Heart rate fusion labeling method and system based on Bayesian prior probability

Abstract

The invention discloses a heart rate fusion labeling method and system based on Bayesian prior probability; reading, preprocessing and signal segmentation are performed on initial electrocardiosignals to obtain electrocardiograph data samples, and feature extraction is performed on the electrocardiograph data samples; according to different types of initial electrocardiosignal sets, a single-lead database or a multi-lead database, a proper heart rate labeling fusion process based on Bayesian prior probability is developed. In addition, probability estimation and iterative solution are carried out on the heart rate tags marked by multiple leads or multiple algorithms through a Bayesian criterion and an expectation maximization algorithm, and a heart rate tag value with higher precision is obtained through fusion. The fusion model automatically knows potential differences among different labeled samples; due to the fact that a label generated by a single algorithm or a lead is possibly unreliable, a label value with higher dependency is obtained through a fusion model, the accuracy of long-term dynamic heart rate estimation is improved, and more accurate information is provided for diagnosis of clinical cardiovascular diseases.

Description

technical field [0001] The invention belongs to the technical field of signal detection and medical equipment electronics, and in particular relates to a heart rate fusion labeling method and system based on Bayesian prior probability. Background technique [0002] Cardiovascular disease is the main cause of death worldwide, and the detection of ECG signals is one of the basic methods for detecting and diagnosing cardiovascular diseases. Therefore, the real-time monitoring and intelligent labeling and analysis of ECG signals have received great attention. In order to achieve this goal, the automatic labeling of the QRS wave and heart rate parameters of ECG signals must be realized first, because they are not only the research basis for intelligent diagnosis of cardiovascular diseases, but also the target parameters monitored by ECG instruments. [0003] Sustained high accuracy is the key to the practical application of Holter's heart rate detection algorithm. In fact, most o...

AI Technical Summary

Problems solved by technology

[0003] Sustained high accuracy is the key to the practical application of Holter's heart rate detection algorithm. In fact, most of the classic and commonly used QRS detection algorithms have high detection accuracy in high-signal quality databases, but they are applied to noise-intensive When the dynamic ECG database is large, the detection accuracy is greatly reduced. The reason for its poor robustness may be that some parameters in the algorithm are empirical values ​​or variables related to the detected signal, such as the window size and threshold of the sliding window Coefficients, etc., these parameters may be applicable to a specific ECG database, not necessarily applicable to other databases, especially the noisy ECG database

However, during the acquisition process of dynamic ECG, it is more likely to be affected by human motion to generate unavoidable noise. How to improve the QRS detection effect of these noisy databases is a current research hotspot.

For ECG signals collected under different environments and different individual characteristics, it is difficult for a single detection algorithm or a single lead detection to always achieve high accuracy. Therefore, when the true value of the annotation is unknown, how to provide The key to the problem is to find consensus among the noisy labels, make up for the gaps, and generate labels that are closer to the true value. This invention proposes a heart rate label fusion method and system based on Bayesian prior probability, which will improve the existing Heart rate labeling method and system Improve the accuracy of heart rate labeling and improve the heart rate labeling method to achieve major improvements and breakthroughs

Images