UCLA Library

Objective: We sought to several develop parsimonious machine learning (ML) models to predict resource utilization and clinical outcomes following cardiac operations using only preoperative factors.

Methods: All patients undergoing coronary artery bypass grafting and/or valve operations were identified in the 2015-2021 University of California Cardiac Surgery Consortium repository. The primary endpoint of the study was length of stay (LOS). Secondary endpoints included 30-day mortality, acute kidney injury (AKI), reoperation, postoperative blood transfusion and duration of intensive care unit admission (ICU LOS). Linear regression, gradient boosted machines (GBM), random forest (RF), extreme gradient boosting (XGBoost) predictive models were developed. The coefficient of determination (R2) and area under the receiver operating characteristic (AUC) were used to compare models. Important predictors of increased resource use were identified using SHapley summary plots.

Results: Compared to all other modeling strategies, GBM demonstrated the greatest performance in the prediction of LOS (R2 0.42), ICU LOS (R2 0.23) and 30-day mortality (AUC 0.69). Advancing age, reduced hematocrit and multiple-valve procedures were associated with increased LOS and ICU LOS. Furthermore, the GBM model best predicted AKI (AUC 0.76), while RF exhibited greatest discrimination in the prediction of postoperative transfusion (AUC 0.73). We observed no difference in performance between modeling strategies for reoperation (AUC 0.80).

Conclusion: Our findings affirm the utility of ML in the estimation of resource use and clinical outcomes following cardiac operations. We identified several risk factors associated with increased resource use, which may be used to guide case scheduling in times of limited hospital capacity.