Machine learning to predict abnormal myocardial perfusion from pre-test features.
Artificial intelligence
CAD
Image analysis
Machine learning
Myocardial perfusion imaging
PET
SPECT
Journal
Journal of nuclear cardiology : official publication of the American Society of Nuclear Cardiology
ISSN: 1532-6551
Titre abrégé: J Nucl Cardiol
Pays: United States
ID NLM: 9423534
Informations de publication
Date de publication:
Oct 2022
Oct 2022
Historique:
received:
29
03
2022
accepted:
22
04
2022
revised:
22
04
2022
pubmed:
8
6
2022
medline:
15
10
2022
entrez:
7
6
2022
Statut:
ppublish
Résumé
Accurately predicting which patients will have abnormal perfusion on MPI based on pre-test clinical information may help physicians make test selection decisions. We developed and validated a machine learning (ML) model for predicting abnormal perfusion using pre-test features. We included consecutive patients who underwent SPECT MPI, with 20,418 patients from a multi-center (5 sites) international registry in the training population and 9019 patients (from 2 separate sites) in the external testing population. The ML (extreme gradient boosting) model utilized 30 pre-test features to predict the presence of abnormal myocardial perfusion by expert visual interpretation. In external testing, the ML model had higher prediction performance for abnormal perfusion (area under receiver-operating characteristic curve [AUC] 0.762, 95% CI 0.750-0.774) compared to the clinical CAD consortium (AUC 0.689) basic CAD consortium (AUC 0.657), and updated Diamond-Forrester models (AUC 0.658, p < 0.001 for all). Calibration (validation of the continuous risk prediction) was superior for the ML model (Brier score 0.149) compared to the other models (Brier score 0.165 to 0.198, all p < 0.001). ML can predict abnormal myocardial perfusion using readily available pre-test information. This model could be used to help guide physician decisions regarding non-invasive test selection.
Sections du résumé
BACKGROUND
Accurately predicting which patients will have abnormal perfusion on MPI based on pre-test clinical information may help physicians make test selection decisions. We developed and validated a machine learning (ML) model for predicting abnormal perfusion using pre-test features.
METHODS
We included consecutive patients who underwent SPECT MPI, with 20,418 patients from a multi-center (5 sites) international registry in the training population and 9019 patients (from 2 separate sites) in the external testing population. The ML (extreme gradient boosting) model utilized 30 pre-test features to predict the presence of abnormal myocardial perfusion by expert visual interpretation.
RESULTS
In external testing, the ML model had higher prediction performance for abnormal perfusion (area under receiver-operating characteristic curve [AUC] 0.762, 95% CI 0.750-0.774) compared to the clinical CAD consortium (AUC 0.689) basic CAD consortium (AUC 0.657), and updated Diamond-Forrester models (AUC 0.658, p < 0.001 for all). Calibration (validation of the continuous risk prediction) was superior for the ML model (Brier score 0.149) compared to the other models (Brier score 0.165 to 0.198, all p < 0.001).
CONCLUSION
ML can predict abnormal myocardial perfusion using readily available pre-test information. This model could be used to help guide physician decisions regarding non-invasive test selection.
Identifiants
pubmed: 35672567
doi: 10.1007/s12350-022-03012-6
pii: 10.1007/s12350-022-03012-6
pmc: PMC9588501
mid: NIHMS1807533
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Langues
eng
Sous-ensembles de citation
IM
Pagination
2393-2403Subventions
Organisme : NHLBI NIH HHS
ID : R01 HL089765
Pays : United States
Organisme : NCATS NIH HHS
ID : UL1 TR001863
Pays : United States
Informations de copyright
© 2022. The Author(s) under exclusive licence to American Society of Nuclear Cardiology.
Références
Comput Biol Med. 2022 Jun;145:105449
pubmed: 35381453
J Nucl Cardiol. 2020 Jun;27(3):1010-1021
pubmed: 29923104
N Engl J Med. 2020 Apr 9;382(15):1408-1419
pubmed: 32227753
J Nucl Cardiol. 2010 Dec;17(6):1050-7
pubmed: 20963537
J Nucl Cardiol. 2005 Jan-Feb;12(1):66-77
pubmed: 15682367
J Am Coll Cardiol. 2021 Nov 30;78(22):e187-e285
pubmed: 34756653
JACC Cardiovasc Imaging. 2018 Jul;11(7):1000-1009
pubmed: 29055639
J Nucl Cardiol. 2022 Apr 13;:
pubmed: 35419699
Eur J Nucl Med Mol Imaging. 2013 May;40(5):649-51
pubmed: 23407991
J Nucl Cardiol. 2018 Apr;25(2):635-651
pubmed: 27444500
BMC Med. 2019 Dec 16;17(1):230
pubmed: 31842878
N Engl J Med. 1979 Jun 14;300(24):1350-8
pubmed: 440357
JACC Cardiovasc Imaging. 2020 Oct;13(10):2193-2202
pubmed: 32563652
Epidemiology. 2010 Jan;21(1):128-38
pubmed: 20010215
Int J Cardiol. 2020 Aug 15;313:9-15
pubmed: 32349938
BMJ. 2012 Jun 12;344:e3485
pubmed: 22692650
J Nucl Cardiol. 2018 Feb;25(1):269-297
pubmed: 29243073
JACC Cardiovasc Imaging. 2021 Mar;14(3):644-653
pubmed: 32828784
Eur Heart J. 2011 Jun;32(11):1316-30
pubmed: 21367834
JACC Cardiovasc Imaging. 2020 Mar;13(3):774-785
pubmed: 31202740
JACC Cardiovasc Imaging. 2017 Jul;10(7):787-794
pubmed: 28330657
Cardiovasc Res. 2022 Jul 20;118(9):2152-2164
pubmed: 34259870
JACC Cardiovasc Imaging. 2022 Feb;15(2):271-280
pubmed: 34656462
J Am Coll Cardiol. 2021 Nov 30;78(22):2218-2261
pubmed: 34756652
J Am Coll Cardiol. 2011 Apr 12;57(15):1622-32
pubmed: 21439754
Heart. 2018 Jul;104(13):1118-1124
pubmed: 29331986
Eur Heart J Cardiovasc Imaging. 2021 May 10;22(6):705-714
pubmed: 32533137
J Nucl Cardiol. 2022 Oct;29(5):2295-2307
pubmed: 34228341
J Nucl Med. 2015 Apr;56(4):545-51
pubmed: 25745089
Circulation. 2019 Dec 10;140(24):1971-1980
pubmed: 31707827
Bioinformatics. 2005 Aug 1;21(15):3301-7
pubmed: 15905277
Eur Heart J. 2020 Jan 14;41(3):407-477
pubmed: 31504439
J Nucl Cardiol. 2021 Aug;28(4):1676-1687
pubmed: 31823328
Circulation. 2002 Jan 29;105(4):539-42
pubmed: 11815441
J Nucl Cardiol. 2020 Feb;27(1):147-155
pubmed: 29790017
J Am Coll Cardiol. 2013 Mar 12;61(10):1054-65
pubmed: 23473411
J Nucl Cardiol. 2021 Oct 4;:
pubmed: 34608604
Eur Heart J. 2017 Apr 1;38(13):991-998
pubmed: 27141095
JAMA Cardiol. 2021 Dec 1;6(12):1465
pubmed: 34586335
Biometrics. 1988 Sep;44(3):837-45
pubmed: 3203132