@InProceedings{ CBMS2026,
	author = "Steven Hicks and Molly Maleckar and Thomas Dreibholz and Vajira Thambawita",
	title = "{DeepECG-Kit: A Practical and Reproducible Framework for ECG Deep Learning}",
	booktitle = "{Proceedings of the 39th IEEE International Symposium on Computer-Based Medical Systems~(CBMS)}",
	day = "5",
	month = jun,
	year = "2026",
	address = "Limassol/Cyprus",
	language = "english",
	keywords = "Electrocardiography, Deep Learning, Arrhythmia Classification, Reproducibility, Open Source, Benchmarking",
	abstract = "{We present DeepECG-Kit, an open-source Python library for reproducible ECG deep learning research. The library unifies multiple datasets under a standardized preprocessing pipeline and provides reference implementations of several neural network architectures. These span convolutional, recurrent, attention-based, and hybrid approaches. Built on PyTorch, DeepECG-Kit supports automated checkpointing, stratified data splitting, and comprehensive evaluation through both a command-line interface and a Python API. We demonstrate the library by benchmarking five architectures on four datasets covering multi-label diagnostic classification, multi-class rhythm detection, binary atrial fibrillation detection, and cross-dataset AF classification under identical conditions. DeepECG-Kit is freely available at https://github.com/stevenah/deepecg-kit and can be installed via pip install deepecgkit.}"
}