Machine Learning Based Predictive Modeling of Readmissions Following Extracorporeal Membrane Oxygenation Hospitalizations

Overview

Journal Surg Open Sci

Specialty General Surgery

Date 2024 Apr 24

PMID 38655069

Authors

Jeffrey Balian

Sara Sakowitz

Arjun Verma

Amulya Vadlakonda

Emma Cruz

Konmal Ali

Peyman Benharash

Affiliations

Soon will be listed here.

Abstract

Background: Despite increasing utilization and survival benefit over the last decade, extracorporeal membrane oxygenation (ECMO) remains resource-intensive with significant complications and rehospitalization risk. We thus utilized machine learning (ML) to develop prediction models for 90-day nonelective readmission following ECMO.

Methods: All adult patients receiving ECMO who survived index hospitalization were tabulated from the 2016-2020 Nationwide Readmissions Database. Extreme Gradient Boosting (XGBoost) models were developed to identify features associated with readmission following ECMO. Area under the receiver operating characteristic (AUROC), mean Average Precision (mAP), and the Brier score were calculated to estimate model performance relative to logistic regression (LR). Shapley Additive Explanation summary (SHAP) plots evaluated the relative impact of each factor on the model. An additional sensitivity analysis solely included patient comorbidities and indication for ECMO as potential model covariates.

Results: Of ∼22,947 patients, 4495 (19.6 %) were readmitted nonelectively within 90 days. The XGBoost model exhibited superior discrimination (AUROC 0.64 vs 0.49), classification accuracy (mAP 0.30 vs 0.20) and calibration (Brier score 0.154 vs 0.165, all P < 0.001) in predicting readmission compared to LR. SHAP plots identified duration of index hospitalization, undergoing heart/lung transplantation, and Medicare insurance to be associated with increased odds of readmission. Upon sub-analysis, XGBoost demonstrated superior disclination compared to LR (AUROC 0.61 vs 0.60, P < 0.05). Chronic liver disease and frailty were linked with increased odds of nonelective readmission.

Conclusions: ML outperformed LR in predicting readmission following ECMO. Future work is needed to identify other factors linked with readmission and further optimize post-ECMO care among this cohort.

References

Schmidt M, Burrell A, Roberts L, Bailey M, Sheldrake J, Rycus P . Predicting survival after ECMO for refractory cardiogenic shock: the survival after veno-arterial-ECMO (SAVE)-score. Eur Heart J. 2015; 36(33):2246-56. DOI: 10.1093/eurheartj/ehv194. View

van Walraven C, Austin P, Jennings A, Quan H, Forster A . A modification of the Elixhauser comorbidity measures into a point system for hospital death using administrative data. Med Care. 2009; 47(6):626-33. DOI: 10.1097/MLR.0b013e31819432e5. View

Abbasi A, Karasu Y, Li C, Sodha N, Eickhoff C, Ventetuolo C . Machine learning to predict hemorrhage and thrombosis during extracorporeal membrane oxygenation. Crit Care. 2020; 24(1):689. PMC: 7727105. DOI: 10.1186/s13054-020-03403-6. View

Morisson L, Duceau B, Do Rego H, Lancelot A, Hariri G, Charfeddine A . A new machine learning algorithm to predict veno-arterial ECMO implantation after post-cardiotomy low cardiac output syndrome. Anaesth Crit Care Pain Med. 2022; 42(1):101172. DOI: 10.1016/j.accpm.2022.101172. View

Mehta A, Taylor J, Day G, Lane T, Douglas I . Disparities in Adult Patient Selection for Extracorporeal Membrane Oxygenation in the United States: A Population-Level Study. Ann Am Thorac Soc. 2023; 20(8):1166-1174. PMC: 10405618. DOI: 10.1513/AnnalsATS.202212-1029OC. View

Pfob A, Lu S, Sidey-Gibbons C . Machine learning in medicine: a practical introduction to techniques for data pre-processing, hyperparameter tuning, and model comparison. BMC Med Res Methodol. 2022; 22(1):282. PMC: 9624048. DOI: 10.1186/s12874-022-01758-8. View

