{"title":"Electrocardiographic parameter profiles for differentiating hypertrophic cardiomyopathy stages","authors":"Naomi Hirota MD, PhD, Shinya Suzuki MD, PhD, Takuto Arita MD, Naoharu Yagi MD, Mikio Kishi MD, Hiroaki Semba MD, PhD, Hiroto Kano MD, Shunsuke Matsuno MD, Yuko Kato MD, PhD, Takayuki Otsuka MD, PhD, Junji Yajima MD, PhD, Tokuhisa Uejima MD, PhD, Yuji Oikawa MD, PhD, Takeshi Yamashita MD, PhD","doi":"10.1002/joa3.70031","DOIUrl":null,"url":null,"abstract":"<div>\n \n \n <section>\n \n <h3> Background</h3>\n \n <p>The efficacy of artificial intelligence (AI)-enhanced electrocardiography (ECG) for detecting hypertrophic cardiomyopathy (HCM) and its dilated phase (dHCM) has been developed, though specific ECG characteristics associated with these conditions remain insufficiently characterized.</p>\n </section>\n \n <section>\n \n <h3> Methods</h3>\n \n <p>This retrospective study included 19,170 patients, with 140 HCM or dHCM cases, from the Shinken Database (2010–2017). The 140 cases (HCM-total) were categorized into basal-only HCM (HCM-basal, <i>n</i> = 75), apical involvement (HCM-apical, <i>n</i> = 46), and dHCM (<i>n</i> = 19). We analyzed 438 ECG parameters across the P-wave (110), QRS complex (194), and ST-T segment (134). High parameter importance (HPI) was defined as 1/<i>p</i> > 10<sup>4</sup> in univariate logistic regression, while multivariate logistic regression was used to determine the area under the receiver operating characteristic curves (AUROC).</p>\n </section>\n \n <section>\n \n <h3> Results</h3>\n \n <p>In HCM-basal and HCM-apical, HPI was predominantly observed in the ST-T segment (49% and 51%, respectively), followed by the QRS complex (29% and 27%). For dHCM, HPI was lower in the ST-T segment (16%) and QRS complex (22%). The P-wave had low HPI across all subtypes. AUROCs for models with total ECG parameters were 0.925 for HCM-basal, 0.981 for HCM-apical, and 0.969 for dHCM. While AUROCs for the top 10 HPI models were lower than the total ECG parameter model for HCM total, they were comparable across HCM subtypes.</p>\n </section>\n \n <section>\n \n <h3> Conclusions</h3>\n \n <p>As HCM progresses to dHCM, a shift in HPI from the ST-T segment to the QRS complex provides clinically relevant insights. For HCM subtypes, the top 10 ECG parameters yield predictive performance similar to the full parameter set, supporting efficient approaches for AI-based diagnostic models.</p>\n </section>\n </div>","PeriodicalId":15174,"journal":{"name":"Journal of Arrhythmia","volume":"41 2","pages":""},"PeriodicalIF":2.2000,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/joa3.70031","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Arrhythmia","FirstCategoryId":"1085","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1002/joa3.70031","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"CARDIAC & CARDIOVASCULAR SYSTEMS","Score":null,"Total":0}
引用次数: 0
Abstract
Background
The efficacy of artificial intelligence (AI)-enhanced electrocardiography (ECG) for detecting hypertrophic cardiomyopathy (HCM) and its dilated phase (dHCM) has been developed, though specific ECG characteristics associated with these conditions remain insufficiently characterized.
Methods
This retrospective study included 19,170 patients, with 140 HCM or dHCM cases, from the Shinken Database (2010–2017). The 140 cases (HCM-total) were categorized into basal-only HCM (HCM-basal, n = 75), apical involvement (HCM-apical, n = 46), and dHCM (n = 19). We analyzed 438 ECG parameters across the P-wave (110), QRS complex (194), and ST-T segment (134). High parameter importance (HPI) was defined as 1/p > 104 in univariate logistic regression, while multivariate logistic regression was used to determine the area under the receiver operating characteristic curves (AUROC).
Results
In HCM-basal and HCM-apical, HPI was predominantly observed in the ST-T segment (49% and 51%, respectively), followed by the QRS complex (29% and 27%). For dHCM, HPI was lower in the ST-T segment (16%) and QRS complex (22%). The P-wave had low HPI across all subtypes. AUROCs for models with total ECG parameters were 0.925 for HCM-basal, 0.981 for HCM-apical, and 0.969 for dHCM. While AUROCs for the top 10 HPI models were lower than the total ECG parameter model for HCM total, they were comparable across HCM subtypes.
Conclusions
As HCM progresses to dHCM, a shift in HPI from the ST-T segment to the QRS complex provides clinically relevant insights. For HCM subtypes, the top 10 ECG parameters yield predictive performance similar to the full parameter set, supporting efficient approaches for AI-based diagnostic models.