Objective To predict the hospitalization expense of pneumonia patients using the CatBoost and LightGBM models and explore associated influencing factors,providing a scientific basis for reasonable control of medical expenses and alle-viating the financial burden on society and patients'families.Methods Data from 1407 inpatients with pneumonia admitted to a tertiary hospital between November 1,2021 and January 31,2023 were collected from the hospital information system.The Cat-Boost and LightGBM models were employed to predict the hospitalization expenses of pneumonia patients and analyze the influen-cing factors.Results The full-variable CatBoost model fitted better than the full-variable LightGBM model,with the R-square value of 0.859 and Mean Absolute Percentage Error(MAPE)of 0.352.The full-variable CatBoost model predicted better than the full-variable LightGBM model,with the R-square value of 0.820 and MAPE of 0.363.Notably,the average length of stay was the most important factor affecting the hospitalization expenses of pneumonia patients.Conclusion Compared to the LightGBM model,the CatBoost model indicated more advantages and higher accuracy in predicting hospitalization expenses for pneumonia patients.These accurate predictions of hospitalization expenses can provide decision-making references for hospital administra-tors.Therefore,on the premise of ensuring medical quality,reasonably reducing the length of hospitalization can effectively con-trol the increase in hospitalization expenses.

Objective To predict the hospitalization expense of pneumonia patients using the CatBoost and LightGBM models and explore associated influencing factors,providing a scientific basis for reasonable control of medical expenses and alle-viating the financial burden on society and patients'families.Methods Data from 1407 inpatients with pneumonia admitted to a tertiary hospital between November 1,2021 and January 31,2023 were collected from the hospital information system.The Cat-Boost and LightGBM models were employed to predict the hospitalization expenses of pneumonia patients and analyze the influen-cing factors.Results The full-variable CatBoost model fitted better than the full-variable LightGBM model,with the R-square value of 0.859 and Mean Absolute Percentage Error(MAPE)of 0.352.The full-variable CatBoost model predicted better than the full-variable LightGBM model,with the R-square value of 0.820 and MAPE of 0.363.Notably,the average length of stay was the most important factor affecting the hospitalization expenses of pneumonia patients.Conclusion Compared to the LightGBM model,the CatBoost model indicated more advantages and higher accuracy in predicting hospitalization expenses for pneumonia patients.These accurate predictions of hospitalization expenses can provide decision-making references for hospital administra-tors.Therefore,on the premise of ensuring medical quality,reasonably reducing the length of hospitalization can effectively con-trol the increase in hospitalization expenses.