![[2507.14206] A Comprehensive Benchmark for Electrocardiogram Time-Series](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2507.14206] A Comprehensive Benchmark for Electrocardiogram Time-Series
Summary
This article presents a comprehensive benchmark for electrocardiogram (ECG) time-series analysis, highlighting its unique characteristics and proposing new evaluation metrics and model architectures.
Why It Matters
The study addresses the gap in ECG analysis by providing a structured benchmark that categorizes applications, identifies limitations in existing metrics, and proposes a new architecture. This is crucial for advancing research in cardiac health diagnostics and improving machine learning applications in healthcare.
Key Takeaways
- Introduces a comprehensive benchmark for ECG time-series analysis.
- Categorizes ECG applications into four distinct evaluation tasks.
- Identifies limitations in traditional evaluation metrics and proposes a novel metric.
- Benchmarks state-of-the-art time-series models for ECG analysis.
- Proposes a new architecture validated through extensive experiments.
Electrical Engineering and Systems Science > Signal Processing arXiv:2507.14206 (eess) [Submitted on 15 Jul 2025 (v1), last revised 25 Feb 2026 (this version, v2)] Title:A Comprehensive Benchmark for Electrocardiogram Time-Series Authors:Zhijiang Tang, Jiaxin Qi, Yuhua Zheng, Jianqiang Huang View a PDF of the paper titled A Comprehensive Benchmark for Electrocardiogram Time-Series, by Zhijiang Tang and 3 other authors View PDF HTML (experimental) Abstract:Electrocardiogram~(ECG), a key bioelectrical time-series signal, is crucial for assessing cardiac health and diagnosing various diseases. Given its time-series format, ECG data is often incorporated into pre-training datasets for large-scale time-series model training. However, existing studies often overlook its unique characteristics and specialized downstream applications, which differ significantly from other time-series data, leading to an incomplete understanding of its properties. In this paper, we present an in-depth investigation of ECG signals and establish a comprehensive benchmark, which includes (1) categorizing its downstream applications into four distinct evaluation tasks, (2) identifying limitations in traditional evaluation metrics for ECG analysis, and introducing a novel metric; (3) benchmarking state-of-the-art time-series models and proposing a new architecture. Extensive experiments demonstrate that our proposed benchmark is comprehensive and robust. The results validate the effectiveness of the pro...
