Machine Learned Cellular Phenotypes in Cardiomyopathy Predict Sudden Death.

Publisher: Lippincott Williams & Wilkins Country of Publication: United States NLM ID: 0047103 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1524-4571 (Electronic) Linking ISSN: 00097330 NLM ISO Abbreviation: Circ Res Subsets: MEDLINE

Publication: Baltimore, MD : Lippincott Williams & Wilkins
Original Publication: Baltimore, Md. Grune & Stratton.

Diagnosis, Computer-Assisted* ; Electrophysiologic Techniques, Cardiac* ; Neural Networks, Computer* ; Signal Processing, Computer-Assisted* ; Support Vector Machine*; Cardiomyopathies/*diagnosis ; Death, Sudden, Cardiac/*etiology ; Tachycardia, Ventricular/*diagnosis ; Ventricular Fibrillation/*diagnosis; Action Potentials ; Aged ; Aged, 80 and over ; Cardiomyopathies/etiology ; Cardiomyopathies/mortality ; Cardiomyopathies/physiopathology ; Female ; Humans ; Male ; Middle Aged ; Myocardial Infarction/complications ; Myocardial Infarction/mortality ; Myocardial Infarction/physiopathology ; Phenotype ; Predictive Value of Tests ; Prognosis ; Prospective Studies ; Risk Assessment ; Risk Factors ; Tachycardia, Ventricular/etiology ; Tachycardia, Ventricular/mortality ; Tachycardia, Ventricular/physiopathology ; Time Factors ; Ventricular Fibrillation/etiology ; Ventricular Fibrillation/mortality ; Ventricular Fibrillation/physiopathology

Rationale: Susceptibility to VT/VF (ventricular tachycardia/fibrillation) is difficult to predict in patients with ischemic cardiomyopathy either by clinical tools or by attempting to translate cellular mechanisms to the bedside.
Objective: To develop computational phenotypes of patients with ischemic cardiomyopathy, by training then interpreting machine learning of ventricular monophasic action potentials (MAPs) to reveal phenotypes that predict long-term outcomes.
Methods and Results: We recorded 5706 ventricular MAPs in 42 patients with coronary artery disease and left ventricular ejection fraction ≤40% during steady-state pacing. Patients were randomly allocated to independent training and testing cohorts in a 70:30 ratio, repeated K=10-fold. Support vector machines and convolutional neural networks were trained to 2 end points: (1) sustained VT/VF or (2) mortality at 3 years. Support vector machines provided superior classification. For patient-level predictions, we computed personalized MAP scores as the proportion of MAP beats predicting each end point. Patient-level predictions in independent test cohorts yielded c-statistics of 0.90 for sustained VT/VF (95% CI, 0.76-1.00) and 0.91 for mortality (95% CI, 0.83-1.00) and were the most significant multivariate predictors. Interpreting trained support vector machine revealed MAP morphologies that, using in silico modeling, revealed higher L-type calcium current or sodium-calcium exchanger as predominant phenotypes for VT/VF.
Conclusions: Machine learning of action potential recordings in patients revealed novel phenotypes for long-term outcomes in ischemic cardiomyopathy. Such computational phenotypes provide an approach which may reveal cellular mechanisms for clinical outcomes and could be applied to other conditions.

Comment in: Circ Res. 2021 Jan 22;128(2):185-187. (PMID: 33476206)

