arxiv: 2605.09848 · v1 · submitted 2026-05-11 · 💻 cs.LG

Recognition: 2 theorem links

· Lean Theorem

Efficient Neural Architectures for Real-Time ECG Interpretation on Limited Hardware

Ashery Mbilinyi , Callum O'Riley , Julia Handra , Ashley Moller-Hansen , Jason Andrade , Marc Deyell , Cameron Hague , Nathaniel Hawkins

show 3 more authors

Kendall Ho Jonathan Leipsic Roger Tam

Authors on Pith no claims yet

Pith reviewed 2026-05-12 04:34 UTC · model grok-4.3

classification 💻 cs.LG

keywords ECG classificationlightweight CNNefficiency metricreal-time diagnosiscardiac signal processingembedded AImultilabel classification

0 comments

The pith

Lightweight CNN variants for ECG signals deliver competitive diagnostic accuracy at far lower computational cost than standard models.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper conducts an empirical comparison of convolutional neural network designs for automated 12-lead ECG classification across binary, multiclass, and multilabel tasks. It benchmarks two established models against three newly proposed lightweight architectures that either split temporal and spatial feature extraction into parallel branches or process both dimensions jointly. Experiments cover three international datasets and test the addition of basic demographic inputs, with all models ranked by a single Efficiency Score that folds together AUC, model size, inference speed, and memory footprint. The central effort is to show that these compact designs can approach the performance of heavier networks while remaining practical for real-time use on constrained hardware.

Core claim

By introducing ParallelCNN with dual temporal-spatial branches, its symmetric-initialization variant ParallelCNNew, and the streamlined SimpleNet that jointly handles both dimensions, the authors produce models whose diagnostic AUCs remain close to those of larger baselines while exhibiting substantially smaller parameter counts, faster inference, and lower memory use; augmenting any of them with age and sex metadata yields further gains at negligible overhead.

What carries the argument

The Efficiency Score, a composite ranking that multiplies normalized AUC by the inverse of model size, inference latency, and peak memory usage, used to identify architectures suitable for deployment on limited hardware.

If this is right

The models can support real-time ECG interpretation on portable or bedside devices without cloud offloading.
Adding only age and sex metadata improves performance across all tested tasks with almost no extra compute.
The same lightweight designs remain effective across ECG datasets collected in three different countries.
A single scalar Efficiency Score suffices to trade off accuracy against resource use when selecting models for clinical deployment.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If the Efficiency Score generalizes, it could serve as a standard yardstick for comparing medical time-series models beyond ECG.
Parallel-branch designs may transfer directly to other physiological signals such as EEG or PPG that also contain separable temporal and spatial structure.
Deployment on truly edge hardware would still require separate profiling of power draw and quantization effects not measured here.

Load-bearing premise

The lightweight models will retain their measured accuracy when run on new patient populations or real clinical hardware without retraining or architecture changes.

What would settle it

A statistically significant drop in AUC or rise in error rate when any of the proposed models is tested on an independent ECG dataset drawn from a previously unseen demographic group or on actual embedded-device hardware.

Figures

Figures reproduced from arXiv: 2605.09848 by Ashery Mbilinyi, Ashley Moller-Hansen, Callum O'Riley, Cameron Hague, Jason Andrade, Jonathan Leipsic, Julia Handra, Kendall Ho, Marc Deyell, Nathaniel Hawkins, Roger Tam.

**Figure 1.** Figure 1: The standard ECG curve with its most common waveforms. Important intervals and points of measurement are depicted. Manual ECG interpretation [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

**Figure 2.** Figure 2: AttiaNet architecture [26] stack, features are flattened and passed through dense layers, with optional integration of demographic metadata. Owing to its simplicity and small parameter footprint, SimpleNet provides a strong efficiency-focused baseline against which more complex architectures can be compared. IV. DATASETS AND TASKS To evaluate the performance and generalizability of the five CNN architectur… view at source ↗

**Figure 3.** Figure 3: ParallelCNN architecture: dual temporal and spatial pathways with parallel convolutional blocks and feature fusion. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** Figure 4: AUC scores for multilabel classification. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

