Comparison of logistic regression and machine learning models predicting low SpO2 during one-lung ventilation in patients undergoing thoracoscopic partial pulmonary resection
Objective To compare the predictive effects of logistic regression and machine learning models on occurrence of low peripheral oxygen saturation(SpO2)during one-lung ventilation(OLV)in pa-tients undergoing thoracoscopic partial pulmonary resection(TPPR),and to explore risk factors of low SpO2.Methods A total of 127 patients undergoing unilateral TPPR from August 1,2022 to April 30,2023 were enrolled,61 males and 66 females,aged 18-80 years,ASA physical status Ⅰ-Ⅲ.Based on whether intraoperative SpO2 during OLV was less than 90％,the patients were divided into two groups:low SpO2 group(n=21)and normal SpO2 group(n=106).Perioperative data were collected and a predic-tive model was constructed using logistic regression.This model was compared with predictive models con-structed using five machine learning models,including random forest(RF),extreme gradient boosting(XGBoost),decision tree(DT),logistic regression(LogR),and support vector machine(SVM).The re-ceiver operating characteristic(ROC)curve was plotted,and the performance of the predictive models were evaluated by the area under the curve(AUC).The best output model was interpreted using Shapley additive explanations(SHAP)to identify the risk factors of low SpO2 during OLV in patients undergoing TPPR.Results Multivariate logistic regression analysis showed that increased age(OR=1.087,95％CI 1.006-1.175,P=0.036),increased BMI(OR=1.299,95％CI 1.050-1.608,P=0.016),increased pre-operative blood glucose(OR=2.028,95％CI 1.378-2.983,P＜0.001),and decreased RV/TLC％Pred(OR=0.936,95％CI 0.892-0.983,P=0.008)were independent risk factors of low SpO2 during OLV.The predictive model was Logit(p)=-10.098+0.08 × age+0.231 × BMI+0.633 × blood glu-cose-0.059 × RV/TLC％Pred,with an AUC of 0.873(95％CI 0.803-0.943,P＜0.001).After optimi-zing parameters of machine learning models using grid search combined with five-fold cross-validation,the model training results were satisfactory.ROC curve analysis showed that the AUC for RF was 0.921(95％CI 0.840-0.979),XGBoost was 0.940(95％CI 0.812-0.981),DT was 0.919(95％CI 0.828-0.982),LogR was 0.892(95％CI 0.831-0.980),and SVM was 0.922(95％CI 0.832-0.982).XG-Boost had the highest AUC,surpassing the logistic regression model.SHAP analysis indicated that the most important risk factors in the XGBoost output model were increased age,BMI,and preoperative blood glucose concentration.Conclusion Increased age,BMI,and preoperative blood glucose concentration are signifi-cant risk factors for low SpO2 during OLV in patients undergoing TPPR.The XGBoost machine learning model outperformed traditional logistic regression in predicting the occurrence of low SpO2 during OLV.XG-Boost can analyze more complex relationships between variables and outcomes and provide more accurate in-dividualized predictions of the risk of low SpO2 during OLV.
Machine learningThoracoscopic partial pulmonary resectionOne-lung ventilationAgeBody mass indexBlood glucose
万方数据
