Optimization of Diagnosis-Related Groups for 14,246 Patients with Uterine Leiomyoma in a Single Center in Western China Using a Machine Learning Model

Yuan Ma,1,2 Li Li,1 Li Yu,1 Wei He,1 Ling Yi,1 Yuxin Tang,1 Jijie Li,1 Zhigang Zhong,3 Meixian Wang,4 Shiyao Huang,5,6 Yiquan Xiong,5,6 Pei Xiao,7 Yuxiang Huang1 1Department of Medical Record Management, West China Second University Hospital, Sichuan University, Chengdu, Sichuan, People’s Republic of China; 2Key Laboratory of Birth Defects and Related Diseases of Women and Children (Sichuan University), Ministry of Education, Chengdu, Sichuan, People’s Republic of China; 3Department of Prevention, Office of Cancer Prevention and Treatment, Sichuan Cancer Hospital & Institute, Sichuan Cancer Center, Cancer Hospital Affiliate to University of Electronic Science and Technology of China, Chengdu, Sichuan, People’s Republic of China; 4National Center for Birth Defects Monitoring, West China Second University Hospital, Sichuan University, Chengdu, Sichuan, People’s Republic of China; 5Chinese Evidence-Based Medicine Center, West China Hospital, Sichuan University, Chengdu, Sichuan, People’s Republic of China; 6NMPA Key Laboratory for Real World Data Research and Evaluation in Hainan, Chengdu, Sichuan, People’s Republic of China; 7Medical Insurance Office, West China Fourth Hospital, Sichuan University, Chengdu, Sichuan, People’s Republic of ChinaCorrespondence: Yuxiang Huang, Email 3593213104@qq.comBackground: Uterine leiomyoma (UL) is one of the most common benign tumors in women, and its incidence is gradually increasing in China. The clinical complications of UL have a negative impact on women’s health, and the cost of treatment poses a significant burden on patients. Diagnosis-related groups (DRG) are internationally recognized as advanced healthcare payment management methods that can effectively reduce costs. However, there are variations in the design and grouping rules of DRG policies across different regions. Therefore, this study aims to analyze the factors influencing the hospitalization costs of patients with UL and optimize the design of DRG grouping schemes to provide insights for the development of localized DRG grouping policies.Methods: The Mann–Whitney U-test or the Kruskal–Wallis H-test was employed for univariate analysis, and multiple stepwise linear regression analysis was utilized to identify the primary influencing factors of hospitalization costs for UL. Case combination classification was conducted using the exhaustive chi-square automatic interactive detection (E-CHAID) algorithm within a decision tree framework.Results: Age, occupation, number of hospitalizations, type of medical insurance, Transfer to other departments, length of stay (LOS), type of UL, admission condition, comorbidities and complications, type of primary procedure, other types of surgical procedures, and discharge method had a significant impact on hospitalization costs (P< 0.05). Among them, the type of primary procedure, other types of surgical procedures, and LOS were the main factors influencing hospitalization costs. By incorporating the type of primary procedure, other types of surgical procedures, and LOS into the decision tree model, patients were divided into 11 DRG combinations.Conclusion: Hospitalization costs for UL are mainly related to the type of primary procedure, other types of surgical procedures, and LOS. The DRG case combinations of UL based on E-CHAID algorithm are scientific and reasonable.Keywords: uterine leiomyoma, diagnosis-related groups, decision tree