CoChat
Sign Up
PreprintCited by 38ECG Monitoring and AnalysisCardiomyopathy and Myosin StudiesCardiac electrophysiology and arrhythmias

Identification of hypertrophic cardiomyopathy on electrocardiographic images with deep learning

Open access
Sangha, Veer, Dhingra, Lovedeep Singh, Aminorroaya, Arya, Croon, Philip M, Sikand, Nikhil V, Sen, Sounok, Martinez, Matthew W, Maron, Martin S +4 more·2023·Nature Cardiovascular Research
DOI 10.1101/2023.12.23.23300490PMC PMC10793540

Preprint — published as 10.1038/s44161-025-00685-3 · Paper page

Open this paper in CoChat — free accountOpen PDF

Abstract

Hypertrophic cardiomyopathy (HCM) is frequently underdiagnosed. While deep learning (DL) models using raw electrocardiographic (ECG) voltage data can enhance detection, their use at the point-of-care is limited. Here we report the development and validation of a DL model that detects HCM from images of 12-lead ECGs across layouts. The model was developed using 124,553 ECGs from 66,987 individuals at the Yale New Haven Hospital (YNHH), with HCM features determined by concurrent imaging (cardiac magnetic resonance [CMR] or echocardiography). External validation included ECG images from MIMIC-IV, Amsterdam University Medical Center (AUMC), and UK Biobank, where HCM was defined by CMR (YNHH, MIMIC-IV, AUMC) and diagnosis codes (UK Biobank). The model demonstrated robust performance across image formats and sites (AUROCs: 0.95 internal testing; 0.94 MIMIC-IV; 0.92 AUMC; 0.91 UK Biobank). Discriminative features localized to anterior/lateral leads (V4-V5) regardless of layout. This approach enables scalable, image-based screening for HCM across clinical settings.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1101/2023.12.23.23300490

Cites

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The low prevalence of individual structural heart diseases limits the potential implementation of models for broad screening because of the low PPVs of the proposed models.

    “While these models provide a strong foundation for the role of ECG-based detection of SHDs, the low prevalence of these individual diseases, their potential implementation for broad screening is limited by the low PPVs of the proposed models.10-12,16,32-34”

    From Ensemble Deep Learning Algorithm for Structural Heart Disease Screening Using Electrocardiographic Images (2025) · cites this paper as [16]

    The Discussion says community-based screening with this HCM model can produce high false-positive rates and that low prevalence limits screening yield. The Results report only 29 HCM cases among 45,389 UK Biobank ECGs. This supports the concern for HCM, but the paper does not establish low PPVs or screening limitations across multiple structural heart diseases and their proposed models. (LLM, conf=0.85)

    Experimental — accuracy not yet measured

  • Previously, applications of artificial intelligence for electrocardiograms have shown potential to detect signatures of structural heart diseases.

    “Previously applications of artificial intelligence for electrocardiograms (AI-ECG) have shown potential to detect signatures of SHDs.10-18”

    From Ensemble Deep Learning Algorithm for Structural Heart Disease Screening Using Electrocardiographic Images (2025) · cites this paper as [16]

  • Prior studies have reported the use of deep learning on 12-lead ECGs to detect individual structural cardiovascular conditions, including LVSD, hypertrophic cardiomyopathy, cardiac amyloidosis, and aortic stenosis.

    “Prior studies have reported the use of deep learning on 12-lead ECGs to detect individual structural cardiovascular conditions, including LVSD,10,11,13 hypertrophic cardiomyopathy,16,32,33 cardiac amyloidosis,34,35 aortic stenosis,12 among others.15,35-37”

    From Ensemble Deep Learning Algorithm for Structural Heart Disease Screening Using Electrocardiographic Images (2025) · cites this paper as [16]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

No references could be resolved yet.

