Assessing Machine Learning for Diagnostic Classification of Hypertension Types Identified by Ambulatory Blood Pressure Monitoring

IF 2.5 Q2 CARDIAC & CARDIOVASCULAR SYSTEMS CJC Open Pub Date : 2024-06-01 DOI:10.1016/j.cjco.2024.03.005

Tran Quoc Bao Tran MSc , Stefanie Lip MBChB , Clea du Toit MSc , Tejas Kumar Kalaria MRCP , Ravi K. Bhaskar MS , Alison Q. O’Neil EngD , Beata Graff MD, PhD , Michał Hoffmann MD, PhD , Anna Szyndler MD, PhD , Katarzyna Polonis PhD , Jacek Wolf MD, PhD , Sandeep Reddy MBBS, PhD , Krzysztof Narkiewicz MD, PhD , Indranil Dasgupta DM , Anna F. Dominiczak MD, FMedSci , Shyam Visweswaran MD, PhD , Linsay McCallum PhD , Sandosh Padmanabhan MD, PhD

{"title":"Assessing Machine Learning for Diagnostic Classification of Hypertension Types Identified by Ambulatory Blood Pressure Monitoring","authors":"Tran Quoc Bao Tran MSc , Stefanie Lip MBChB , Clea du Toit MSc , Tejas Kumar Kalaria MRCP , Ravi K. Bhaskar MS , Alison Q. O’Neil EngD , Beata Graff MD, PhD , Michał Hoffmann MD, PhD , Anna Szyndler MD, PhD , Katarzyna Polonis PhD , Jacek Wolf MD, PhD , Sandeep Reddy MBBS, PhD , Krzysztof Narkiewicz MD, PhD , Indranil Dasgupta DM , Anna F. Dominiczak MD, FMedSci , Shyam Visweswaran MD, PhD , Linsay McCallum PhD , Sandosh Padmanabhan MD, PhD","doi":"10.1016/j.cjco.2024.03.005","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><p>Inaccurate blood pressure (BP) classification results in inappropriate treatment. We tested whether machine learning (ML), using routine clinical data, can serve as a reliable alternative to ambulatory BP monitoring (ABPM) in classifying BP status.</p></div><div><h3>Methods</h3><p>This study employed a multicentre approach involving 3 derivation cohorts from Glasgow, Gdańsk, and Birmingham, and a fourth independent evaluation cohort. ML models were trained using office BP, ABPM, and clinical, laboratory, and demographic data, collected from patients referred for hypertension assessment. Seven ML algorithms were trained to classify patients into 5 groups, named as follows: Normal/Target; Hypertension-Masked; Normal/Target-White-Coat (WC); Hypertension-WC; and Hypertension. The 10-year cardiovascular outcomes and 27-year all-cause mortality risks were calculated for the ML-derived groups using the Cox proportional hazards model.</p></div><div><h3>Results</h3><p>Overall, extreme gradient boosting (using XGBoost open source software) showed the highest area under the receiver operating characteristic curve of 0.85-0.88 across derivation cohorts, Glasgow (n = 923; 43% female; age 50.7 ± 16.3 years), Gdańsk (n = 709; 46% female; age 54.4 ± 13 years), and Birmingham (n = 1222; 56% female; age 55.7 ± 14 years). But accuracy (0.57-0.72) and F1 (harmonic mean of precision and recall) scores (0.57-0.69) were low across the 3 patient cohorts. The evaluation cohort (n = 6213; 51% female; age 51.2 ± 10.8 years) indicated elevated 10-year risks of composite cardiovascular events in the Normal/Target-WC and the Hypertension-WC groups, with heightened 27-year all-cause mortality observed in all groups, except the Hypertension-Masked group, compared to the Normal/Target group.</p></div><div><h3>Conclusions</h3><p>ML has limited potential in accurate BP classification when ABPM is unavailable. Larger studies including diverse patient groups and different resource settings are warranted.</p></div>","PeriodicalId":36924,"journal":{"name":"CJC Open","volume":"6 6","pages":"Pages 798-804"},"PeriodicalIF":2.5000,"publicationDate":"2024-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2589790X24001367/pdfft?md5=f906baaf76b4cab310f0b4114739014a&pid=1-s2.0-S2589790X24001367-main.pdf","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"CJC Open","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2589790X24001367","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

Inaccurate blood pressure (BP) classification results in inappropriate treatment. We tested whether machine learning (ML), using routine clinical data, can serve as a reliable alternative to ambulatory BP monitoring (ABPM) in classifying BP status.

