基于主成分分析的医院科研评价方法及实证研究

An empirical study on the evaluation system of hospital scientific research based on principal component analysis

目的通过主成分分析的方法, 构建科研评价模型, 探索科学的医院科研评价方法。方法收集2019—2020年吉林大学第一医院科研人员职称、学历、岗位以及科研产出信息, 采用德尔菲专家咨询的方法, 确定各变量的赋值, 应用SPSS 21.0软件构建主成分分析模型并进行模型验证。结果研究共收集了1 882名科研人员的信息, 效度检验的KMO值(0.731)及Bartlett球形检验(P<0.05)符合主成分分析要求;模型共获得7个主成分, 其中, 主成分一代表科研人员发表SCI论文、申请国家级、省部级科研项目和学会兼职情况, 主成分二代表科研人员申请专利和发表论著的情况, 主成分三代表科研人员科技获奖情况;按照学科对科研人员科研产出分值进行汇总排序, 结果发现神经内科、内分泌代谢科、神经外科、普外科和骨科排名较好;模型验证结果发现职称为高级、学历为博士的科研人员中位加权综合评分最高(P<0.05), 提示职称等级越高、学历越高的科研人员综合科研产出评分越高。结论该研究构建的科研评价模型能为医院科研评价提供科学的数据参考。

Objective To construct a scientific research evaluation model through principal component analysis,and to explore scientific research evaluation methods for hospitals.Methods The professional title,educational background,positions and scientific research output information of the scientific research personnel in the First Hospital of Jilin University from 2019 to 2020 were collected.Delphi expert consultation was used to determine the assignment value of each variable,and use SPSS 21.0 software was used to build the principal component analysis model and conduct model verification.Results The study collected a total of 1882 researchers′information.The KMO value of the validity test and the Bartlett sphere test meet the requirements of principal component analysis(KMO=0.731,P<0.05);the model obtained a total of 7 principal components.Among them,principal component 1 represents researchers who published SCI papers,applying for national,provincial and ministerial level scientific research projects,and their part-time positions in academic societies.The second principal component represents the status of applying for patents and publications,and the third principal component represents the status of the awards.The scores of scientific research output of researchers were summarized and sorted according to disciplines,according to which the neurology,endocrinology and metabolism,neurosurgery,general surgery and orthopedics ranked better.The model verification results found that researchers with senior professional titles and doctoral degrees had the highest median weighted comprehensive score(P<0.05),suggesting that scholars with higher professional title levels and higher education received higher comprehensive scientific research output scores.Conclusions The scientific research evaluation model constructed by this study can provide scientific data reference for the hospital scientific research evaluation.