Improved predictive models for acute kidney injury with IDEA: Intraoperative Data Embedded Analytics.
Acute Kidney Injury
/ etiology
Adult
Aged
Algorithms
Area Under Curve
Cohort Studies
Female
Humans
Intraoperative Period
Machine Learning
Male
Middle Aged
Models, Statistical
Postoperative Complications
/ etiology
Predictive Value of Tests
Prognosis
Retrospective Studies
Risk Factors
Sensitivity and Specificity
Journal
PloS one
ISSN: 1932-6203
Titre abrégé: PLoS One
Pays: United States
ID NLM: 101285081
Informations de publication
Date de publication:
2019
2019
Historique:
received:
29
05
2018
accepted:
18
03
2019
entrez:
5
4
2019
pubmed:
5
4
2019
medline:
3
1
2020
Statut:
epublish
Résumé
Acute kidney injury (AKI) is a common complication after surgery that is associated with increased morbidity and mortality. The majority of existing perioperative AKI risk prediction models are limited in their generalizability and do not fully utilize intraoperative physiological time-series data. Thus, there is a need for intelligent, accurate, and robust systems to leverage new information as it becomes available to predict the risk of developing postoperative AKI. A retrospective single-center cohort of 2,911 adults who underwent surgery at the University of Florida Health between 2000 and 2010 was utilized for this study. Machine learning and statistical analysis techniques were used to develop perioperative models to predict the risk of developing AKI during the first three days after surgery, first seven days after surgery, and overall (after surgery during the index hospitalization). The improvement in risk prediction was examined by incorporating intraoperative physiological time-series variables. Our proposed model enriched a preoperative model that produced a probabilistic AKI risk score by integrating intraoperative statistical features through a machine learning stacking approach inside a random forest classifier. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, and Net Reclassification Improvement (NRI). The predictive performance of the proposed model is better than the preoperative data only model. The proposed model had an AUC of 0.86 (accuracy of 0.78) for the seven-day AKI outcome, while the preoperative model had an AUC of 0.84 (accuracy of 0.76). Furthermore, by integrating intraoperative features, the algorithm was able to reclassify 40% of the false negative patients from the preoperative model. The NRI for each outcome was AKI at three days (8%), seven days (7%), and overall (4%). Postoperative AKI prediction was improved with high sensitivity and specificity through a machine learning approach that dynamically incorporated intraoperative data.
Sections du résumé
BACKGROUND
Acute kidney injury (AKI) is a common complication after surgery that is associated with increased morbidity and mortality. The majority of existing perioperative AKI risk prediction models are limited in their generalizability and do not fully utilize intraoperative physiological time-series data. Thus, there is a need for intelligent, accurate, and robust systems to leverage new information as it becomes available to predict the risk of developing postoperative AKI.
METHODS
A retrospective single-center cohort of 2,911 adults who underwent surgery at the University of Florida Health between 2000 and 2010 was utilized for this study. Machine learning and statistical analysis techniques were used to develop perioperative models to predict the risk of developing AKI during the first three days after surgery, first seven days after surgery, and overall (after surgery during the index hospitalization). The improvement in risk prediction was examined by incorporating intraoperative physiological time-series variables. Our proposed model enriched a preoperative model that produced a probabilistic AKI risk score by integrating intraoperative statistical features through a machine learning stacking approach inside a random forest classifier. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, and Net Reclassification Improvement (NRI).
RESULTS
The predictive performance of the proposed model is better than the preoperative data only model. The proposed model had an AUC of 0.86 (accuracy of 0.78) for the seven-day AKI outcome, while the preoperative model had an AUC of 0.84 (accuracy of 0.76). Furthermore, by integrating intraoperative features, the algorithm was able to reclassify 40% of the false negative patients from the preoperative model. The NRI for each outcome was AKI at three days (8%), seven days (7%), and overall (4%).
CONCLUSIONS
Postoperative AKI prediction was improved with high sensitivity and specificity through a machine learning approach that dynamically incorporated intraoperative data.
Identifiants
pubmed: 30947282
doi: 10.1371/journal.pone.0214904
pii: PONE-D-18-16132
pmc: PMC6448850
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
e0214904Subventions
Organisme : NIGMS NIH HHS
ID : R01 GM110240
Pays : United States
Organisme : NIGMS NIH HHS
ID : P50 GM111152
Pays : United States
Organisme : NCATS NIH HHS
ID : UL1 TR000064
Pays : United States
Déclaration de conflit d'intérêts
The authors have declared that no competing interests exist.
