Hierarchical Transformer for Electrocardiogram Diagnosis

Xiaoya Tang, Jake Berquist, Benjamin A. Steinberg, Tolga Tasdizen
11/1/2024
cs.LG

No official code found. Best implementation: xiaoyatang/3stageFormer (1 star, BSD-2-Clause, last commit 2025-07-20).

Abstract

We propose a hierarchical Transformer for ECG analysis that combines depth-wise convolutions, multi-scale feature aggregation via a CLS token, and an attention-gated module to learn inter-lead relationships and enhance interpretability. The model is lightweight, flexible, and eliminates the need for complex attention or downsampling strategies.

View on arXivView PDF

Code Implementations(1)

xiaoyatang/3stageFormerCommunity54%
BSD-2-Clause

Implementation of paper 'Hierarchical Transformer for Electrocardiogram Diagnosis'(MIDL 2025 short), applicable for PhysioNet2020 and your customized ECG dataset.

10MATLAB, PythonJan 6, 20251 years agoBSD-2-Clause

Similar Papers

CardioPatternFormer: Pattern-Guided Attention for Interpretable ECG Classification with Transformer Architecture

86% match2025

Hierarchical Attention Network for Interpretable ECG-based Heart Disease Classification

84% match2025

MSW-Transformer: Multi-Scale Shifted Windows Transformer Networks for 12-Lead ECG Classification

84% match2023

FoundationalECGNet: A Lightweight Foundational Model for ECG-based Multitask Cardiac Analysis

83% match2025

ECG-Soup: Harnessing Multi-Layer Synergy for ECG Foundation Models

82% match2025

Tracing the Heart's Pathways: ECG Representation Learning from a Cardiac Conduction Perspective

82% match2025

Cite this paper

@article{tang2024hierarchical,
  title  = {Hierarchical Transformer for Electrocardiogram Diagnosis},
  author = {Xiaoya Tang and Jake Berquist and Benjamin A. Steinberg and Tolga Tasdizen},
  year   = {2024},
  eprint = {2411.00755},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2411.00755v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2411.00755.svg)](https://codeofpaper.com/paper/2411.00755)

Discussion