Deep learning-based T-wave electrical alternation detection method

Abstract

The invention discloses a deep learning-based T-wave electrical alternation detection method. The characteristics of T-wave electrical alternation is further excavated by using a deep learning method, ECG waveform data containing T-wave electrical alternation can be detected effectively by using the deep learning-based T-wave electrical alternation detection method, and the effect of TWA can be analyzed and detected quantitatively by using an F1 value. The deep learning-based T-wave electrical alternation detection method has a promotion effect on TWA identification research. According to the deep learning-based T-wave electrical alternation detection method, the problems that the detection of T-wave electrical alternation still lacks a medical gold standard and the detection accuracy cannot be quantitatively analyzed are solved.

Description

technical field [0001] The invention relates to the technical field of electrocardiographic signals, in particular to a method for detecting T-wave alternation based on deep learning Background technique [0002] The existing detection methods of T-wave alternation (TWA) include spectrum analysis method, time-domain analysis method, moving average correction method, Laplace likelihood ratio and some similar improved methods. The accuracy of waveform positioning in the traditional TWA detection method will affect the TWA detection effect, and since there is no accurate standard for TWA to refer to, the detection effect of various methods cannot be accurately measured. [0003] Deep learning has made significant progress in recent years, and it has been proven to be able to mine complex structures in high-dimensional data for learning. Advantages of deep learning: 1. Judging from the results, the performance of deep learning is very good, and the learning ability is very stro...

AI Technical Summary

Problems solved by technology

Disadvantages of deep learning: 1. Deep learning requires a large amount of data for training

2. Deep learning requires high computing power, and ordinary CPUs can no longer meet the requirements of deep learning

The mainstream computing power is in GPU and TPU, which requires high hardware and high cost

3. Deep learning model design is complex

4. Since deep learning relies on data, the interpretability is not high

Images