Circ Res. 2020 Apr 10;126(8):947-964. (PMID: 32091972)
J Physiol. 2017 Nov 1;595(21):6599-6612. (PMID: 28815597)
Heart Rhythm. 2015 Feb;12(2):360-6. (PMID: 25446153)
Circulation. 2014 Jan 28;129(4):516-26. (PMID: 24470473)
Nat Med. 2019 Jan;25(1):65-69. (PMID: 30617320)
J Pharmacol Toxicol Methods. 2014 Nov-Dec;70(3):246-54. (PMID: 25087753)
J Am Coll Cardiol. 2007 Dec 18;50(25):2385-92. (PMID: 18154963)
Sci Rep. 2016 Aug 26;6:32390. (PMID: 27561321)
Circ Res. 2019 Mar 15;124(6):904-919. (PMID: 30661465)
J Am Coll Cardiol. 2012 Sep 18;60(12):1103-10. (PMID: 22883636)
Front Physiol. 2018 Oct 16;9:1453. (PMID: 30374311)
Circulation. 2018 Sep 25;138(13):e272-e391. (PMID: 29084731)
Front Physiol. 2018 Mar 13;9:213. (PMID: 29593570)
Cardiovasc Res. 2012 Aug 1;95(3):308-16. (PMID: 22660482)
Am J Physiol Heart Circ Physiol. 2019 Oct 1;317(4):H667-H673. (PMID: 31347917)
Nat Med. 2019 Jan;25(1):70-74. (PMID: 30617318)
J Cardiovasc Electrophysiol. 2017 Nov;28(11):1345-1351. (PMID: 28744959)
J Am Coll Cardiol. 2019 Jan 8;73(1):70-88. (PMID: 30621954)
Circulation. 2015 Jun 2;131(22):1950-9. (PMID: 25810336)
JACC Clin Electrophysiol. 2015 Dec 1;1(6):542-550. (PMID: 26949741)
Circ Arrhythm Electrophysiol. 2019 Jul;12(7):e007316. (PMID: 31216884)
J Am Coll Cardiol. 2011 Sep 20;58(13):1309-24. (PMID: 21920259)
Heart Rhythm. 2010 Apr;7(4):549-58. (PMID: 20153265)
PLoS One. 2013;8(2):e56359. (PMID: 23437117)
J Mol Cell Cardiol. 1975 Jan;7(1):1-15. (PMID: 1121029)
Eur Heart J. 2017 Oct 21;38(40):3017-3025. (PMID: 28662567)
J Electrocardiol. 2015 Jul-Aug;48(4):551-7. (PMID: 25912974)
Heart Lung Circ. 2019 Jan;28(1):22-30. (PMID: 30389366)
Heart Rhythm. 2016 Sep;13(9):1922-31. (PMID: 27215536)
Science. 2019 May 31;364(6443):859-865. (PMID: 31147514)
Cardiovasc Res. 2019 Sep 1;115(11):1659-1671. (PMID: 30753358)
Circulation. 2004 Sep 14;110(11):1343-50. (PMID: 15353505)
N Engl J Med. 2017 Dec 14;377(24):2325-2336. (PMID: 29236642)
Heart Rhythm. 2007 Aug;4(8):1087-9. (PMID: 17675086)
Circulation. 2018 Mar 20;137(12):e67-e492. (PMID: 29386200)
Pacing Clin Electrophysiol. 2005 Mar;28(3):219-27. (PMID: 15733182)
PLoS Comput Biol. 2011 May;7(5):e1002061. (PMID: 21637795)
Eur Heart J. 1986 Jan;7(1):34-41. (PMID: 2420597)
J Am Coll Cardiol. 2008 Nov 25;52(22):1782-92. (PMID: 19022157)
Circ Res. 2009 Jan 2;104(1):19-31. (PMID: 19118284)
Annu Rev Med. 2018 Jan 29;69:147-164. (PMID: 29414264)
Circ Res. 2019 Aug 30;125(6):653-658. (PMID: 31465267)

R01 HL122384 United States HL NHLBI NIH HHS; R21 HL145500 United States HL NHLBI NIH HHS; R01 HL083359 United States HL NHLBI NIH HHS; R01 HL149134 United States HL NHLBI NIH HHS; K23 HL145017 United States HL NHLBI NIH HHS; F32 HL144101 United States HL NHLBI NIH HHS

Keywords: artificial intelligence; coronary disease; death, sudden, cardiac; heart failure; ion channels; systems biology

Date Created: 20201110 Date Completed: 20210726 Latest Revision: 20220123

20231215

PMC7855939

10.1161/CIRCRESAHA.120.317345

33167779