Machine-learning predictions for acute kidney injuries after coronary artery bypass grafting: a real-life muticenter retrospective cohort study.
Acute kidney injuries
Coronary artery bypass grafting
Machine learning
Prediction model
Journal
BMC medical informatics and decision making
ISSN: 1472-6947
Titre abrégé: BMC Med Inform Decis Mak
Pays: England
ID NLM: 101088682
Informations de publication
Date de publication:
23 Nov 2023
23 Nov 2023
Historique:
received:
24
05
2023
accepted:
16
11
2023
medline:
27
11
2023
pubmed:
24
11
2023
entrez:
23
11
2023
Statut:
epublish
Résumé
Acute kidney injury (AKI) after coronary artery bypass grafting (CABG) surgery is associated with poor outcomes. The objective of this study was to apply a new machine learning (ML) method to establish prediction models of AKI after CABG. A total of 2,780 patients from two medical centers in East China who underwent primary isolated CABG were enrolled. The dataset was randomly divided for model training (80%) and model testing (20%). Four ML models based on LightGBM, Support vector machine (SVM), Softmax and random forest (RF) algorithms respectively were established in Python. A total of 2,051 patients from two other medical centers were assigned to an external validation group to verify the performances of the ML prediction models. The models were evaluated using the area under the receiver operating characteristics curve (AUC), Hosmer-Lemeshow goodness-of-fit statistic, Bland-Altman plots, and decision curve analysis. The outcome of the LightGBM model was interpreted using SHapley Additive exPlanations (SHAP). The incidence of postoperative AKI in the modeling group was 13.4%. Similarly, the incidence of postoperative AKI of the two medical centers in the external validation group was 8.2% and 13.6% respectively. LightGBM performed the best in predicting, with an AUC of 0.8027 in internal validation group and 0.8798 and 0.7801 in the external validation group. The SHAP revealed the top 20 predictors of postoperative AKI ranked according to the importance, and the top three features on prediction were the serum creatinine in the first 24 h after operation, the last preoperative Scr level, and body surface area. This study provides a LightGBM predictive model that can make accurate predictions for AKI after CABG surgery. The LightGBM model shows good predictive ability in both internal and external validation. It can help cardiac surgeons identify high-risk patients who may experience AKI after CABG surgery.
Sections du résumé
BACKGROUND
BACKGROUND
Acute kidney injury (AKI) after coronary artery bypass grafting (CABG) surgery is associated with poor outcomes. The objective of this study was to apply a new machine learning (ML) method to establish prediction models of AKI after CABG.
METHODS
METHODS
A total of 2,780 patients from two medical centers in East China who underwent primary isolated CABG were enrolled. The dataset was randomly divided for model training (80%) and model testing (20%). Four ML models based on LightGBM, Support vector machine (SVM), Softmax and random forest (RF) algorithms respectively were established in Python. A total of 2,051 patients from two other medical centers were assigned to an external validation group to verify the performances of the ML prediction models. The models were evaluated using the area under the receiver operating characteristics curve (AUC), Hosmer-Lemeshow goodness-of-fit statistic, Bland-Altman plots, and decision curve analysis. The outcome of the LightGBM model was interpreted using SHapley Additive exPlanations (SHAP).
RESULTS
RESULTS
The incidence of postoperative AKI in the modeling group was 13.4%. Similarly, the incidence of postoperative AKI of the two medical centers in the external validation group was 8.2% and 13.6% respectively. LightGBM performed the best in predicting, with an AUC of 0.8027 in internal validation group and 0.8798 and 0.7801 in the external validation group. The SHAP revealed the top 20 predictors of postoperative AKI ranked according to the importance, and the top three features on prediction were the serum creatinine in the first 24 h after operation, the last preoperative Scr level, and body surface area.
CONCLUSION
CONCLUSIONS
This study provides a LightGBM predictive model that can make accurate predictions for AKI after CABG surgery. The LightGBM model shows good predictive ability in both internal and external validation. It can help cardiac surgeons identify high-risk patients who may experience AKI after CABG surgery.
