聚类分析和关联分析在三级甲等医院科研绩效评价中的应用  

Application of cluster and association analyses in research performance assessment in a grade A tertiary hospital

作　　者：郑富豪[1] 沈金波[2] Zheng Fuhao;Shen Jinbo(Research Administration Department of Fujian Provincial Hospital,Fujian Provincial Clinical College of Fujian Provincial Hospital,Fujian Provincial Hospital of Geriatrics,Fuzhou 350001,China;Center for Information Management of Fujian Provincial Hospital,Fuzhou 350001,China)

机构地区：[1]福建省立医院科研处、福建省立临床学院、福建省老年医院,福州350001 [2]福建省立医院信息管理中心

出　　处：《中国医疗管理科学》2020年第6期23-28,共6页Chinese Journal Of Medical Management Sciences

基　　金：福建省科技厅软科学课题“基于人工智能的公立医院科研评价体系研究和应用”(2019R0111);福建省卫生健康软科学研究课题“基于知识发现的科研管理决策新模式研究与应用”(2019-RK-6)。

摘　　要：目的优化医院科研绩效评价,以提升不同等级科研人员的科研能力。方法收集汇总2014年~2018年医院科研人员的基本信息、科研成果等数据,采用K-means聚类分析算法、Apriori关联分析算法对科研绩效数据进行分析评价。结果优秀、良好两个等级科研人员占比为25.8%,聚类结果与医院实际科研现状较接近;临床医生是医院科研主力,分别在整体、优秀、良好中占比为69.5%、83.33%、72.92%;评价等级为优秀的与博士研究生学历、正高职称有较强相关性;评价等级为合格的与工龄不超过10年、学历为大学本科或专科、职称为初级或中级、科室类别为护理或行政等特征有一定相关性,置信度超过80%。结论分析聚类分析、关联在科研绩效评价中应用,可以总结出不同等级科研人员的特点规律,医院应为不同等级科研人员制定有针对性的科研管理措施,制定相应激励政策,加强不同等级科研人员交流和互动,组建优秀科研团队,加强与知名院校交流合作,提升影响力和促进资源整合。Objective To optimize the research performance assessment of hospitals and improve the capabilities of researchers at different academic levels.Methods The basic information and scientific research outcome of hospital researchers from 2014 to 2018 were collected,and the performance data of scientific research was evaluated using K-means clustering method and Apriori association algorithm.Results The performance evaluation was divided into three levels:excellent,good,qualified.The percentage of excellent and good scientific researchers was 25.8%.The clustering results were close to the actual research status of the hospital.Clinicians are the main scientific research force in hospitals,accounting for 69.5%,83.33%,and 72.92%of the overall,excellent,and good levels,respectively.The excellent level had a strong correlation with doctoral degree and high academic titles.The qualified level had certain correlation with the characteristics of not more than 10 years of work experience,bachelor degree or college degree of educational background,primary or middle academic titles,and nursing or administration department category.Conclusion The application of clustering analysis and association analysis in scientific research performance evaluation can be used to summarize the characteristics of different levels of researchers.Hospitals should develop targeted scientific research management measures for researchers at different levels,formulate corresponding incentive policies,strengthen exchanges and interactions between researchers at different levels,form an excellent scientific research team,enhance exchange and cooperation with well-known universities,and upgrade reputation and resource integration.

关 键 词：聚类分析 K-means 关联分析 Apriori 科研绩效评价 

分 类 号：TP399[自动化与计算机技术—计算机应用技术] R197[自动化与计算机技术—计算机科学与技术]