Research works printed but not linked: 43

  1. [1]Maron B. J. Hypertrophic cardiomyopathy. JAMA 287, 1308– 1320 (2002). 11886323
  2. [2]Maron B. J. & Maron M. S. Hypertrophic cardiomyopathy. Lancet 381, 242– 255 (2013). 22874472
  3. [3]Semsarian C., Ingles J., Maron M. S. & Maron B. J. New perspectives on the prevalence of hypertrophic cardiomyopathy. J. Am. Coll. Cardiol. 65, 1249– 1254 (2015). 25814232
  4. [4]Sabater-Molina M., Pérez-Sánchez I., Hernández del Rincón J. P. & Gimeno J. R. Genetics of hypertrophic cardiomyopathy: A review of current state. Clin. Genet. 93, 3– 14 (2018). 28369730
  5. [5]Geske J. B., Ommen S. R. & Gersh B. J. Hypertrophic cardiomyopathy. JACC Heart Fail. 6, 364– 375 (2018). 29655825
  6. [6]Maron B. J. Diagnosis and evaluation of hypertrophic cardiomyopathy. J. Am. Coll. Cardiol. 79, 372– 389 (2022). 35086660
  7. [8]Goto S. Multinational Federated Learning Approach to Train ECG and Echocardiogram Models for Hypertrophic Cardiomyopathy Detection. Circulation 146, 755– 769 (2022). 35916132 PMC9439630
  8. [9]Ko W.-Y. Detection of Hypertrophic Cardiomyopathy Using a Convolutional Neural Network-Enabled Electrocardiogram. J. Am. Coll. Cardiol. 75, 722– 733 (2020). 32081280
  9. [10]Finocchiaro G. The electrocardiogram in the diagnosis and management of patients with hypertrophic cardiomyopathy. Heart Rhythm 17, 142– 151 (2020). 31349064
  10. [11]McLeod C. J. Outcome of patients with hypertrophic cardiomyopathy and a normal electrocardiogram. J. Am. Coll. Cardiol. 54, 229– 233 (2009). 19589435
  11. [12]Siontis K. C. Detection of hypertrophic cardiomyopathy by an artificial intelligence electrocardiogram in children and adolescents. Int. J. Cardiol. 340, 42– 47 (2021). 34419527
  12. [13]Magnusson P., Palm A., Branden E. & Mörner S. Misclassification of hypertrophic cardiomyopathy: validation of diagnostic codes. Clin. Epidemiol. 9, 403– 410 (2017). 28848363 PMC5557115
  13. [14]Farahani N. Z. Explanatory analysis of a machine learning model to identify hypertrophic cardiomyopathy patients from EHR using diagnostic codes. Proceedings (IEEE Int. Conf. Bioinformatics Biomed.) 2020, 1932– 1937 (2020). 34316386 PMC8313105
  14. [15]Daskalov I. K., Dotsinsky I. A. & Christov I. I. Developments in ECG acquisition, preprocessing, parameter measurement, and recording. IEEE Eng. Med. Biol. Mag. 17, 50– 58 (1998). 9548081
  15. [16]Blanco-Velasco M., Weng B. & Barner K. E. ECG signal denoising and baseline wander correction based on the empirical mode decomposition. Comput. Biol. Med. 38, 1– 13 (2008). 17669389
  16. [17]Sangha V. Automated multilabel diagnosis on electrocardiographic images and signals. Nat Commun 13, 1583 (2022). 35332137 PMC8948243
  17. [18]Sangha V. Detection of Left Ventricular Systolic Dysfunction From Electrocardiographic Images. Circulation (2023) doi: 10.1161/CIRCULATIONAHA.122.062646. PMC10982757 37489538 · DOI · Paper page
  18. [19]Dhingra L. S. Ensemble deep learning algorithm for structural heart disease screening using electrocardiographic images: PRESENT SHD. J. Am. Coll. Cardiol. (2025) doi: 10.1016/j.jacc.2025.01.030. PMC12199746 40139886 · DOI · Paper page
  19. [20]Tan Mingxing and Le Quoc V. EfficientNet: Rethinking model scaling for convolutional neural networks. International Conference on Machine Learning, 2019.
  20. [21]Sangha V. Biometric contrastive learning for data-efficient deep learning from electrocardiographic images. J. Am. Med. Inform. Assoc. (2024) doi: 10.1093/jamia/ocae002. PMC10990541 38269618 · DOI
  21. [22]Selvaraju R. R. Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization. 2017 Ieee International Conference on Computer Vision (Iccv) 618– 626 (2017).
  22. [23]Gow B. MIMIC-IV-ECG: Diagnostic electrocardiogram matched subset. PhysioNet 10.13026/4NQG-SB35 (2023). · DOI · Paper page
  23. [24]Johnson A., Pollard T., Horng S., Celi L. A. & Mark R. MIMIC-IV-Note: Deidentified free-text clinical notes. PhysioNet 10.13026/1N74-NE17 (2023). · DOI · Paper page
  24. [25]Petersen S. E. Reference ranges for cardiac structure and function using cardiovascular magnetic resonance (CMR) in Caucasians from the UK Biobank population cohort. J. Cardiovasc. Magn. Reson. 19, (2017). PMC5304550 28178995
  25. [26]Naderi H. Predicting left ventricular hypertrophy from the 12-lead electrocardiogram in the UK Biobank imaging study using machine learning. Eur. Heart J. Digit. Health 4, 316– 324 (2023). 37538142 PMC10393938
  26. [27]Oikonomou E. K. Artificial intelligence-enhanced risk stratification of cancer therapeutics-related cardiac dysfunction using electrocardiographic images. Circ. Cardiovasc. Qual. Outcomes (2024) doi: 10.1161/CIRCOUTCOMES.124.011504. PMC11745701 39221857 · DOI · Paper page
  27. [28]Dhingra L. S. Heart failure risk stratification using artificial intelligence applied to electrocardiogram images: a multinational study. Eur. Heart J. (2025) doi: 10.1093/eurheartj/ehae914. PMC12086686 39804243 · DOI
  28. [29]Delcrè S. D. L. Relationship of ECG findings to phenotypic expression in patients with hypertrophic cardiomyopathy: A cardiac magnetic resonance study. Int. J. Cardiol. 167, 1038– 1045 (2013). 22464482
  29. [30]Charron P. Diagnostic value of electrocardiography and echocardiography for familial hypertrophic cardiomyopathy in a genotyped adult population. Circulation 96, 214– 219 (1997). 9236436
  30. [31]Sharma S. International recommendations for electrocardiographic interpretation in athletes. Eur. Heart J. 39, 1466– 1480 (2018). 28329355
  31. [32]Dhingra L. S. A multicenter evaluation of the impact of therapies on deep learning-based electrocardiographic hypertrophic cardiomyopathy markers. Am. J. Cardiol. 237, 35– 40 (2024). 39581517 PMC11761372
  32. [33]Lamichhane B. & Neupane N. Improved healthcare access in low-resource regions: A review of technological solutions. arXiv [cs.CY] (2022).
  33. [34]Sau A. A comparison of artificial intelligence-enhanced electrocardiography approaches for prediction of time-to-mortality using electrocardiogram images. Eur. Heart J. Digit. Health (2024) doi: 10.1093/ehjdh/ztae090. PMC11914724 40110221 · DOI
  34. [35]Norrish G. Yield of clinical screening for hypertrophic cardiomyopathy in child firstdegree relatives. Circulation 140, 184– 192 (2019). 31006259 PMC6636798
  35. [36]Lafreniere-Roula M. Family screening for hypertrophic cardiomyopathy: Is it time to change practice guidelines? Eur. Heart J. 40, 3672– 3681 (2019). 31170284 PMC6885133
  36. [37]Corrado D., Basso C., Schiavon M. & Thiene G. Screening for hypertrophic cardiomyopathy in young athletes. N. Engl. J. Med. 339, 364– 369 (1998). 9691102
  37. [39]Anderson B. R., McElligott S., Polsky D. & Vetter V. L. Electrocardiographic screening for hypertrophic cardiomyopathy and long QT syndrome: The drivers of cost-effectiveness for the prevention of sudden cardiac death. Pediatr. Cardiol. 35, 323– 331 (2014). 24005901
  38. [40]Aminorroaya A. Development and multinational validation of an ensemble deep learning algorithm for detecting and predicting structural heart disease using noisy single-lead electrocardiograms. Eur. Heart J. Digit. Health (2025) doi: 10.1093/ehjdh/ztaf034. PMC12282373 40703117 · DOI
  39. [41]Albahri A. S. A systematic review of trustworthy and explainable artificial intelligence in healthcare: Assessment of quality, bias risk, and data fusion. Inf. Fusion 96, 156– 191 (2023).
  40. [43]Basavarajaiah S. Prevalence of hypertrophic cardiomyopathy in highly trained athletes. J. Am. Coll. Cardiol. 51, 1033– 1039 (2008). 18325444
  41. [44]Goldberger A. L. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 101, E215– 20 (2000). 10851218
  42. [45]Sun X. & Xu W. Fast implementation of DeLong’s algorithm for comparing the areas under correlated receiver operating characteristic curves. IEEE Signal Process. Lett. 21, 1389– 1393 (2014).
  43. [46]DeLong E. R., DeLong D. M. & Clarke-Pearson D. L. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 44, 837– 845 (1988). 3203132

