Prediction of acute kidney injury risk after cardiac surgery: using a hybrid machine learning algorithm

• Published in: BMC medical informatics and decision making Vol. 22; no. 1; pp. 137 - 10
• Main Authors: Petrosyan, Yelena, Mesana, Thierry G., Sun, Louise Y.
• Format: Journal Article
• Language: English
• Published: London BioMed Central 18.05.2022; BioMed Central Ltd; Springer Nature B.V; BMC
• Subjects: Acute Kidney Injury - diagnosis; Acute Kidney Injury - epidemiology; Acute Kidney Injury - etiology; Acute renal failure; Adult; Algorithms; Artificial intelligence; Calibration; Cardiac surgery-associated acute kidney injury; Cardiac Surgical Procedures - adverse effects; Cardiovascular disease; Care and treatment; Classification; Computer applications; Coronary artery bypass; Data mining; Datasets; Diagnosis; Feature selection; Generalized linear models; Goodness of fit; Health aspects; Health Informatics; Health risk assessment; Health risks; Heart; Heart surgery; Humans; Information Systems and Communication Service; Kidneys; Learning algorithms; Machine Learning; Management of Computing and Information Systems; Mathematical models; Medicine; Medicine & Public Health; Methods; Model accuracy; Mortality; Patient outcomes; Patients; Predictive modeling; Random Forests; Regression models; Retrospective Studies; Risk; Risk Assessment; Risk Factors; Risk groups; Statistical analysis; Statistical methods; Statistical models; Statistical tests; Surgery; Canada; United States > US; Cardiac surgery-associated acute kidney injury; Random Forests; Predictive modeling; Data mining; Machine Learning
• ISSN: 1472-6947; 1472-6947
• DOI: 10.1186/s12911-022-01859-w

Background Acute kidney injury (AKI) is a serious complication after cardiac surgery. We derived and internally validated a Machine Learning preoperative model to predict cardiac surgery-associated AKI of any severity and compared its performance with parametric statistical models. Methods We conducted a retrospective study of adult patients who underwent major cardiac surgery requiring cardiopulmonary bypass between November 1st, 2009 and March 31st, 2015. AKI was defined according to the KDIGO criteria as stage 1 or greater, within 7 days of surgery. We randomly split the cohort into derivation and validation datasets. We developed three AKI risk models: (1) a hybrid machine learning (ML) algorithm, using Random Forests for variable selection, followed by high performance logistic regression; (2) a traditional logistic regression model and (3) an enhanced logistic regression model with 500 bootstraps, with backward variable selection. For each model, we assigned risk scores to each of the retained covariate and assessed model discrimination (C statistic) and calibration (Hosmer–Lemeshow goodness-of-fit test) in the validation datasets. Results Of 6522 included patients, 1760 (27.0%) developed AKI. The best performance was achieved by the hybrid ML algorithm to predict AKI of any severity. The ML and enhanced statistical models remained robust after internal validation (C statistic = 0.75; Hosmer–Lemeshow p  = 0.804, and AUC = 0.74, Hosmer–Lemeshow p  = 0.347, respectively). Conclusions We demonstrated that a hybrid ML model provides higher accuracy without sacrificing parsimony, computational efficiency, or interpretability, when compared with parametric statistical models. This score-based model can easily be used at the bedside to identify high-risk patients who may benefit from intensive perioperative monitoring and personalized management strategies.