计算机辅助诊断模型助力基层医疗机构诊断阵发性心房颤动

Computer-aided Diagnosis Model Assists Paroxysmal Atrial Fibrillation Diagnosis in Primary Care

心房颤动(简称房颤)是一种可导致多种严重并发症的心律失常。其中,阵发性房颤具有阵发性和无症状性特点,故难以诊断。长程心电所获得心电大数据可以提高阵发性房颤的检出率。但心电大数据的判读却成为基层医疗机构的负担和难题。为解决以上问题,多种基于心电特征的浅层学习模型不断出现,这些模型高度依赖人工提取特征,均有局限性。深度学习是一种数据驱动的自动特征学习算法,弥补浅层学习的不足。Lorenz散点图作为心电大数据快速分析的新兴方法,其输出的二维图形是深度学习的优质素材。本文综述房颤计算机辅助诊断模型的心电特征研究进展以及机器学习在房颤诊断中的应用现状,为辅助诊断模型构建提供新思路,同时为解决基层心电大数据的判读难题提供新视角。

Atrial fibrillation is a common arrhythmia that can cause many serious complications.Among them,paroxysmal atrial fibrillation is difficult to diagnose due to paroxysmal and asymptomatic characteristics.Big ECG data obtained by long-term ECG can improve the detection rate of paroxysmal atrial fibrillation.However,the interpretation of ECG big data has become a burden and problem for primary medical institutions.To solve the problems,a variety of shallow learning models based on ECG features have been developed,which highly rely on manual feature extraction and have limitations.Deep learning is a data-driven automatic feature learning algorithm,which can make up for the shortcomings of shallow learning.As an emerging method for rapid analysis of ECG big data,the Lorenz scatterplots using two-dimensional graphs is high-quality materials for deep learning.This paper reviews the latest advances in ECG features of atrial fibrillation using computer-aided model for diagnosis,and the application of machine learning in atrial fibrillation diagnosis,providing new insights into the development of a good computer-aided diagnosis model,and a new perspective for the interpretation of ECG big data in primary care.