Department of Cardiology, Charité University Medicine Berlin, Campus Virchow Klinikum (CVK), Berlin, Germany. Austrian Institute of Technology (AIT), Digital Health Information Systems, Graz, Austria. Department of Cardiology, Deutsches Herzzentrum Berlin (DHZB), Berlin, Germany. Graz University of Technology, Institute of Neural Engineering, Graz, Austria.
Physiol Meas. 2018 Oct 30;39(11):114001. doi: 10.1088/1361-6579/aae13e.
Recent advantages in mHealth-enabled ECG recorders boosted the demand for algorithms, which are able to automatically detect cardiac anomalies with high accuracy.
We present a combined method of classical signal analysis and machine learning which has been developed during the Computing in Cardiology Challenge (CinC) 2017. Almost 400 hand-crafted features have been developed to reflect the complex physiology of cardiac arrhythmias and their appearance in single-channel ECG recordings. For the scope of this article, we performed several experiments on the publicly available challenge dataset to improve the classification accuracy. We compared the performance of two tree-based algorithms-gradient boosted trees and random forests-using different parameters for learning. We assessed the influence of five different sets of training annotations on the classifiers performance. Further, we present a new web-based ECG viewer to review and correct the training labels of a signal data set. Moreover, we analysed the feature importance and evaluated the model performance when using only a subset of the features. The primary data source used in the analysis was the dataset of the CinC 2017, consisting of 8528 signals from four classes. Our best results were achieved using a gradient boosted tree model which worked significantly better than random forests.
Official results of the challenge follow-up phase provided by the Challenge organizers on the full hidden test set are 90.8% (Normal), 84.1% (AF), 74.5% (Other), resulting in a mean F1-score of 83.2%, which was only 1.6% behind the challenge winner and 0.2% ahead of the next-best algorithm. Official results were rounded to two decimal places which lead to the equal-second best F1 -score of 83% with five others.
The algorithm achieved the second-best score among 80 algorithms of the Challenge follow-up phase equal with five others.
移动医疗心电图记录仪的最新进展提高了对算法的需求，这些算法能够以高精度自动检测心脏异常。
我们提出了一种结合经典信号分析和机器学习的方法，该方法是在 2017 年计算心脏病学挑战赛（CinC）中开发的。已经开发了近 400 个手工制作的特征，以反映心律失常的复杂生理学及其在单通道心电图记录中的出现。在本文的范围内，我们使用公开可用的挑战数据集进行了几次实验，以提高分类准确性。我们比较了两种基于树的算法（梯度提升树和随机森林）的性能，这些算法使用不同的学习参数。我们评估了五种不同的训练注释集对分类器性能的影响。此外，我们提出了一种新的基于网络的 ECG 查看器，用于审查和纠正信号数据集的训练标签。此外，我们分析了特征重要性，并评估了仅使用特征子集时的模型性能。分析中使用的主要数据源是 CinC 2017 数据集，该数据集由四个类别的 8528 个信号组成。我们使用梯度提升树模型获得了最佳结果，该模型的性能明显优于随机森林。
挑战赛组织者在完整隐藏测试集上提供的挑战赛后续阶段的官方结果是正常 90.8%、AF 84.1%、其他 74.5%，平均 F1 得分为 83.2%，仅比挑战赛冠军低 1.6%，比下一个最佳算法高 0.2%。官方结果四舍五入到小数点后两位，导致与其他五人并列第二的最佳 F1 得分为 83%。
该算法在挑战赛后续阶段的 80 种算法中获得了第二名，与其他五人并列。
