Optimization of hospitalization duration and cost structure for chronic schizophrenia patients in psychiatric...

Chronic schizophrenia patients in psychiatric hospitals often have prolonged stays, high insurance resource consumption, and low efficiency. Studies quantifying the joint optimization of payment and operational efficiency are lacking, complicating the trade-off between cost control and relapse prevention. This study aimed to identify factors associated with hospitalization duration and costs using real-world data and machine learning.

A retrospective study was conducted on inpatients with chronic schizophrenia admitted to Huai'an No.3 People's Hospital from 2022 to 2024. Data on demographics, clinical features, comorbidities, medications, and insurance costs were collected. Factors were screened by descriptive, univariate, and Gamma stepwise regression. Machine learning models (Lasso, Random Forest, XGBoost) were built to predict length of stay, with SHAP used for feature interpretation.

A total of 3607 inpatients with chronic schizophrenia were enrolled. Univariate analysis revealed 29 significant variables; Gamma regression identified 18 independent factors. The rehabilitation treatment cost ratio was the strongest risk factor, and the examination cost ratio the strongest protective factor. The Random Forest model performed best (cross-validation R²=0.7143, test R²=0.7318). SHAP analysis showed total hospitalization cost and rehabilitation treatment cost ratio as core predictors.

This study identified key factors influencing hospitalization duration and costs in chronic schizophrenia, and developed an accurate and stable model that can support lean management in psychiatric hospitals.