ECG Agent

This repository provides tools for [💼ECG analysis], including an interactive Gradio web interface and a unified test suite for verifying model performance across multiple tasks (Classification, Detection, Forecasting, Generation).

Prerequisites

Ensure you have the required Python packages installed:

pip install gradio wfdb torch pandas numpy matplotlib scipy scikit-learn omegaconf hydra-core

Hardware Requirements

• Inference: CPU or any CUDA-capable GPU (VRAM > 2GB).
• Training: NVIDIA GPU with VRAM > 6GB recommended (tested on RTX 3090).

Roadmap

• Introduce visualization interface for model training process.

1. Interactive Web Interface (Gradio)

The Gradio app provides a user-friendly interface to visualize ECG signals, crop segments, and run analysis tasks.

How to Run

1. Start the application:

   python gradio_app.py

2. Access the interface:

   • Open your web browser and go to: http://localhost:7860 (or the URL provided in the terminal).

Usage Steps

1. Upload: Drag and drop your .dat and .hea files (must be paired) into the upload box.
2. Load: Click "Load & Preprocess". The signal will be resampled to 100Hz and displayed.
3. Select: Use the "Start Index" slider to choose a 500-point window. The selected red region will update on the chart.
4. Analyze: Select a task (e.g., Classification) from the dropdown and click "Analyze Selected Segment".
5. View Results: The result visualization and text info will appear on the right. You can also download the result CSV.

2. Unified Test Suite

The unified test script verifies the functionality of all 4 tasks using a standard test file and generates a report.

How to Run

Execute the test script:

python unified_test.py

Output

• Console: Logs of the loading and inference process.
• Report: A test_report.md file summarizing the results.
• Visualization: A visualization.png image showing the signal processing steps.

3. Model Weights

The model weights used by these applications are stored in model_weights_backup/. Refer to MANIFEST.md for details on the source checkpoints and configurations.

🧑‍🤝‍🧑 Brothers link

• [💼 ECG-Benchmark]
• [🤖 ECG-Agent]
• [📖 A Comprehensive Benchmark for Electrocardiogram Time-Series]

📝 Citation

If you find this repo helpful, please cite our paper:

@inproceedings{tang2025comprehensive,
  title={A Comprehensive Benchmark for Electrocardiogram Time-Series},
  author={Tang, Zhijiang and Qi, Jiaxin and Zheng, Yuhua and Huang, Jianqiang},
  booktitle={Proceedings of the 33rd ACM International Conference on Multimedia},
  pages={6490--6499},
  year={2025}
}