Ji-Hoon Choi, Sung-Hee Song, Hongryul Kim, Jongwoo Kim, Heesun Park, JaeHu Jeon, JoongSik Hong, Hye Bin Gwag, Sung Ho Lee, Jaichan Lee, Soo Jin Cho, Seung-Jung Park, Young Keun On, Ju Youn Kim, Kyoung-Min Park
{"title":"使用基于左心房重塑的串行 12 导联心电图预测心房颤动的机器学习算法。","authors":"Ji-Hoon Choi, Sung-Hee Song, Hongryul Kim, Jongwoo Kim, Heesun Park, JaeHu Jeon, JoongSik Hong, Hye Bin Gwag, Sung Ho Lee, Jaichan Lee, Soo Jin Cho, Seung-Jung Park, Young Keun On, Ju Youn Kim, Kyoung-Min Park","doi":"10.1161/JAHA.123.034154","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>We hypothesized that analysis of serial ECGs could predict new-onset atrial fibrillation (AF) more accurately than analysis of a single ECG by detecting the subtle cardiac remodeling that occurs immediately before AF occurrence. Our aim in this study was to compare the performance of 2 types of machine learning (ML) algorithms.</p><p><strong>Methods and results: </strong>Standard 12-lead ECGs of patients selected by cardiologists between January 2010 and May 2021 were used for ML model development. Two ML models (single ECG and serial ECG) were developed using a light gradient boosting machine-learning algorithm. Model performance was evaluated based on the area under the receiver operating characteristic curve, sensitivity, specificity, accuracy, and F1 score. We trained the ML models on 415 964 ECGs from 176 090 patients. When testing the 2 ML models using external validation data sets, the performance of the serial-ML model was significantly better than that of the single-ML model for predicting new-onset AF (single- versus serial-ML model: sensitivity 0.744 versus 0.810; specificity 0.742 versus 0.822; accuracy 0.743 versus 0.816; F1 score 0.743 versus 0.815; area under the receiver operating characteristic curve 0.812 versus 0.880; <i>P</i><0.001). The Shapley Additive Explanations analysis ranked P-wave duration and amplitude among the top 10 ECG parameters.</p><p><strong>Conclusions: </strong>An ML model based on serial ECGs from an individual had greater ability to predict new-onset AF than the ML model based on a single ECG. P-wave morphologies were associated with future AF prediction.</p>","PeriodicalId":54370,"journal":{"name":"Journal of the American Heart Association","volume":null,"pages":null},"PeriodicalIF":5.0000,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Machine Learning Algorithm to Predict Atrial Fibrillation Using Serial 12-Lead ECGs Based on Left Atrial Remodeling.\",\"authors\":\"Ji-Hoon Choi, Sung-Hee Song, Hongryul Kim, Jongwoo Kim, Heesun Park, JaeHu Jeon, JoongSik Hong, Hye Bin Gwag, Sung Ho Lee, Jaichan Lee, Soo Jin Cho, Seung-Jung Park, Young Keun On, Ju Youn Kim, Kyoung-Min Park\",\"doi\":\"10.1161/JAHA.123.034154\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Background: </strong>We hypothesized that analysis of serial ECGs could predict new-onset atrial fibrillation (AF) more accurately than analysis of a single ECG by detecting the subtle cardiac remodeling that occurs immediately before AF occurrence. Our aim in this study was to compare the performance of 2 types of machine learning (ML) algorithms.</p><p><strong>Methods and results: </strong>Standard 12-lead ECGs of patients selected by cardiologists between January 2010 and May 2021 were used for ML model development. Two ML models (single ECG and serial ECG) were developed using a light gradient boosting machine-learning algorithm. Model performance was evaluated based on the area under the receiver operating characteristic curve, sensitivity, specificity, accuracy, and F1 score. We trained the ML models on 415 964 ECGs from 176 090 patients. When testing the 2 ML models using external validation data sets, the performance of the serial-ML model was significantly better than that of the single-ML model for predicting new-onset AF (single- versus serial-ML model: sensitivity 0.744 versus 0.810; specificity 0.742 versus 0.822; accuracy 0.743 versus 0.816; F1 score 0.743 versus 0.815; area under the receiver operating characteristic curve 0.812 versus 0.880; <i>P</i><0.001). The Shapley Additive Explanations analysis ranked P-wave duration and amplitude among the top 10 ECG parameters.