Nuqali A, Goyal A, Acharya P, Mastoris I, Dalia T, Chan W . Thirty-day readmissions among patients with cardiogenic shock who underwent extracorporeal membrane oxygenation support in the United States: Insights from the nationwide readmissions database. Am Heart J Plus. 2024; 13:100076. PMC: 10978167. DOI: 10.1016/j.ahjo.2021.100076. View

Xue B, Li D, Lu C, King C, Wildes T, Avidan M . Use of Machine Learning to Develop and Evaluate Models Using Preoperative and Intraoperative Data to Identify Risks of Postoperative Complications. JAMA Netw Open. 2021; 4(3):e212240. PMC: 8010590. DOI: 10.1001/jamanetworkopen.2021.2240. View

McCarthy F, McDermott K, Kini V, Gutsche J, Wald J, Xie D . Trends in U.S. Extracorporeal Membrane Oxygenation Use and Outcomes: 2002-2012. Semin Thorac Cardiovasc Surg. 2015; 27(2):81-8. PMC: 4780346. DOI: 10.1053/j.semtcvs.2015.07.005. View

10.

Nieman C, Pitman K, Tufaro A, Eisele D, Frick K, Gourin C . The effect of frailty on short-term outcomes after head and neck cancer surgery. Laryngoscope. 2017; 128(1):102-110. DOI: 10.1002/lary.26735. View

11.

Saito T, Rehmsmeier M . The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLoS One. 2015; 10(3):e0118432. PMC: 4349800. DOI: 10.1371/journal.pone.0118432. View

12.

Ayers B, Wood K, Gosev I, Prasad S . Predicting Survival After Extracorporeal Membrane Oxygenation by Using Machine Learning. Ann Thorac Surg. 2020; 110(4):1193-1200. DOI: 10.1016/j.athoracsur.2020.03.128. View

13.

LaValley M . Logistic regression. Circulation. 2008; 117(18):2395-9. DOI: 10.1161/CIRCULATIONAHA.106.682658. View

14.

Herrin J, St Andre J, Kenward K, Joshi M, Audet A, Hines S . Community factors and hospital readmission rates. Health Serv Res. 2014; 50(1):20-39. PMC: 4319869. DOI: 10.1111/1475-6773.12177. View

15.

Sandrio S, Thiel M, Krebs J . The Outcome Relevance of Pre-ECMO Liver Impairment in Adults with Acute Respiratory Distress Syndrome. J Clin Med. 2023; 12(14). PMC: 10381435. DOI: 10.3390/jcm12144860. View

16.

Tashtish N, Al-Kindi S, Karnib M, Zanath E, Mitchell S, Di Felice C . Causes and predictors of 30-day readmissions in patients with cardiogenic shock requiring extracorporeal membrane oxygenation support. Int J Artif Organs. 2019; 43(4):258-267. DOI: 10.1177/0391398819882025. View

17.

Isabona J, Imoize A, Kim Y . Machine Learning-Based Boosted Regression Ensemble Combined with Hyperparameter Tuning for Optimal Adaptive Learning. Sensors (Basel). 2022; 22(10). PMC: 9146345. DOI: 10.3390/s22103776. View

18.

Sanaiha Y, Rudasill S, Sareh S, Mardock A, Khoury H, Ziaeian B . Impact of hospital safety-net status on failure to rescue after major cardiac surgery. Surgery. 2019; 166(5):778-784. PMC: 7700062. DOI: 10.1016/j.surg.2019.05.034. View

19.

Cuzick J . A Wilcoxon-type test for trend. Stat Med. 1985; 4(1):87-90. DOI: 10.1002/sim.4780040112. View

20.

Gandjian M, Williamson C, Xia Y, Maturana C, Chervu N, Verma A . Association of Hospital Safety Net Status With Outcomes and Resource Use for Extracorporeal Membrane Oxygenation in the United States. J Intensive Care Med. 2021; 37(4):535-542. DOI: 10.1177/08850666211007062. View