Anomaly detection method and system for intelligent biological signals

Abstract

The invention relates to an intelligent biological signal anomaly detection method and a system capable of carrying out efficient, accurate and low-cost automatic analysis on biological signals. The system comprises a model training stage and a model application stage, wherein the model training stage comprises the following steps: acquiring a biological signal; preprocessing the acquired signal;sending the preprocessed data to a CNN-BilSTM-Attention combination model to extract features, wherein the signal firstly enters a CNN convolution neural network to extract the spatial features, thenenters a two-way long-short time memory network to extract time features, then passes through each network connection layer and finally passes through a SoftMax layer to calculate the final result, each link of the network is established, and the emphasis degree of different sequence points of signals are optimized by setting an Attention mechanism; calculating loss of the network module until themodule loss is less than the threshold, then the model training is completed, and otherwise Adam is used to optimize model parameters. The invention effectively improves the analysis effect of biomedical signals.

Description

Technical field: [0001] The invention relates to the technical field of biological signal detection, in particular to an intelligent biological signal anomaly detection method and system capable of performing efficient, accurate and low-cost automatic analysis on biological signals. Background technique: [0002] All living organisms that exist in nature, whether animals or humans, have their own vital signs. With the development of economy and society, these vital signs can be collected in the form of signal sequences through existing instruments. These biological The signal is a partial mapping of the physiological function of the living body, which can effectively detect the physiological state and physiological function of the living body. For humans, in recent years, people's pace of life has accelerated, work pressure has increased, and the morbidity and mortality of cardiovascular and cerebrovascular diseases have increased. People's bodies are being seriously threate...

AI Technical Summary

Problems solved by technology

[0011] 3. For signal processing and analysis, fragmented processing and analysis is usually adopted, but for biological signals, sometimes it is more important to monitor continuously for a long time and effectively respond to changing trends. The continuous monitoring of signals under the current method There is very little work done with analysis, and the continuous monitoring of signals has been realized on the acquisition instrument, and the obtained information has not been fully utilized

[0013] 1. The artificially extracted signal features can be used to identify signals using traditional machine learning methods, such as support vector machines, multi-layer perceptrons, decision trees, random forests, etc. Although these algorithms have strong mathematical interpretability, However, the requirements for signal features are very high, and experienced people are required to effectively extract features to make the recognition accuracy meet a basic requirement. This method is time-consuming and laborious, and cannot have a good generalization ability.

[0014] 2. The morphological characteristics of a certain aspect of the signal are difficult to reflect all the characteristic information of the signal sequence. For a specific classification task, the manually extracted features may not be able to meet the needs of machine intelligence learning this classification task. For example, for some small changes , imperceptible abnormal heart beats, the intelligent algorithm may lose its role

[0015] 3. For biological signal classification tasks, in order to make up for the shortcomings of traditional machine learning methods, many deep learning frameworks have emerged in recent years, such as CNN, RNN, LSTM, SAE, etc., which can automatically extract features from signal sequences, but The recognition accuracy of the deep learning framework algorithms of these network structures needs to be improved, and the interpretability of signal classification ideas is poor

Images