Application of machine learning to the prediction of postoperative sepsis after appendectomy.
Journal
Surgery
ISSN: 1532-7361
Titre abrégé: Surgery
Pays: United States
ID NLM: 0417347
Informations de publication
Date de publication:
03 2021
03 2021
Historique:
received:
04
05
2020
revised:
22
07
2020
accepted:
25
07
2020
pubmed:
22
9
2020
medline:
27
4
2021
entrez:
21
9
2020
Statut:
ppublish
Résumé
We applied various machine learning algorithms to a large national dataset to model the risk of postoperative sepsis after appendectomy to evaluate utility of such methods and identify factors associated with postoperative sepsis in these patients. The National Surgery Quality Improvement Program database was used to identify patients undergoing appendectomy between 2005 and 2017. Logistic regression, support vector machines, random forest decision trees, and extreme gradient boosting machines were used to model the occurrence of postoperative sepsis. In the study, 223,214 appendectomies were identified; 2,143 (0.96%) were indicated as having postoperative sepsis. Logistic regression (area under the curve 0.70; 95% confidence interval, 0.68-0.73), random forest decision trees (area under the curve 0.70; 95% confidence interval, 0.68-0.73), and extreme gradient boosting (area under the curve 0.70; 95% confidence interval, 0.68-0.73) afforded similar performance, while support vector machines (area under the curve 0.51; 95% confidence interval, 0.50-0.52) had worse performance. Variable importance analyses identified preoperative congestive heart failure, transfusion, and acute renal failure as predictors of postoperative sepsis. Machine learning methods can be used to predict the development of sepsis after appendectomy with moderate accuracy. Such predictive modeling has potential to ultimately allow for preoperative recognition of patients at risk for developing postoperative sepsis after appendectomy thus facilitating early intervention and reducing morbidity.
Sections du résumé
BACKGROUND
We applied various machine learning algorithms to a large national dataset to model the risk of postoperative sepsis after appendectomy to evaluate utility of such methods and identify factors associated with postoperative sepsis in these patients.
METHODS
The National Surgery Quality Improvement Program database was used to identify patients undergoing appendectomy between 2005 and 2017. Logistic regression, support vector machines, random forest decision trees, and extreme gradient boosting machines were used to model the occurrence of postoperative sepsis.
RESULTS
In the study, 223,214 appendectomies were identified; 2,143 (0.96%) were indicated as having postoperative sepsis. Logistic regression (area under the curve 0.70; 95% confidence interval, 0.68-0.73), random forest decision trees (area under the curve 0.70; 95% confidence interval, 0.68-0.73), and extreme gradient boosting (area under the curve 0.70; 95% confidence interval, 0.68-0.73) afforded similar performance, while support vector machines (area under the curve 0.51; 95% confidence interval, 0.50-0.52) had worse performance. Variable importance analyses identified preoperative congestive heart failure, transfusion, and acute renal failure as predictors of postoperative sepsis.
CONCLUSION
Machine learning methods can be used to predict the development of sepsis after appendectomy with moderate accuracy. Such predictive modeling has potential to ultimately allow for preoperative recognition of patients at risk for developing postoperative sepsis after appendectomy thus facilitating early intervention and reducing morbidity.
Identifiants
pubmed: 32951903
pii: S0039-6060(20)30518-3
doi: 10.1016/j.surg.2020.07.045
pmc: PMC7927311
mid: NIHMS1673741
pii:
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Langues
eng
Sous-ensembles de citation
IM
Pagination
671-677Subventions
Organisme : NIAAA NIH HHS
ID : T32 AA013527
Pays : United States
Organisme : NIGMS NIH HHS
ID : T32 GM008750
Pays : United States
Informations de copyright
Copyright © 2020 Elsevier Inc. All rights reserved.
Références
Curr Opin Crit Care. 2011 Aug;17(4):396-401
pubmed: 21677580
PLoS One. 2018 Jan 19;13(1):e0191176
pubmed: 29351327
N Engl J Med. 2016 Sep 29;375(13):1216-9
pubmed: 27682033
BMC Med Inform Decis Mak. 2018 Dec 29;18(1):139
pubmed: 30594159
Health Aff (Millwood). 2014 Jul;33(7):1163-70
pubmed: 25006142
Acad Emerg Med. 2016 Mar;23(3):269-78
pubmed: 26679719
Crit Care Med. 2012 Mar;40(3):754-61
pubmed: 21963582
JAMA Netw Open. 2018 Aug 3;1(4):e181018
pubmed: 30646095
Arch Surg. 2010 Jul;145(7):695-700
pubmed: 20644134
NPJ Digit Med. 2018 May 8;1:18
pubmed: 31304302
Surg Infect (Larchmt). 2019 Dec;20(8):601-606
pubmed: 31009326
N Engl J Med. 2010 Feb 4;362(5):382-5
pubmed: 20042745
Neurosurgery. 2019 Jul 1;85(1):E83-E91
pubmed: 30476188
J Am Med Inform Assoc. 2014 Jul-Aug;21(4):578-82
pubmed: 24821743
Int J Methods Psychiatr Res. 2011 Mar;20(1):40-9
pubmed: 21499542
Br J Surg. 2014 Jan;101(1):e9-22
pubmed: 24272924
Ann Surg. 2010 Dec;252(6):895-900
pubmed: 21107099
JAMA. 2017 Aug 8;318(6):517-518
pubmed: 28727867
Ann Surg. 2003 Jul;238(1):59-66
pubmed: 12832966
J Thorac Cardiovasc Surg. 2004 Jul;128(1):138-46
pubmed: 15224033