校准
医学
脉搏波分析
脉冲波速
脉冲波
血压
计算机科学
电信
统计
内科学
数学
抖动
作者
Ramakrishna Mukkamala,Sanjeev G. Shroff,Konstantinos G. Kyriakoulis,Alberto Avolio,George S. Stergiou
出处
期刊:Hypertension
[Lippincott Williams & Wilkins]
日期:2025-04-15
被引量:2
标识
DOI:10.1161/hypertensionaha.125.24822
摘要
Cuffless blood pressure (BP) measurement offers considerable potential for clinical practice but is a challenging technological field. Many are investigating pulse wave analysis with or without pulse arrival time in which machine learning is applied to pulsatile waveforms obtained with mobile devices (eg, wristbands, smartphones) to estimate BP. These methods generally require individual user calibration with cuff BP measurements or demographics (eg, age, sex). This calibration makes it difficult to evaluate the method’s accuracy, and many studies claiming accuracy used inadequate testing procedures. Yet, publications and regulatory-cleared devices continue to rise, seemingly implying technological advancements. An update is provided on the flurry of activity in cuffless BP technologies over the last 2 to 3 years, covering the clinical need, the latest devices, recent publications based on pulse wave analysis and pulse arrival time, progress in developing validation standards for cuffless BP devices, and recent publications on other cuffless BP measurement principles. Despite the high volume of research and development, to date, there is no compelling evidence that pulse wave analysis and pulse arrival time can provide significant added value in BP measurement accuracy beyond the cuff BP or demographic data for calibration. Thus, it is reasonable to at least be skeptical of published and future studies on pulse wave analysis and pulse arrival time for cuffless BP measurement with uncertain testing procedures. It is important to focus on establishing robust validation standards for cuffless BP devices requiring individual user calibration and also pursuing cuffless and calibration-free BP measurement methodologies going forward.
科研通智能强力驱动
Strongly Powered by AbleSci AI