**Figure 5.** Figure 5: AUC scores for multiclass classification. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗

**Figure 6.** Figure 6: AUC scores for binary classification. These findings suggest that lightweight architectures already capture most of the discriminatory signals from ECG waveforms, leaving limited additional benefit from demographic features. Nevertheless, the consistent (albeit small) improvements highlight the potential value of integrating inexpensive metadata in clinical AI systems, particularly for models that strugg… view at source ↗

read the original abstract

Electrocardiogram (ECG) interpretation is essential for diagnosing a wide range of cardiac abnormalities. While deep learning has shown strong potential for automating ECG classification, many existing models rely on large, computationally intensive architectures that hinder practical deployment. In this paper, we present an empirical study of convolutional neural network (CNN) architectures, exploring tradeoffs between diagnostic accuracy and computational efficiency. We benchmark two established baselines: AttiaNet, a compact model composed of sequential temporal and spatial blocks, and DeepResidualCNN, the winning architecture of the 2021 PhysioNet/Computing in Cardiology Challenge. Building on these, we propose three lightweight models: (i) ParallelCNN, which employs dual temporal and spatial branches for parallel pattern extraction; (ii) ParallelCNNew, a variant with symmetric weight initialization for balanced feature learning; and (iii) SimpleNet, a streamlined architecture that jointly processes temporal and spatial dimensions. Our experiments span three publicly available 12-lead ECG datasets from Germany, China, and the United States, covering binary, multiclass, and multilabel classification tasks across diverse patient populations. We further evaluate the impact of integrating low-cost demographic metadata (age and sex) to improve performance with minimal overhead. To ensure fair comparison, we introduce a unified Efficiency Score that integrates model size, inference speed, memory usage, and AUC performance. By balancing diagnostic performance and efficiency, our models offer a scalable and viable foundation for next-generation AI systems in cardiovascular care.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This is a solid but incremental empirical comparison of lightweight CNN variants for ECG classification on public datasets, with a new composite efficiency metric and some demographic inputs.

read the letter

Hey, about this ECG paper on efficient architectures for limited hardware. The main takeaway is that it runs a practical bake-off of three tweaked CNN models against two established baselines, testing them across three public 12-lead datasets from different countries and tasks. They add age and sex as cheap extra features and propose an Efficiency Score that folds in model size, speed, memory, and AUC to pick winners on the tradeoff curve. The experiments look straightforward with no internal contradictions in the setup or protocol. What the paper does well is cover a range of classification types and geographies while keeping the focus on deployability rather than just chasing higher AUC. The demographic addition is a low-overhead move that could matter in real settings, and benchmarking on standard hardware gives concrete numbers for inference and memory. The soft spots are minor but real. The proposed models are direct variants of the cited baselines with parallel branches or symmetric init, so the novelty is mostly in the specific combinations and the ad-hoc score rather than a new approach. The Efficiency Score itself lacks much motivation for its exact weighting, which makes it feel like a reasonable but arbitrary summary metric. The abstract itself has no numbers, which makes the balancing claim hard to judge without the full results, though the stress-test indicates the comparisons are direct and use external data. This paper is for readers working on medical signal ML for edge or clinical devices who need concrete tradeoff data rather than theory. It would give value to someone implementing ECG classifiers who wants to see how small changes affect the efficiency frontier. It deserves a serious referee because the question is relevant, the data are public, and the evaluation protocol is reproducible enough to check. I would send it to peer review with a note to strengthen the justification for the new score and perhaps add more ablation on the demographic features.

Referee Report

2 major / 2 minor

Summary. The manuscript presents an empirical benchmarking study of CNN architectures for 12-lead ECG classification. It compares two baselines (AttiaNet and DeepResidualCNN) against three proposed lightweight variants (ParallelCNN, ParallelCNNew, SimpleNet), evaluates them across binary/multiclass/multilabel tasks on three public datasets from Germany, China, and the US, introduces a composite Efficiency Score, and tests the addition of age/sex metadata.

Significance. If the empirical results and Efficiency Score hold up under scrutiny, the work could offer practical guidance for deploying real-time ECG models on resource-constrained hardware. The multi-dataset, multi-task evaluation and explicit focus on efficiency metrics address a genuine deployment gap in cardiovascular AI.

major comments (2)

[Abstract] Abstract: The abstract describes the experimental setup, datasets, and Efficiency Score but supplies no quantitative results, error analysis, ablation details, or specific AUC/efficiency numbers to support the central claim of balanced diagnostic performance and efficiency. This makes it impossible to assess whether the lightweight models actually preserve accuracy relative to the baselines.
[Methods (Efficiency Score definition)] The Efficiency Score is presented as the unifying metric integrating model size, inference speed, memory, and AUC, yet no justification, weighting scheme, or sensitivity analysis is provided for how these components are combined. Without this, it is unclear whether the score meaningfully reflects practical deployment tradeoffs or simply re-ranks models in an ad-hoc manner.

