Development and performance assessment of novel machine learning models to predict pneumonia after liver transplantation

Abstract Background Pneumonia is the most frequently encountered postoperative pulmonary complications (PPC) after orthotopic liver transplantation (OLT), which cause high morbidity and mortality rates. We aimed to develop a model predict pneumonia in OLT patients using machine learning (ML) methods. Methods Data of 786 adult underwent at Third Affiliated Hospital Sun Yat-sen University from January 2015 September 2019 was retrospectively extracted electronic medical records randomly subdivided into training set testing set. With set, six ML models including logistic regression (LR), support vector (SVM), random forest (RF), adaptive boosting (AdaBoost), extreme gradient (XGBoost) (GBM) were developed. These assessed by area under curve (AUC) receiver operating characteristic on The related risk factors outcomes also probed based chosen model. Results 591 eventually included 253 (42.81%) diagnosed with pneumonia, associated increased hospitalization ( P < 0.05). Among models, XGBoost performed best. AUC 0.734 (sensitivity: 52.6%; specificity: 77.5%). notably 14 items features: INR, HCT, PLT, ALB, ALT, FIB, WBC, PT, serum Na + , TBIL, anesthesia time, preoperative length stay, total fluid transfusion operation time. Conclusion Our study firstly demonstrated that common variables might patients.