Amanda C Filiberto1, Tezcan Ozrazgat-Baslanti2, Tyler J Loftus3, Ying-Chih Peng4, Shounak Datta2, Philip Efron5, Gilbert R Upchurch1, Azra Bihorac2, Michol A Cooper6. 1. Department of Surgery, University of Florida, Gainesville, FL. 2. Department of Medicine, University of Florida, Gainesville, FL; Precision and Intelligent Systems in Medicine (PrismaP), University of Florida, Gainesville FL. 3. Department of Surgery, University of Florida, Gainesville, FL; Precision and Intelligent Systems in Medicine (PrismaP), University of Florida, Gainesville FL. 4. Precision and Intelligent Systems in Medicine (PrismaP), University of Florida, Gainesville FL. 5. Department of Surgery, University of Florida, Gainesville, FL; Department of Anesthesia, University of Florida, Gainesville, FL. 6. Department of Surgery, University of Florida, Gainesville, FL. Electronic address: Michol.cooper@surgery.ufl.edu.
Abstract
BACKGROUND: Postoperative acute kidney injury is common after major vascular surgery and is associated with increased morbidity, mortality, and cost. High-performance risk stratification using a machine learning model can inform strategies that mitigate harm and optimize resource use. It is hypothesized that incorporating intraoperative data would improve machine learning model accuracy, discrimination, and precision in predicting acute kidney injury among patients undergoing major vascular surgery. METHODS: A single-center retrospective cohort of 1,531 adult patients who underwent nonemergency major vascular surgery, including open aortic, endovascular aortic, and lower extremity bypass procedures, was evaluated. The validated, automated MySurgeryRisk analytics platform used electronic health record data to forecast patient-level probabilistic risk scores for postoperative acute kidney injury using random forest models with preoperative data alone and perioperative data (preoperative plus intraoperative). The MySurgeryRisk predictions were compared with each other as well as with the American Society of Anesthesiologists physical status classification. RESULTS: Machine learning models using perioperative data had greater accuracy, discrimination, and precision than models using either preoperative data alone or the American Society of Anesthesiologists physical status classification (accuracy: 0.70 vs 0.64 vs 0.62, area under the receiver operating characteristics curve: 0.77 vs 0.68 vs 0.61, area under the precision-recall curve: 0.70 vs 0.58 vs 0.48). CONCLUSION: In predicting acute kidney injury after major vascular surgery, machine learning approaches that incorporate dynamic intraoperative data had greater accuracy, discrimination, and precision than models using either preoperative data alone or the American Society of Anesthesiologists physical status classification. Machine learning methods have the potential for real-time identification of high-risk patients who may benefit from personalized risk-reduction strategies.
BACKGROUND: Postoperative acute kidney injury is common after major vascular surgery and is associated with increased morbidity, mortality, and cost. High-performance risk stratification using a machine learning model can inform strategies that mitigate harm and optimize resource use. It is hypothesized that incorporating intraoperative data would improve machine learning model accuracy, discrimination, and precision in predicting acute kidney injury among patients undergoing major vascular surgery. METHODS: A single-center retrospective cohort of 1,531 adult patients who underwent nonemergency major vascular surgery, including open aortic, endovascular aortic, and lower extremity bypass procedures, was evaluated. The validated, automated MySurgeryRisk analytics platform used electronic health record data to forecast patient-level probabilistic risk scores for postoperative acute kidney injury using random forest models with preoperative data alone and perioperative data (preoperative plus intraoperative). The MySurgeryRisk predictions were compared with each other as well as with the American Society of Anesthesiologists physical status classification. RESULTS: Machine learning models using perioperative data had greater accuracy, discrimination, and precision than models using either preoperative data alone or the American Society of Anesthesiologists physical status classification (accuracy: 0.70 vs 0.64 vs 0.62, area under the receiver operating characteristics curve: 0.77 vs 0.68 vs 0.61, area under the precision-recall curve: 0.70 vs 0.58 vs 0.48). CONCLUSION: In predicting acute kidney injury after major vascular surgery, machine learning approaches that incorporate dynamic intraoperative data had greater accuracy, discrimination, and precision than models using either preoperative data alone or the American Society of Anesthesiologists physical status classification. Machine learning methods have the potential for real-time identification of high-risk patients who may benefit from personalized risk-reduction strategies.
Authors: Jilles B Bijker; Wilton A van Klei; Yvonne Vergouwe; Douglas J Eleveld; Leo van Wolfswinkel; Karel G M Moons; Cor J Kalkman Journal: Anesthesiology Date: 2009-12 Impact factor: 7.892
Authors: Ira L Leeds; Andrew J Rosenblum; Paul E Wise; Anthony C Watkins; Matthew I Goldblatt; Elliott R Haut; Jonathan E Efron; Fabian M Johnston Journal: Surgery Date: 2018-08-24 Impact factor: 3.982
Authors: Karl Y Bilimoria; Yaoming Liu; Jennifer L Paruch; Lynn Zhou; Thomas E Kmiecik; Clifford Y Ko; Mark E Cohen Journal: J Am Coll Surg Date: 2013-09-18 Impact factor: 6.113
Authors: Matthew Huber; Tezcan Ozrazgat-Baslanti; Paul Thottakkara; Philip A Efron; Robert Feezor; Charles Hobson; Azra Bihorac Journal: Ann Vasc Surg Date: 2015-07-14 Impact factor: 1.466
Authors: Ward H van der Ven; Denise P Veelo; Marije Wijnberge; Björn J P van der Ster; Alexander P J Vlaar; Bart F Geerts Journal: Surgery Date: 2020-12-11 Impact factor: 3.982