minor comments (2)

[Results] The manuscript would benefit from explicit reporting of confidence intervals or statistical significance tests on the AUC differences across models and datasets.
[Model Architectures] Clarify whether the lightweight variants were derived by systematic pruning/search or by manual simplification of the baselines; this affects reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments on our manuscript. We have revised the abstract to include quantitative results and expanded the methods section with justification and analysis for the Efficiency Score.

read point-by-point responses

Referee: [Abstract] Abstract: The abstract describes the experimental setup, datasets, and Efficiency Score but supplies no quantitative results, error analysis, ablation details, or specific AUC/efficiency numbers to support the central claim of balanced diagnostic performance and efficiency. This makes it impossible to assess whether the lightweight models actually preserve accuracy relative to the baselines.

Authors: We agree that the original abstract was primarily descriptive and lacked specific quantitative support. In the revised manuscript, we have updated the abstract to report key AUC values for the proposed models versus baselines across the three datasets, along with efficiency metrics such as model size reduction and inference speed improvements. We have also added brief references to the ablation studies and error analysis to substantiate the claims of balanced performance and efficiency. revision: yes
Referee: [Methods (Efficiency Score definition)] The Efficiency Score is presented as the unifying metric integrating model size, inference speed, memory, and AUC, yet no justification, weighting scheme, or sensitivity analysis is provided for how these components are combined. Without this, it is unclear whether the score meaningfully reflects practical deployment tradeoffs or simply re-ranks models in an ad-hoc manner.

Authors: We acknowledge the need for greater transparency in the Efficiency Score definition. The revised manuscript now includes an expanded subsection in Methods that justifies the composite formulation, specifies the weighting scheme (equal weights applied after min-max normalization of each component to ensure balanced contribution), and reports a sensitivity analysis in which we vary the weights by ±20% and demonstrate that model rankings remain stable. These additions show that the score captures meaningful deployment tradeoffs rather than producing arbitrary rankings. revision: yes

Circularity Check

0 steps flagged

No significant circularity in empirical benchmarking study

full rationale

This paper is an empirical benchmarking study that compares CNN architectures on three independent public ECG datasets (from Germany, China, and the US) covering binary, multiclass, and multilabel tasks. It evaluates baselines (AttiaNet, DeepResidualCNN) and proposes lightweight variants (ParallelCNN, ParallelCNNew, SimpleNet), measures model size/inference speed/memory/AUC directly, and defines a composite Efficiency Score from those measured quantities. No derivation chain, first-principles predictions, or fitted parameters are claimed; all results are obtained by running the models on external data. The work contains no self-definitional steps, no predictions that reduce to their own inputs by construction, and no load-bearing self-citations that substitute for independent evidence. The central claims rest on reproducible experimental comparisons rather than any closed logical loop.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is an applied empirical machine learning study. No explicit free parameters, mathematical axioms, or invented entities are described in the abstract; performance depends on standard CNN training practices and public data.

pith-pipeline@v0.9.0 · 5596 in / 955 out tokens · 68430 ms · 2026-05-12T04:34:24.053833+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We propose a unified Efficiency Score that integrates model size, inference speed, memory usage, and AUC performance. EfficiencyScore=λ·AUC+(1−λ)·(1−ResourceCost)
IndisputableMonolith/Foundation/AbsoluteFloorClosure.lean absolute_floor_iff_bare_distinguishability unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

AttiaNet, a compact model composed of sequential temporal and spatial blocks

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages

[1]

Leading causes of death in the us, 2019-2023,

F. B. Ahmad, J. A. Cisewski, and R. N. Anderson, “Leading causes of death in the us, 2019-2023,”JAMA, 2024

work page 2019
[2]

The top 10 causes of death

WHO, “The top 10 causes of death.” https://www.who.int/news-room/ fact-sheets/detail/the-top-10-causes-of-death, 2024. Retrieved Au- gust 13, 2024, from https://www.who.int/news-room/fact-sheets/detail/ the-top-10-causes-of-death

work page 2024
[3]

The role of machine learning in the detection of cardiac fibrosis in electro- cardiograms: Scoping review,