Methods

This study employed a multicentre approach involving 3 derivation cohorts from Glasgow, Gdańsk, and Birmingham, and a fourth independent evaluation cohort. ML models were trained using office BP, ABPM, and clinical, laboratory, and demographic data, collected from patients referred for hypertension assessment. Seven ML algorithms were trained to classify patients into 5 groups, named as follows: Normal/Target; Hypertension-Masked; Normal/Target-White-Coat (WC); Hypertension-WC; and Hypertension. The 10-year cardiovascular outcomes and 27-year all-cause mortality risks were calculated for the ML-derived groups using the Cox proportional hazards model.

Results

Overall, extreme gradient boosting (using XGBoost open source software) showed the highest area under the receiver operating characteristic curve of 0.85-0.88 across derivation cohorts, Glasgow (n = 923; 43% female; age 50.7 ± 16.3 years), Gdańsk (n = 709; 46% female; age 54.4 ± 13 years), and Birmingham (n = 1222; 56% female; age 55.7 ± 14 years). But accuracy (0.57-0.72) and F1 (harmonic mean of precision and recall) scores (0.57-0.69) were low across the 3 patient cohorts. The evaluation cohort (n = 6213; 51% female; age 51.2 ± 10.8 years) indicated elevated 10-year risks of composite cardiovascular events in the Normal/Target-WC and the Hypertension-WC groups, with heightened 27-year all-cause mortality observed in all groups, except the Hypertension-Masked group, compared to the Normal/Target group.

Conclusions

ML has limited potential in accurate BP classification when ABPM is unavailable. Larger studies including diverse patient groups and different resource settings are warranted.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

评估通过动态血压监测对高血压类型进行诊断分类的机器学习方法

背景不准确的血压（BP）分类会导致不恰当的治疗。我们测试了使用常规临床数据进行机器学习（ML）是否可作为非卧床血压监测（ABPM）的可靠替代方法来对血压状态进行分类。方法这项研究采用了多中心方法，包括格拉斯哥、格但斯克和伯明翰的三个衍生队列以及第四个独立评估队列。使用从转诊进行高血压评估的患者处收集的诊室血压、ABPM 以及临床、实验室和人口统计学数据对 ML 模型进行了训练。训练了七种 ML 算法，将患者分为以下 5 组：正常/目标；高血压-掩蔽；正常/目标-白大褂（WC）；高血压-白大褂；高血压。使用 Cox 比例危险度模型计算了 ML 衍生组的 10 年心血管结局和 27 年全因死亡率风险。结果总体而言，极梯度增强（使用 XGBoost 开放源码软件）在所有衍生组中显示出最高的接收器操作特征曲线下面积（0.85-0.88）。在格拉斯哥（n = 923；43% 为女性；年龄为 50.7 ± 16.3 岁）、格但斯克（n = 709；46% 为女性；年龄为 54.4 ± 13 岁）和伯明翰（n = 1222；56% 为女性；年龄为 55.7 ± 14 岁）的衍生队列中，接受者操作特征曲线下面积最高，为 0.85-0.88。但 3 个患者群的准确度（0.57-0.72）和 F1（准确度和召回率的调和平均值）得分（0.57-0.69）都很低。评估队列（n = 6213；51% 女性；年龄 51.2 ± 10.8 岁）显示，与正常/目标组相比，正常/目标-WC 组和高血压-WC 组的 10 年复合心血管事件风险升高，除高血压-掩蔽组外，所有组的 27 年全因死亡率均升高。在没有 ABPM 的情况下，MML 在准确血压分类方面的潜力有限。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助

来源期刊