</p><p><strong>Conclusions: </strong>An ML model based on serial ECGs from an individual had greater ability to predict new-onset AF than the ML model based on a single ECG. P-wave morphologies were associated with future AF prediction.</p>\",\"PeriodicalId\":54370,\"journal\":{\"name\":\"Journal of the American Heart Association\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":5.0000,\"publicationDate\":\"2024-10-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of the American Heart Association\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1161/JAHA.123.034154\",\"RegionNum\":1,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2024/9/30 0:00:00\",\"PubModel\":\"Epub\",\"JCR\":\"Q1\",\"JCRName\":\"CARDIAC & CARDIOVASCULAR SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of the American Heart Association","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1161/JAHA.123.034154","RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2024/9/30 0:00:00","PubModel":"Epub","JCR":"Q1","JCRName":"CARDIAC & CARDIOVASCULAR SYSTEMS","Score":null,"Total":0}
引用次数: 0
摘要
背景:我们假设,通过检测心房颤动发生前的微妙心脏重塑,分析连续心电图比分析单张心电图更能准确预测新发心房颤动(AF)。本研究旨在比较两种机器学习(ML)算法的性能:在开发 ML 模型时使用了心脏病专家在 2010 年 1 月至 2021 年 5 月期间选取的患者的标准 12 导联心电图。使用轻梯度提升机器学习算法开发了两种 ML 模型(单个心电图和序列心电图)。模型性能根据接收者操作特征曲线下面积、灵敏度、特异性、准确性和 F1 分数进行评估。我们在来自 176090 名患者的 415 964 张心电图上训练了 ML 模型。在使用外部验证数据集测试两种 ML 模型时,序列 ML 模型在预测新发房颤方面的性能明显优于单一 ML 模型(单一模型与序列 ML 模型相比:灵敏度为 0.744 对 0.810;特异性为 0.742 对 0.822;准确性为 0.743 对 0.816;F1 评分为 0.743 对 0.815;接收器操作特征曲线下面积为 0.812 对 0.880;PConclusions:与基于单张心电图的 ML 模型相比,基于个人连续心电图的 ML 模型预测新发房颤的能力更强。P波形态与未来房颤预测有关。
Machine Learning Algorithm to Predict Atrial Fibrillation Using Serial 12-Lead ECGs Based on Left Atrial Remodeling.
Background: We hypothesized that analysis of serial ECGs could predict new-onset atrial fibrillation (AF) more accurately than analysis of a single ECG by detecting the subtle cardiac remodeling that occurs immediately before AF occurrence. Our aim in this study was to compare the performance of 2 types of machine learning (ML) algorithms.
Methods and results: Standard 12-lead ECGs of patients selected by cardiologists between January 2010 and May 2021 were used for ML model development. Two ML models (single ECG and serial ECG) were developed using a light gradient boosting machine-learning algorithm. Model performance was evaluated based on the area under the receiver operating characteristic curve, sensitivity, specificity, accuracy, and F1 score. We trained the ML models on 415 964 ECGs from 176 090 patients. When testing the 2 ML models using external validation data sets, the performance of the serial-ML model was significantly better than that of the single-ML model for predicting new-onset AF (single- versus serial-ML model: sensitivity 0.744 versus 0.810; specificity 0.742 versus 0.822; accuracy 0.743 versus 0.816; F1 score 0.743 versus 0.815; area under the receiver operating characteristic curve 0.812 versus 0.880; P<0.001). The Shapley Additive Explanations analysis ranked P-wave duration and amplitude among the top 10 ECG parameters.
Conclusions: An ML model based on serial ECGs from an individual had greater ability to predict new-onset AF than the ML model based on a single ECG. P-wave morphologies were associated with future AF prediction.
期刊介绍:
As an Open Access journal, JAHA - Journal of the American Heart Association is rapidly and freely available, accelerating the translation of strong science into effective practice.
JAHA is an authoritative, peer-reviewed Open Access journal focusing on cardiovascular and cerebrovascular disease. JAHA provides a global forum for basic and clinical research and timely reviews on cardiovascular disease and stroke. As an Open Access journal, its content is free on publication to read, download, and share, accelerating the translation of strong science into effective practice.