J. Handra, H. James, A. Mbilinyi, A. Moller-Hansen, C. O’Riley, J. Andrade, M. Deyell, C. Hague, N. Hawkins, K. Ho,et al., “The role of machine learning in the detection of cardiac fibrosis in electro- cardiograms: Scoping review,”JMIR cardio, vol. 8, no. 1, p. e60697, 2024

work page 2024
[4]

Automatic diagnosis of the 12-lead ecg using a deep neural network,

A. H. Ribeiro, M. H. Ribeiro, G. M. Paix ˜ao, D. M. Oliveira, P. R. Gomes, J. A. Canazart, M. P. Ferreira, C. R. Andersson, P. W. Macfarlane, W. Meira Jr,et al., “Automatic diagnosis of the 12-lead ecg using a deep neural network,”Nature communications, vol. 11, no. 1, p. 1760, 2020

work page 2020
[5]

Deep learning for ecg arrhythmia detection and classification: an overview of progress for period 2017–2023,

Y . Ansari, O. Mourad, K. Qaraqe, and E. Serpedin, “Deep learning for ecg arrhythmia detection and classification: an overview of progress for period 2017–2023,”Frontiers in Physiology, vol. 14, p. 1246746, 2023

work page 2017
[6]

Accuracy of physicians’ elec- trocardiogram interpretations: a systematic review and meta-analysis,

D. A. Cook, S.-Y . Oh, and M. V . Pusic, “Accuracy of physicians’ elec- trocardiogram interpretations: a systematic review and meta-analysis,” JAMA internal medicine, vol. 180, no. 11, pp. 1461–1471, 2020

work page 2020
[7]

Exploring the impact of expertise, clinical history, and visual search on electrocardiogram interpretation,

G. Wood, J. Batt, A. Appelboam, A. Harris, and M. R. Wilson, “Exploring the impact of expertise, clinical history, and visual search on electrocardiogram interpretation,”Medical Decision Making, vol. 34, no. 1, pp. 75–83, 2014

work page 2014
[8]

Competency in interpretation of 12-lead electrocardiogram among swiss doctors.,

J. J. Goy, J. Schlaepfer, and J.-C. Stauffer, “Competency in interpretation of 12-lead electrocardiogram among swiss doctors.,”Swiss medical weekly, vol. 143, 2013

work page 2013
[9]

Accuracy of electrocardiogram interpretation by cardiologists in the setting of incorrect computer analysis,

D. Anh, S. Krishnan, and F. Bogun, “Accuracy of electrocardiogram interpretation by cardiologists in the setting of incorrect computer analysis,”Journal of electrocardiology, vol. 39, no. 3, pp. 343–345, 2006

work page 2006
[10]

Ecg interpretation: clinical relevance, challenges, and advances,

N. Rafie, A. H. Kashou, and P. A. Noseworthy, “Ecg interpretation: clinical relevance, challenges, and advances,”Hearts, vol. 2, no. 4, pp. 505–513, 2021

work page 2021
[11]

State-of-the-art deep learning methods on electrocardiogram data: systematic review,

G. Petmezas, L. Stefanopoulos, V . Kilintzis, A. Tzavelis, J. A. Rogers, A. K. Katsaggelos, and N. Maglaveras, “State-of-the-art deep learning methods on electrocardiogram data: systematic review,”JMIR medical informatics, vol. 10, no. 8, p. e38454, 2022

work page 2022
[12]

Machine learning of electrophysiological signals for the prediction of ventric- ular arrhythmias: systematic review and examination of heterogeneity between studies,

M. Z. Kolk, B. Deb, S. Ruip ´erez-Campillo, N. K. Bhatia, P. Clopton, A. A. Wilde, S. M. Narayan, R. E. Knops, and F. V . Tjong, “Machine learning of electrophysiological signals for the prediction of ventric- ular arrhythmias: systematic review and examination of heterogeneity between studies,”EBioMedicine, vol. 89, 2023

work page 2023
[13]

Validation of an automated artificial intelligence system for 12-lead ecg interpretation,

R. Herman, A. Demolder, B. Vavrik, M. Martonak, V . Boza, V . Kres- nakova, A. Iring, T. Palus, J. Bahyl, O. Nelis,et al., “Validation of an automated artificial intelligence system for 12-lead ecg interpretation,” Journal of Electrocardiology, vol. 82, pp. 147–154, 2024

work page 2024
[14]

The importance of resource awareness in artificial intelligence for healthcare,

Z. Jia, J. Chen, X. Xu, J. Kheir, J. Hu, H. Xiao, S. Peng, X. S. Hu, D. Chen, and Y . Shi, “The importance of resource awareness in artificial intelligence for healthcare,”Nature Machine Intelligence, vol. 5, no. 7, pp. 687–698, 2023

work page 2023
[15]

