Continuous physiological signal quality assessment method

Abstract

The invention discloses a continuous physiological signal quality evaluation method. The method comprises the following steps: segmenting an original physiological signal through windowing; preprocessing the segmented signal, wherein the preprocessing comprises signal baseline removal, band-pass filtering and median filtering; carrying out feature value extraction on the preprocessed signal; inputting extracted feature values into a pre-training model to obtain a score; and judging the score according to set threshold values L1 and L2 to obtain an evaluation result of the signal quality, wherein the pre-training model is an isolated forest model.

Description

technical field [0001] The present application relates to the detection of physiological signals, in particular to the quality assessment of physiological signals. Background technique [0002] Continuous physiological signals refer to continuously recorded signals such as ECG, respiration, pulse, etc. that are highly related to human health and disease states. At present, they are more and more valuable in disease diagnosis, adverse event prediction, human body state assessment, and rehabilitation prognosis evaluation. Recognized. For example, the deterioration of some clinical diseases is usually manifested in physiological signals 8-24 hours before the occurrence of more serious clinical outcomes (such as sudden cardiac death), and clinicians and researchers in this field are increasingly aware of To the importance of continuous monitoring and deep mining analysis of physiological signals. [0003] In recent years, a variety of measurement methods and sensors have been ...

AI Technical Summary

Problems solved by technology

[0006] However, the above methods have strong limitations: First, the signals in the early research / patents mostly come from bedside monitors, and the collected signals are fundamentally different from real-world signals. The problems faced by wearable devices are different; secondly, despite the rapid development of high-performance machine learning / deep learning models, most of these methods require a large number of labels to complete the model learning steps, the number of labels is insufficient or the model training samples are too limited. Strong uniformity will lead to insufficient generalization performance of the model, making it difficult to meet the needs of use

The problem is that it is relatively easy to obtain a large amount of data, but it takes a lot of manpower and time to label all the data. The rapid iteration of hardware and software of wearable devices has made the task of labeling even more serious.

Images