Department of Biomedical Engineering, Purdue University, USA.
Department of Mechanical Engineering, Pennsylvania State University, USA.
Comput Biol Med. 2024 Sep;179:108872. doi: 10.1016/j.compbiomed.2024.108872. Epub 2024 Jul 15.
We present a novel method for detecting atrial fibrillation (AFib) by analyzing Lead II electrocardiograms (ECGs) using a unique set of features.
For this purpose, we used specific signal processing techniques, such as proper orthogonal decomposition, continuous wavelet transforms, discrete cosine transform, and standard cross-correlation, to extract 48 features from the ECGs. Thus, our approach aims to more effectively capture AFib signatures, such as beat-to-beat variability and fibrillatory waves, than traditional metrics. Moreover, our features were designed to be physiologically interpretable. Subsequently, we incorporated an XGBoost-based ECG classifier to train and evaluate it using the publicly available 'Training' dataset of the 2017 PhysioNet Challenge, which includes 'Normal,' 'AFib,' 'Other,' and 'Noisy' ECG categories.
Our method achieved an accuracy of 96 % and an F1-score of 0.83 for 'AFib' detection and 80 % accuracy and 0.85 F1-score for 'Normal' ECGs. Finally, we compared our method's performance with the top-classifiers from the 2017 PhysioNet Challenge, namely ENCASE, Random Forest, and Cascaded Binary. As reported by Clifford et al., 2017, these three best performing models scored 0.80, 0.83, 0.82, in terms of F1-score for 'AFib' detection, respectively.
Despite using significantly fewer features than the competition's state-of-the-art ECG detection algorithms (48 vs. 150-622), our model achieved a comparable F1-score of 0.83 for successful 'AFib' detection.
The interpretable features specifically designed for 'AFib' detection results in our method's adaptability in clinical settings for real-time arrhythmia detection and is even effective with short ECGs (<10 heartbeats).
我们提出了一种通过分析导联心电图（ECG）来检测心房颤动（AFib）的新方法，该方法使用了一组独特的特征。
为此，我们使用了特定的信号处理技术，如适当的正交分解、连续小波变换、离散余弦变换和标准互相关，从 ECG 中提取 48 个特征。因此，我们的方法旨在比传统指标更有效地捕捉 AFib 特征，如心跳间变异性和纤维性波。此外，我们的特征旨在具有生理可解释性。随后，我们结合了基于 XGBoost 的 ECG 分类器，使用 2017 年 PhysioNet 挑战赛提供的公开“Training”数据集对其进行训练和评估，该数据集包括“正常”、“AFib”、“其他”和“嘈杂”ECG 类别。
我们的方法在“AFib”检测中达到了 96%的准确率和 0.83 的 F1 分数，在“正常”ECG 中达到了 80%的准确率和 0.85 的 F1 分数。最后，我们将我们的方法与 2017 年 PhysioNet 挑战赛的顶级分类器进行了比较，即 ENCASE、随机森林和级联二进制。正如 Clifford 等人 2017 年所报道的那样，这三个表现最好的模型在“AFib”检测方面的 F1 分数分别为 0.80、0.83 和 0.82。
尽管我们的模型使用的特征明显少于竞赛的最先进的 ECG 检测算法（48 个 vs. 150-622 个），但我们的模型在成功“AFib”检测方面的 F1 分数达到了 0.83，这表明我们的模型具有可比性。
专门设计用于“AFib”检测的可解释特征使得我们的方法能够在临床环境中适应实时心律失常检测，即使在短 ECG（<10 个心跳）的情况下也有效。