Ecg interpretation characteristics of the normal ecg (p- wave, qrs complex, st segment, t-wave),

E. E. Learning, “Ecg interpretation characteristics of the normal ecg (p- wave, qrs complex, st segment, t-wave),” 2023. Accessed: November 23, 2023

work page 2023
[16]

U-net: Convolutional networks for biomedical image segmentation,

O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” inMedical image computing and computer-assisted intervention–MICCAI 2015: 18th international con- ference, Munich, Germany, October 5-9, 2015, proceedings, part III 18, pp. 234–241, Springer, 2015

work page 2015
[17]

High-resolution swin transformer for automatic medical image segmentation,

C. Wei, S. Ren, K. Guo, H. Hu, and J. Liang, “High-resolution swin transformer for automatic medical image segmentation,”Sensors, vol. 23, no. 7, p. 3420, 2023

work page 2023
[18]

Energy and policy consider- ations for modern deep learning research,

E. Strubell, A. Ganesh, and A. McCallum, “Energy and policy consider- ations for modern deep learning research,” inProceedings of the AAAI conference on artificial intelligence, vol. 34, pp. 13693–13696, 2020

work page 2020
[19]

Rt-rcg: Neural network and accelerator search towards effective and real-time ecg reconstruction from intracardiac electrograms,

Y . Zhang, A. Banta, Y . Fu, M. M. John, A. Post, M. Razavi, J. Cavallaro, B. Aazhang, and Y . Lin, “Rt-rcg: Neural network and accelerator search towards effective and real-time ecg reconstruction from intracardiac electrograms,”ACM Journal on Emerging Technologies in Computing Systems (JETC), vol. 18, no. 2, pp. 1–25, 2022

work page 2022
[20]

Fairprune: Achieving fairness through pruning for dermatological disease diagnosis,

Y . Wu, D. Zeng, X. Xu, Y . Shi, and J. Hu, “Fairprune: Achieving fairness through pruning for dermatological disease diagnosis,” inInternational Conference on Medical Image Computing and Computer-Assisted Inter- vention, pp. 743–753, Springer, 2022

work page 2022
[21]

Efficient hardware implementation of cellular neural networks with incremental quantization and early exit,

X. Xu, Q. Lu, T. Wang, Y . Hu, C. Zhuo, J. Liu, and Y . Shi, “Efficient hardware implementation of cellular neural networks with incremental quantization and early exit,”ACM Journal on Emerging Technologies in Computing Systems (JETC), vol. 14, no. 4, pp. 1–20, 2018

work page 2018
[22]

Medq: Lossless ultra-low-bit neural network quantization for medical image segmentation,

R. Zhang and A. C. Chung, “Medq: Lossless ultra-low-bit neural network quantization for medical image segmentation,”Medical Image Analysis, vol. 73, p. 102200, 2021

work page 2021
[23]

A foundational vision transformer improves diagnostic performance for electrocardiograms,

A. Vaid, J. Jiang, A. Sawant, S. Lerakis, E. Argulian, Y . Ahuja, J. Lam- pert, A. Charney, H. Greenspan, J. Narula,et al., “A foundational vision transformer improves diagnostic performance for electrocardiograms,” NPJ Digital Medicine, vol. 6, no. 1, p. 108, 2023

work page 2023
[24]

Ecg-fm: An open electrocardiogram foundation model,

K. McKeen, L. Oliva, S. Masood, A. Toma, B. Rubin, and B. Wang, “Ecg-fm: An open electrocardiogram foundation model,”arXiv preprint arXiv:2408.05178, 2024

work page arXiv 2024
[25]

Convolutional neural networks for electrocardiogram classification,

M. M. Al Rahhal, Y . Bazi, M. Al Zuair, E. Othman, and B. BenJdira, “Convolutional neural networks for electrocardiogram classification,” Journal of Medical and Biological Engineering, vol. 38, pp. 1014–1025, 2018

work page 2018
[26]

Screening for cardiac contractile dysfunction using an artificial intelligence–enabled electrocardiogram,

Z. I. Attia, S. Kapa, F. Lopez-Jimenez, P. M. McKie, D. J. Ladewig, G. Satam, P. A. Pellikka, M. Enriquez-Sarano, P. A. Noseworthy, T. M. Munger,et al., “Screening for cardiac contractile dysfunction using an artificial intelligence–enabled electrocardiogram,”Nature medicine, vol. 25, no. 1, pp. 70–74, 2019

work page 2019
[27]