Cited by

Sorted by citations

Authors
Sangha, V., Dhingra, L. S., Aminorroaya, A., Croon, P. M., Sikand, N. V., Sen, S., Martinez, M. W., Maron, M. S., Krumholz, H. M., Asselbergs, F. W., Oikonomou, E. K., & Khera, R.
Affiliations
  • Yale University (US)
  • Amsterdam University Medical Centers (NL)
  • Morristown Medical Center (US)
  • Tufts Medical Center (US)
  • Yale New Haven Hospital (US)
  • National Institute for Health and Care Research (GB)
Corresponding author
Khera, Rohan
ORCID
Venue
Nature Cardiovascular Research
Publisher
openRxiv
Published
2023-12-28
Volume
4
Issue
8
Pages
991-1000
ISSN
2731-0590 (ISSN-L)
DOI
10.1101/2023.12.23.23300490
PMC
PMC10793540
OpenAlex
W4412552358
Cited by
38
References
0
Citations by year
2026: 18 · 2025: 13 · 2024: 7
FWCI
19.28
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Cardiovascular Research (publishedVersion)
  • medRxiv (acceptedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion, cc-by-nc-nd)
  • UCL Discovery (University College London) (submittedVersion)
  • Pure Amsterdam UMC (submittedVersion)
Funders
  • U.S. Department of Health and Human Services (1F32HL170592-01, K23HL153775, R01HL167858)
  • National Institutes of Health (1r01ag089981-01)
  • National Heart, Lung, and Blood Institute (F32HL170592, R01HL167858, K23HL153775)
Topics
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › ECG Monitoring and Analysis
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › Cardiomyopathy and Myosin Studies
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › Cardiac electrophysiology and arrhythmias
Keywords
hypertrophic cardiomyopathy, electrocardiogram, 12-lead ECG, deep learning, automated disease detection, cardiac magnetic resonance, echocardiography, external validation, MIMIC-IV, UK Biobank

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review