Optimization of diagnosis-related groups for patients with acute appendicitis using a machine learning model.

Background: The diagnosis-related groups prospective payment system (DRG-PPS) is widely implemented worldwide. Its core components include disease classification and pricing mechanisms. Developing a disease grouping and pricing approach that aligns with local conditions is essential. This study examines the factors influencing hospitalization costs for acute appendicitis (AA) patients and proposes strategies for disease grouping and pricing. Methods: Stratified random sampling was used to select research sites from provincial, municipal, and county hospitals in Hefei, China. Data were obtained from the hospitalization information systems of three hospitals from 2017 to 2019. The primary diagnosis was defined as AA. Single-factor analysis and multiple linear stepwise regression were used to identify the main factors influencing hospitalization costs. Additionally, a classification and regression tree (CART) model, based on the exhaustive chi-square automatic interaction detection (E-CHAID) algorithm, was applied to establish the DRG grouping model. Results: A total of 4,066 patients were included. Significant differences in hospitalization costs were observed based on length of stay (LOS), marital status, surgery, and hospital level ( Conclusion: Hospitalization costs for AA patients are primarily influenced by LOS, marital status, surgery, and hospital level. The decision tree model provides a basis for DRG grouping. Health administration departments may consider implementing precise and individualized hospitalization cost reimbursement mechanisms accordingly.