Machine learning-based prediction of postoperative mortality in emergency colorectal surgery: A retrospective, multicenter cohort study using Tokushukai Medical Database

Abstract

Abstract Background Emergency colorectal surgery may constitute surgical challenges, resulting in high mortality and morbidity rates. Although prognostic factors associated with mortality in patients with emergency colorectal surgery have been identified, an accurate mortality risk assessment is still necessary to determine the range of therapeutic resources in accordance with the severity of patients. We established machine-learning models with nonlinear feature extraction to predict in-hospital mortality for patients who had emergency colorectal surgery using clinical data at admission and attempted to identify prognostic factors associated with in-hospital mortality.Methods This retrospective cohort study included adult patients undergoing emergency colorectal surgery in 42 hospitals between 2012 and 2020. Patients were divided into those hospitalized between July 2010 and June 2018 (training/validation dataset) and those hospitalized between July 2018 and June 2020 (testing dataset). We employed logistic regression and three supervised machine-learning models: random forests, gradient-boosting decision trees (GBDT), and multilayer perceptron (MLP) in the training dataset. The prediction models were tested using all testing datasets, and the area under the receiver operating characteristics curve (AUROC) was calculated for each model. The Shapley additive explanations (SHAP) values are also calculated to identify the significant variables in GBDT.Results There were 8,792 patients who underwent emergency colorectal surgery. The in-hospital mortality rates were 11.9% and 11.3% for the training/validation and testing datasets, respectively. After model training, the AUROC was calculated for in-hospital mortality prediction with each trained machine-learning model. Therefore, the AUROC values of 0.742, 0.782, 0.814, and 0.768 were obtained for logistic regression, random forests, GBDT, and MLP. According to SHAP values, age, colorectal cancer, use of laparoscopy, and some laboratory variables, including serum lactate dehydrogenase serum albumin, and blood urea nitrogen, were significantly associated with in-hospital mortality.Conclusion We successfully generated the machine-learning prediction model, including GBDT, with the best prediction performance and exploited the potential for use in evaluating in-hospital mortality risk for patients who undergo emergency colorectal surgery.