Machine learning for the prediction of the in-hospital mortality of post-cardiac arrest patients ：a retrospective observational study

Abstract

Abstract Background Worldwide, cardiac arrest is highly prevalent and associated with a high mortality rate. Despite timely CPR, a substantial proportion of cardiac arrest deaths occur in patients who return to spontaneous circulation (ROSC).Therefore, the purpose of this study was to explore the relevant factors affecting the prognosis of patients with cardiac arrest and develop an accurate and fast prognostic prediction model through machine learning with convenient clinical information. Methods We conducted a retrospective observational study. Data from 1772 cardiac arrest patients above 18 years of age from the MIMIC database were used to develop three machine learning models, including SVM, LR, and XGBoost models, for predicting in-hospital mortality. The areas under the receiver operating characteristic curve (AUC), accuracy, precision, recall and F1 score were calculated to evaluate these models. Results In our study, the XGBoost algorithm outperformed the other algorithms. The accuracy, recall value, precision value and F1 score of the XGBoost algorithm were 0.762, 0.812, 0.765, and 0.788, respectively. In addition, the AUC of the XGBoost model was larger than those of the LR and SVM models (0.847 vs. 0.834 vs. 0.747, respectively). The top 10 most important features of the XGboost algorithm were lactate_min,gcs_min,temperature_max,weight_kg,CK_MB_max,bun_min,glucose_min,spo2_min,wbc_min,and heart_rate_min. The XGBoost algorithm provided more personalized and reliable prognostic information for cardiac arrest patients than the other algorithms. Conclusions The prognostic prediction model for patients with cardiac arrest established by the XGBoost algorithm includes indicators that had certain predictive value for disease severity in previous studies. Compared with other models, this model can provide more accurate and considerable prognostic information, facilitate communication between patients' families and doctors about the disease, and help doctors make clinical decisions.