Références
J Thorac Cardiovasc Surg. 2014 Jun;147(6):1875-83, 1883.e1
pubmed: 23993316
JAMA. 2009 Sep 16;302(11):1179-85
pubmed: 19755696
Nephrol Dial Transplant. 2012 Feb;27(2):589-94
pubmed: 21712489
Nat Rev Nephrol. 2017 Dec 13;14(1):8-10
pubmed: 29234162
Nephrol Dial Transplant. 2012 Jan;27(1):153-60
pubmed: 21677302
J Chronic Dis. 1987;40(5):373-83
pubmed: 3558716
Anesthesiology. 2015 Sep;123(3):515-23
pubmed: 26181335
Ann Intern Med. 2015 Jan 6;162(1):55-63
pubmed: 25560714
J Vasc Surg. 2006 Mar;43(3):460-466; discussion 466
pubmed: 16520155
Crit Care Med. 2011 Jun;39(6):1493-9
pubmed: 21336114
Crit Care. 2004 Aug;8(4):R204-12
pubmed: 15312219
Ann Surg. 2015 Jun;261(6):1207-14
pubmed: 24887982
Ann Surg. 2019 Apr;269(4):652-662
pubmed: 29489489
Crit Care Med. 2013 Nov;41(11):2570-83
pubmed: 23928835
Intensive Care Med. 2017 Jun;43(6):764-773
pubmed: 28130688
Clin J Am Soc Nephrol. 2016 Nov 7;11(11):1935-1943
pubmed: 27633727
Ann Surg. 2016 Dec;264(6):987-996
pubmed: 26756753
Ann Vasc Surg. 2016 Jan;30:72-81.e1-2
pubmed: 26187703
Ann Surg. 2016 Jun;263(6):1219-1227
pubmed: 26181482
JAMA. 2017 Oct 10;318(14):1346-1357
pubmed: 28973220
Cancer. 1950 Jan;3(1):32-5
pubmed: 15405679
J Clin Anesth. 2017 Aug;40:91-98
pubmed: 28625460
Anesthesiology. 2016 Aug;125(2):438-9
pubmed: 27433767
Anesthesiol Res Pract. 2013;2013:174091
pubmed: 24324489
Crit Care Clin. 2015 Oct;31(4):705-23
pubmed: 26410139
Ann Surg. 2009 May;249(5):851-8
pubmed: 19387314
Surgery. 2016 Aug;160(2):463-72
pubmed: 27238354
Crit Care Clin. 2017 Apr;33(2):379-396
pubmed: 28284301
Ann Thorac Surg. 2012 Jan;93(1):337-47
pubmed: 22186469
JAMA Surg. 2016 May 1;151(5):441-50
pubmed: 26720406
Am J Kidney Dis. 2015 Jun;65(6):860-9
pubmed: 25488106
Am J Med. 2002 Oct 15;113(6):456-61
pubmed: 12427493
BMJ Open. 2017 Sep 27;7(9):e016591
pubmed: 28963291
Crit Care. 2014 Nov 20;18(6):606
pubmed: 25673427
J Cardiothorac Vasc Anesth. 2015 Dec;29(6):1588-95
pubmed: 26159745
Curr Opin Anaesthesiol. 2017 Feb;30(1):113-117
pubmed: 27841788
Ann Intern Med. 2009 May 5;150(9):604-12
pubmed: 19414839
Nephrol Dial Transplant. 2013 Nov;28(11):2787-99
pubmed: 24081864
Anesthesiology. 2017 Jan;126(1):16-27
pubmed: 27775997
BMJ. 2010 Jul 05;341:c3365
pubmed: 20603317
Sci Transl Med. 2010 Sep 8;2(48):48ra65
pubmed: 20826840
Stat Med. 2011 Jan 15;30(1):11-21
pubmed: 21204120
PLoS One. 2016 May 27;11(5):e0155705
pubmed: 27232332
J Vasc Surg. 2018 Sep;68(3):916-928
pubmed: 30146038
Kidney Int Rep. 2016 Dec 09;2(3):342-349
pubmed: 29142963