电子测量与仪器学报

周帆,赵莉娜,李钰雯,李建清,刘澄玉.房颤智能检测中的心电特征选择和机器学习[J].电子测量与仪器学报,2021,35(3):1-10

房颤智能检测中的心电特征选择和机器学习

ECG feature selection and machine learning in intelligent detection of atrial fibrillation

DOI：

英文关键词:electrocardiogram atrial fibrillation feature selection machine learning

基金项目:国家自然科学基金(81871444，62001111)、江苏省自然科学基金(BK20200364)、中央高校基本科研业务费专项资金(2242020K40140)项目资助

作者	单位
周帆	1.东南大学仪器科学与工程学院生物电子学国家重点实验室南京210096；
赵莉娜	1.东南大学仪器科学与工程学院生物电子学国家重点实验室南京210096；
李钰雯	1.东南大学仪器科学与工程学院生物电子学国家重点实验室南京210096；
李建清	1.东南大学仪器科学与工程学院生物电子学国家重点实验室南京210096；2.南京医科大学生物医学工程与信息学院南京211166
刘澄玉	1.东南大学仪器科学与工程学院生物电子学国家重点实验室南京210096；

Author	Institution
Zhou Fan	1.State Key Laboratory of Bioelectronics, School of Instrument Science and Engineering, Southeast University, Nanjing210096, China;
Zhao Lina	1.State Key Laboratory of Bioelectronics, School of Instrument Science and Engineering, Southeast University, Nanjing210096, China;
Li Yuwen	1.State Key Laboratory of Bioelectronics, School of Instrument Science and Engineering, Southeast University, Nanjing210096, China;
Li Jianqing	1.State Key Laboratory of Bioelectronics, School of Instrument Science and Engineering, Southeast University, Nanjing210096, China; 2.School of Biomedical Engineering and Informatics, Nanjing Medical University, Nanjing 211166, China
Liu Chengyu	1.State Key Laboratory of Bioelectronics, School of Instrument Science and Engineering, Southeast University, Nanjing210096, China;

摘要点击次数: 1517

全文下载次数: 1669

中文摘要:

房颤是最常见的心律失常疾病，因其临床诊断率低而促进了实时自动检测算法的发展。但大多算法缺乏模型或数据库间的对比，难以评判模型的优劣。为此考虑选择3种机器学习算法（支持向量机、随机森林、逻辑回归）构建3个独立的房颤检测模型。3种模型分别在MIT BIH房颤数据库上训练，并在3个独立数据库上进行测试和对比，同时进一步分析特征选择对模型性能的影响。结果表明选择12个特征（3个时域特征和9个非线性特征）时，3种模型在2018年中国生理挑战赛公开数据集和可穿戴式动态心电数据集上的灵敏度、特异度、准确性和F1分数均达到95%以上，且随机森林相较于另两种算法具有更强的稳定性和泛化能力。

英文摘要:

Atrial fibrillation is the most common cardiac arrhythmia in clinical practice, while some real time automatic detection algorithms have been developed rapidly for improving clinical diagnosis. However, it is difficult to judge the advantages and disadvantages of the existing atrial fibrillation detection algorithms for their lack of comparisons between models or databases. Three different machine learning algorithms, including support vector machine, random forest and logistic regression, were selected to build three separate models to detect atrial fibrillation. The three models were trained on the MIT BIH atrial fibrillation database, and were tested and compared on three independent databases respectively. Furthermore, the influence of feature selection on model performance was analyzed. Experimental results showed that when applied 12 features (3 domain features and 9 nonlinear features), the sensitivity, specificity, accuracy and F1 score of the three models reached more than 95% on the China physiological signal challenge 2018 public database and the wearable dynamic ECG database. In addition, the random forest algorithm has stronger stability and generalization ability compared with the other two algorithms.

查看全文查看/发表评论下载PDF阅读器