道.中国保险,2004(11)
10.刘鹏、雷蕾.缺失数据处理方法的比较研究.计算机科学,2004(10)
11.陈滔.国外健康保险精算现状分析及启示研究.上海保险,2004(5)
12.王爱荣.浅析医疗保险数据信息的采集、开发和利用.中国卫生统计, 2004(1)
13.陈滔、任仕泉.住院医疗保险精算中预期住院费用和天数估计.中国卫生统计,2001(6)
14.陈滔.健康保险精算:模型、方法和应用.北京:中国统计出版社,2006
15.卓志等.保险精算通论.成都:西南财经大学出版社,2006
16.陈文辉.健康保险管理办法指引.北京:中国劳动社会保障出版社,2006
17.任仕泉.医疗保险风险管理办法.成都:四川出版集体·四川科学技术出版社,2006
18.吴岚、王燕.风险理论.北京:中国财政经济出版社,2006
19.刘畅.中资保险公司核心竞争力培育研究.哈尔滨:东北林业大学出版社,2006
20.潘国臣.保险企业创新能力问题研究.武汉:武汉大学出版社,2006
21.张晓.商业健康保险.北京:中国劳动社保障出版社,2005
22.北京保监局《寿险公司内部控制研究》课题组.寿险公司内部控制研究.北京:经济科学出版社,2005
23.孙荣、兰虹.保险学原理.成都:西南财经大学出版社,2004
24.王绪瑾.保险学.北京:经济管理出版社,2004
25.程晓明.医疗保险学.上海:复旦大学出版社,2003
26.陈滔.医疗保险精算和风险控制方法.成都:西南财经大学出版社,2002
27.陈滔.健康保险.成都:西南财经大学出版社,2002
28.刘子操、陶阳.健康保险.北京:中国金融出版社,2001
29.陈朝先.人身保险.北京:中国金融出版社,2000
30.李秀芳.寿险精算实务.天津:南开大学出版社,2000
31.王虎峰.社会保险管理信息系统.北京:改革出版社,1999
32.韦生琼等.人身保险.成都:西南财经大学出版社,1996
33.徐文虎.保险学.上海:上海人民出版社,1994
34. http://www.circ.gov.cn
35. http://www.soa.org