Classification of ecg using ensemble of residual cnns with attention mechanism,

P. Nejedly, A. Ivora, R. Smisek, I. Viscor, Z. Koscova, P. Jurak, and F. Plesinger, “Classification of ecg using ensemble of residual cnns with attention mechanism,” in2021 Computing in Cardiology (CinC), vol. 48, pp. 1–4, IEEE, 2021

work page 2021
[28]

Will two do? varying dimensions in electrocardiography: the physionet/computing in cardiology challenge 2021,

M. A. Reyna, N. Sadr, E. A. P. Alday, A. Gu, A. J. Shah, C. Robichaux, A. B. Rad, A. Elola, S. Seyedi, S. Ansari,et al., “Will two do? varying dimensions in electrocardiography: the physionet/computing in cardiology challenge 2021,” in2021 computing in cardiology (CinC), vol. 48, pp. 1–4, IEEE, 2021

work page 2021
[29]

Deep residual learning for image recognition,

K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2016

work page 2016
[30]

PTB-XL, a large publicly available electrocardiography dataset (version 1.0.3),

P. Wagner, N. Strodthoff, R. Bousseljot, W. Samek, and T. Schaeffter, “PTB-XL, a large publicly available electrocardiography dataset (version 1.0.3),” 2022

work page 2022
[31]

Ptb-xl, a large publicly available electro- cardiography dataset,

P. Wagner, N. Strodthoff, R.-D. Bousseljot, D. Kreiseler, F. I. Lunze, W. Samek, and T. Schaeffter, “Ptb-xl, a large publicly available electro- cardiography dataset,”Scientific data, vol. 7, no. 1, pp. 1–15, 2020

work page 2020
[32]

A 12-lead electrocardiogram database for arrhythmia research covering more than 10,000 patients,

J. Zheng, J. Zhang, S. Danioko, H. Yao, H. Guo, and C. Rakovski, “A 12-lead electrocardiogram database for arrhythmia research covering more than 10,000 patients,”Scientific data, vol. 7, no. 1, p. 48, 2020

work page 2020
[33]

C. T. January, L. S. Wann, H. Calkins, L. Y . Chen, J. E. Cigarroa, J. C. Cleveland Jr, P. T. Ellinor, M. D. Ezekowitz, M. E. Field, K. L. Furie,et al., “2019 aha/acc/hrs focused update of the 2014 aha/acc/hrs guideline for the management of patients with atrial fibrillation: a report of the american college of cardiology/american heart association task f...

work page 2019
[34]

R. L. Page, J. A. Joglar, M. A. Caldwell, H. Calkins, J. B. Conti, B. J. Deal, N. M. Estes III, M. E. Field, Z. D. Goldberger, S. C. Hammill, et al., “2015 acc/aha/hrs guideline for the management of adult patients with supraventricular tachycardia: a report of the american college of cardiology/american heart association task force on clinical practice g...

work page 2015
[35]

Esc guidelines for the management of atrial fibrillation developed in collaboration with eacts,

P. Kirchhof, S. Benussi, D. Kotecha, A. Ahlsson, D. Atar, B. Casadei, et al., “Esc guidelines for the management of atrial fibrillation developed in collaboration with eacts,”European Heart Journal, vol. 37, no. 7, pp. 2893–2962, 2016

work page 2016
[36]

Physiobank, physiotoolkit, and physionet: components of a new research resource for complex physiologic signals,

A. L. Goldberger, L. A. Amaral, L. Glass, J. M. Hausdorff, P. C. Ivanov, R. G. Mark, J. E. Mietus, G. B. Moody, C.-K. Peng, and H. E. Stanley, “Physiobank, physiotoolkit, and physionet: components of a new research resource for complex physiologic signals,”circulation, vol. 101, no. 23, pp. e215–e220, 2000

work page 2000
[37]

Mimic- iv-ecg: Diagnostic electrocardiogram matched subset,

B. Gow, T. Pollard, L. A. Nathanson, A. Johnson, B. Moody, C. Fernan- des, N. Greenbaum, J. W. Waks, P. Eslami, T. Carbonati,et al., “Mimic- iv-ecg: Diagnostic electrocardiogram matched subset,”Type: dataset, vol. 6, pp. 13–14, 2023

work page 2023