Automatic electrocardiogram analysis method and device based on artificial intelligence self-learning

Abstract

An embodiment of the invention relates to an automatic electrocardiogram analysis method and device based on artificial intelligence self-learning. The method includes: preprocessing data, detecting heart beat features, performing interference signal detection and heart beat classification based on a deep learning method, combining signal quality evaluation and leading, performing heart beat checking, performing analytical calculation of electrocardiogram events and parameters, and automatically outputting report data. The automatic electrocardiogram analysis method has the advantages that anintegral fast flow is achieved, the modification information of an automatic analysis result can be recorded, the modified data can be collected and fed back to a deep learning model to continue training, and the accuracy of the automatic electrocardiogram analysis method can be constantly improved.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence data analysis, in particular to an automatic electrocardiogram analysis method and device based on artificial intelligence self-learning. Background technique [0002] In 1908, Einthoven began to use electrocardiography (ECG) to monitor the electrophysiological activity of the heart. At present, non-invasive electrocardiography has become one of the important methods for the diagnosis and screening of heart diseases in the clinical cardiovascular field. According to the clinical application, electrocardiogram can be divided into several categories: static electrocardiogram, dynamic electrocardiogram, and exercise electrocardiogram. The static electrocardiogram adopts the 12-lead system (standard lead system) invented by Einthoven-Wilson-Goldberger, and records the electrocardiogram signal for 8-30 seconds for analysis. The earliest, most commonly used and most basic diagnostic met...

AI Technical Summary

Problems solved by technology

However, this method has some problems that need to be further solved: (a) the types of cardiac rhythm events identified are not perfect, and it does not specifically deal with low voltage, asystole, etc.; (b) the model is trained for single-lead Holter, but the actual In use, there are a large number of multi-lead ECG examinations, especially the standard 12-lead, 3-lead equipment commonly used in dynamic ECG, multi-lead analysis can more comprehensively and correctly analyze arrhythmia, conduction block and locate ST These functions cannot be reflected in single-lead model analysis; (c) Although this method considers noise, it only treats noise as a special cardiac rhythm event, which is not in line with medical laws. Analyze and identify the nature of the noise to eliminate the impact of the noise on the subsequent analysis; (d) This method does not have the ability to automatically generate reports, especially after analyzing all digital signals, automatically intercept the most typical and best signal quality fragments Generate an image report; (e) In addition, this method does not explain how to apply it to the whole process of 8-30 seconds static ECG analysis and 24-hour dynamic ECG analysis, and combine with existing technologies to improve the overall performance of clinical analysts. Work quality and efficiency, and these are the most common and urgent needs in ECG analysis

Images