Exploration of hospitalization expenses for retinal disease patients based on XGboost model

Corresponding author: Wang ping, wpwy88@163.com

DOI: 10.12201/bmr.202604.00168

•  

  Abstract: Objective To explore the differences in hospitalization costs of patients with retinal diseases and the influencing factors of hospitalization expenses, in order to provide data support for hospital cost control. Method The data of patients with retinal diseases from January 1, 2021 to December 31, 2024 were collected. The Mann-Whitney U test was employed to compare the differences in hospitalization costs among various physician groups across different stratifications. Additionally, an XGBoost regression model, combined with SHAP values, was utilized to predict hospitalization costs and to analyze the factors influencing them. Results The differences in hospitalization costs among different physician groups under different stratification were statistically significant (P<0.05), and the effect size is small (δ<0.25). The impact of various factors on hospitalization costs: high-impact factors such as drug costs and DRG payment standards, medium-impact factors including laboratory fees, examination fees, age, and physician group, and low-impact factors such as fee types, consumables fees, and surgical fees. XGBoost exhibits strong predictive performance, with a training set MAPE of 7.11% and an R2 of 0.96, and a validation set MAPE of 18.64% and an R2 of 0.80. Conclusion This study reveals the key factors influencing hospitalization costs using interpretable machine learning models, which can offer a scientific basis for optimizing the allocation of medical resources, regulating diagnostic and treatment behaviors, and formulating cost control strategies that are reasonable.

  Key words: Hospitalization expenses; Attending physician groups; XGboost model; Retinal diseases; Influencing factors1

  Submit time: 28 April 2026

  Copyright: The copyright holder for this preprint is the author/funder, who has granted biomedRxiv a license to display the preprint in perpetuity.
• html

• 图表

• HUANG-yike, XUE-yunlian, LUO-yijing, ZHANG-yuqi, LIN-zipeng, JI-lianrong. Impact of DIP Reform on Hospitalization Costs and Structural Changes: An Analysis Based on 12-Year Time-Series Data from Tertiary A Hospitals. 2026. doi: 10.12201/bmr.202603.00043

  wangwenjuan, liangjuanjuan, huyangyi, caiyuanqing, fanrong, yingbaisong, zhangyuwei. Evaluation of the Impact of DRG Reform on Hospitalization Costs Based on Synthetic Control Method: An Empirical Study in Beijing Tertiary Public Hospitals. 2024. doi: 10.12201/bmr.202404.00024

  LuoYanhong, GuoYarong. Construction of a prediction model for postoperative survival of pancreatic cancer based on SMOTE-ENN combined with XGBoost algorithm. 2025. doi: 10.12201/bmr.202506.00058

  Yangjie, Taoxiangyi, Dengzhihong, Dongweili. The research for effect on visual related quality of life in patients with retinal vein occlusion and macular edema treated by Intravitreal injection. 2026. doi: 10.12201/bmr.202601.00042

  bai xingchi, li yanna, dufang, wu liqun. Analysis on the influencing factors of the medical expense burden of the dead inpatients in Shenzhen based on the first page data of medical records. 2023. doi: 10.12201/bmr.202305.00003

  DU Tao, YANG Jia Ling, ZANG Wei. Research on the influencing factors of patients willingness to first visit at the grass-roots level based on the extended TAM-TPB integrated model. 2023. doi: 10.12201/bmr.202303.00014

  LIU Yuehong, LU Xuan, DING Chaofei, JIN Fubi, CHEN Aiai, SUN Honger. Research progress on health education of patients with diabetic retinopathy. 2025. doi: 10.12201/bmr.202507.00049

  Construction of specific medical records assessment system for CAP patients disease severity based on PDCA. 2024. doi: 10.12201/bmr.202409.00064

  fengfengxiang, renhuiling, wangjunhui, caimiaozhi, zhaojiawei, wangweijie. Qualitative systematic evaluation of influencing factors and avoidance strategies of health information avoidance behavior in patients with chronic diseases. 2022. doi: 10.12201/bmr.202207.00017

  HE Lanlan, LI Danyang, SHEN Li, WU Zhonghua, ZHANG Jun, YE Yongqiang. Construction and validation of nomogram model for prolonged length of stay in patients with acute cerebral infarction based on total cerebral small vessel disease burden scores. 2025. doi: 10.12201/bmr.202501.00008

• ID	Submit time	Number	Download
  1	2026-03-30	10.12201/bmr.202604.00168V1	Download

• Public  Anonymous  To author only Submit Comments