A Deep Bottleneck Feature Extraction Method of Cardiac Shock Signal Based on Deep Neural Network

Abstract

The invention relates to the technical field of biological feature extraction, and is a method for extracting deep bottleneck features of heart shock signals based on deep neural networks. The heart shock signal is used as the feature extraction object, and its waveform amplitude is weak and easily subject to external interference. Synchronously collected ECG signals, using the deep neural network to mine the mechanism of deep features, and extract its deep bottleneck feature parameters. This feature uses the cardiac shock signal as the input vector, and the synchronous ECG signal as the target vector. It is trained through a pre-designed 9-layer neural network to obtain deep bottleneck features and realize the organic combination of cardiac dynamic performance and electrophysiological features. This feature takes cardiac shock signals and ECG signals, which are easier to obtain in daily life, as the research object. It can not only overcome the dependence of conventional waveform characteristic parameters on waveform fluctuations, but also improve the representation performance of a single characteristic parameter. It is an application of deep learning. A new attempt to theoretically conduct daily cardiac function analysis.

Description

Technical field [0001] The present invention relates to the field of biometric extraction techniques, and more particularly to an extraction method of deep bottleneck characteristics based on a deep neural network based on a deep neural network. Background technique [0002] With the increasing popularity of wearable equipment, the field of non-invasive heart function assessment has become a hot spot today. Conventional cardiac function detection means, including electrocardiograph (ECG), heart magnetograph, heart sound, heart impedance map, etc., all need to detect the electrode and other detection equipment such as human body table, have certain requirements for monitoring environment, conditions, and operators And there is a great inconvenience to the daily life of the subject. Ballistocardiogram, BCG) By detecting the weak vibration of cardiac pulsation, indirectly reflects the working conditions of the heart, is an advanced testing means without a human cardiac kinetic perfo...

AI Technical Summary

Problems solved by technology

Since the signal itself is relatively weak and easily interfered by the external environment, the extraction of conventional waveform characteristic parameters has certain limitations.

Images