Development and external validation of a machine learning model for predicting immediate postoperative instability in adults undergoing cardiac surgery

Background Patients undergoing cardiac surgery are at high risk of major early postoperative ficomplications that may lead to hemodynamic instability. Predicting and managing immediate postoperative adverse events remain challenging owing to the complex and nonlinear interplay of numerous risk factors. This study aimed to develop and externally validate a machine-learning (ML) model to predict composite instability outcomes during the immediate postoperative period after cardiac surgery. Methods Adult patients who underwent cardiac surgery at Seoul National University Hospital (SNUH) between October 2004 and October 2021 were included in the model development and internal validation. Thirty-seven preoperative and intraoperative variables were used as the model inputs. The primary outcome was a composite of reoperation for bleeding, death, cardiac arrest, and initiation of mechanical circulatory support with extracorporeal membrane oxygenation (ECMO) or intra-aortic balloon pump (IABP). The model performance was assessed using the area under the receiver operating characteristic curve (AUROC) and corresponding 95% confidence intervals (CIs). For external validation, we analyzed data from Seoul National University Bundang Hospital (SNUBH) and Severance Hospital. Results A total of 7,946 patients from SNUH were used for model development, with external validation performed in 2,270 patients from SNUBH and 1,966 patients from Severance Hospital. The incidence of the composite outcome was 5.93%, 1.37%, and 2.49% at SNUH, SNUBH, and Severance Hospital, respectively. The gradient boosting machine model achieved an AUROC of 0.804 (95% CI, 0.707–0.891) in internal validation, and 0.735 (95% CI, 0.636–0.829) and 0.712 (95% CI, 0.630–0.795) in external validations at SNUBH and Severance Hospital, respectively. Conclusions The proposed ML model, which incorporates preoperative and intraoperative data, demonstrated robust performance and generalizability in predicting immediate postoperative adverse events associated with severe hemodynamic instability after cardiac surgery.