Use of deep learning to predict acute kidney injury after intravenous contrast media administration (Preprint)

BACKGROUND Precise prediction of contrast media-induced acute kidney injury (CIAKI) is an important issue because of its relationship with worse outcomes. OBJECTIVE Herein, we examined whether a deep learning algorithm could predict the risk of intravenous CIAKI better than other machine learning and logistic regression models in patients undergoing computed tomography. METHODS A total of 14,185 cases that underwent intravenous contrast media for computed tomography under the preventive and monitoring facility in Seoul National University Hospital were reviewed. CIAKI was defined as an increase in serum creatinine ≥0.3 mg/dl within 2 days and/or ≥50% within 7 days. Using both time-varying and time-invariant features, machine learning models, such as the recurrent neural network (RNN), light gradient boosting machine, extreme boosting machine, random forest, decision tree, support vector machine, κ-nearest neighboring, and logistic regression, were developed using a training set, and their performance was compared using the area under the receiver operating characteristic curve (AUROC) in a test set. RESULTS CIAKI developed in 261 cases (1.8%). The RNN model had the highest AUROC value of 0.755 (0.708–0.802) for predicting CIAKI, which was superior to those obtained from other machine learning models. Although CIAKI was defined as an increase in serum creatinine ≥0.5 mg/dl and/or ≥25% within 3 days, the highest performance was achieved in the RNN model with an AUROC of 0.716 (0.664–0.768). In the feature ranking analysis, albumin level was the most highly contributing factor to RNN performance, followed by time-varying kidney function. CONCLUSIONS Application of a deep learning algorithm improves the predictability of intravenous CIAKI after computed tomography, representing a basis for future clinical alarming and preventive systems.