Application of K－means cluster analysis in the assessment of scientific research performance of clinical departments in a tertiary general hospital
10.3760/cma.j.cn113565-20250321-00071
- VernacularTitle:K－means聚类分析在某三甲综合性医院临床科室科研绩效评估中的应用
- Author:
Liujing CHEN
1
;
Tingyu MA
Author Information
1. 新疆医科大学公共卫生学院，乌鲁木齐 830017
- Publication Type:Journal Article
- Keywords:
K-means clustering method;
Elbow method;
Gap Statistic;
Scientific research performance;
Assessment
- From:
Chinese Journal of Medical Science Research Management
2025;38(5):394-400
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To determine the optimal number of clusters by joint use of the Elbow Method and Gap Statistic and to classify the scientific research performance of each clinical department in a tertiary general hospital by K-means cluster analysis and analyze the weak points of scientific research in each type of department, and provide targeted measures to enhance the scientific research capability of the department.Methods:Based on the scientific research performance evaluation system of clinical departments in established tertiary general hospitals, the scientific research data from 2017—2021 were substituted into the optimization system to generate quantitative scores, and the number of optimized clusters was derived through the Elbow method and the Gap Statistic algorithm, and the K-means algorithm was applied to achieve departmental hierarchical clustering.Results:The 45 clinical departments were clustered into 4 categories, and the mean of the total research performance scores of all departments was 23.118, with the highest mean score for the management system and the lowest for the research organization platform. Except for the indicator of books of publications, the remaining 15 secondary indicators had good differentiation ( P<0.05). Conclusions:The combined use of the Elbow method and the Gap Statistic algorithm balances computational efficiency with statistical rigor and strengthens the scientific explanatory power of the clustering results. K-means cluster analysis effectively divides the types of departments. It provides a certain reference basis for identifying the strengths and shortcomings of scientific research development in the department, optimizing the allocation of scientific research resources, adjusting the focus of assessment and formulating the development strategy of the discipline.