Identifiants
pubmed: 37996844
doi: 10.1186/s12911-023-02376-0
pii: 10.1186/s12911-023-02376-0
pmc: PMC10668365
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
270Informations de copyright
© 2023. The Author(s).
Références
Lancet. 1986 Feb 8;1(8476):307-10
pubmed: 2868172
Braz J Cardiovasc Surg. 2017 Nov-Dec;32(6):481-486
pubmed: 29267610
N Engl J Med. 2017 Jan 5;376(1):11-20
pubmed: 27959707
Interact Cardiovasc Thorac Surg. 2013 Oct;17(4):638-43
pubmed: 23793709
J Extra Corpor Technol. 2015 Mar;47(1):16-28
pubmed: 26390675
Interact Cardiovasc Thorac Surg. 2015 Oct;21(4):465-9
pubmed: 26160965
Heart Surg Forum. 2018 Jan 31;21(1):E036-E039
pubmed: 29485962
Front Cardiovasc Med. 2022 Sep 15;9:951881
pubmed: 36186995
Am J Cardiol. 2017 Jan 1;119(1):91-99
pubmed: 27788934
Am Heart J. 2016 Jan;171(1):40-7
pubmed: 26699599
Semin Thorac Cardiovasc Surg. 2021 Autumn;33(3):735-745
pubmed: 32979479
Surg Clin North Am. 2023 Apr;103(2):317-333
pubmed: 36948721
Sci Rep. 2017 Dec 1;7(1):16806
pubmed: 29196738
J Med Internet Res. 2023 Jan 5;25:e41142
pubmed: 36603200
Stat Med. 2017 Dec 10;36(28):4455-4467
pubmed: 27426413
Ther Clin Risk Manag. 2020 May 21;16:451-459
pubmed: 32547042
J Med Internet Res. 2022 Aug 9;24(8):e38082
pubmed: 35943767
Sci Rep. 2022 Sep 14;12(1):15438
pubmed: 36104386
Heart Lung Circ. 2021 Feb;30(2):288-295
pubmed: 32690359
BMC Cardiovasc Disord. 2020 Oct 28;20(1):463
pubmed: 33115418
PeerJ. 2018 Feb 23;6:e4413
pubmed: 29492345
Ann Intern Med. 1999 Mar 16;130(6):461-70
pubmed: 10075613
Medicine (Baltimore). 2017 Nov;96(45):e8395
pubmed: 29137024
Braz J Cardiovasc Surg. 2018 Jul-Aug;33(4):323-329
pubmed: 30184028
J Med Internet Res. 2016 Dec 16;18(12):e323
pubmed: 27986644
J Cardiothorac Surg. 2021 Oct 26;16(1):315
pubmed: 34702324
J Transl Med. 2020 Apr 6;18(1):157
pubmed: 32252780
Int J Gen Med. 2021 Nov 19;14:8509-8519
pubmed: 34824547
Lancet Digit Health. 2021 Mar;3(3):e158-e165
pubmed: 33549512
J Cardiothorac Vasc Anesth. 2021 Jul;35(7):2052-2062
pubmed: 33414071
Int Urol Nephrol. 2019 Sep;51(9):1605-1611
pubmed: 31161519
J Crit Care. 2016 Feb;31(1):249-54
pubmed: 26700609
Saudi J Anaesth. 2019 Apr;13(Suppl 1):S31-S34
pubmed: 30930717
Stat Med. 2017 Dec 10;36(28):4482-4490
pubmed: 27699818
Basic Res Cardiol. 2023 Mar 20;118(1):10
pubmed: 36939941
Cardiol Clin. 2019 Aug;37(3):297-305
pubmed: 31279423
J Clin Med. 2022 Oct 24;11(21):
pubmed: 36362493
J Cardiothorac Vasc Anesth. 2022 Apr;36(4):962-970
pubmed: 34969562
J Transl Med. 2019 Apr 11;17(1):119
pubmed: 30971285