基于血常规、血气分析、肺功能及凝血功能指标构建的模型对慢性阻塞性肺疾病急性加重期患者预后的预测价值OA

Objective To explore the prognostic value of a model constructed based on routine blood tests,blood gas analysis,pulmonary function tests and coagulation function indicators in patients with acute exacerbation of chronic obstructive pulmonary disease(AECOPD).Methods A total of 210 patients with AECOPD admitted to Liuzhou Liutie Central Hospital from July 2023 to July 2024 were retrospectively collected as the training set,and another 90 patients with AECOPD treated from August 2024 to January 2025 were selected as the validation set.After 3 months of follow-up,they were subdivided into good prognosis subgroup and poor prognosis subgroup according to 3-month readmission status.The differences in routine blood tests,blood gas analysis,pulmonary function tests and coagulation function indicators between the two prognosis subgroups in the training set and the validation set were analyzed.LASSO regression screening was performed,and XGBoost was used to construct a prediction model for the prognosis of AECOPD patients.The importance of each feature was quantified by Shapley additive explanations(SHAP)value.The receiver operating characteristic(ROC)curve,calibration curve and decision curve analysis(DCA)were performed to validate the model.Results In the training set and validation set,there were significant differences in pulmonary function classification,smoking history,neutrophil-to-lymphocyte ratio(NLR),C-reactive protein(CRP),partial pressure of carbon dioxide(PCO2),percentage of predicted forced vital capacity(FVC％),percentage of predicted forced expiratory volume in one second(FEV1％),FEV1/FVC,fibrinogen(FIB)and D-dimer(D-D)between the good prognosis subgroup and poor prognosis subgroup(P＜0.05).Seven variables were selected by Lasso regression,which were smoking history,NLR,CRP,PCO2,FEV1/FVC,FIB and D-D,and the XGBoost model was established accordingly.In the training set and validation set,the AUC values of the ROC curve of the XGBoost model for predicting the risk of poor prognosis in AECOPD patients were higher.Hosmer-Lemeshow test revealed no significant difference between the prediction probability and the observation results of the XGBoost model in the training set and the validation set(P＞0.05),and the prediction results were in good agreement with the actual results.DCA demonstrated that the XGBoost model in the training set and validation set had good clinical practicability in the range of 35％-80％threshold probability.Conclusion The XGBoost model based on routine blood tests,blood gas analysis,pulmonary function test and coagulation function indicators has a good predictive value for the prognosis of AECOPD patients.The model also has good discrimination and calibration,which can provide some reference for clinical prediction of the prognosis of AECOPD patients.