Sleep rhythm disorder detection method and apparatus based on long-term signal and multi-feature extraction

Abstract

The present invention relates to a sleep rhythm disorder detection method and apparatus based on a long-term signal and multi-feature extraction. The method comprises: step S1, obtaining a physiological signal of a target user, wherein the physiological signal at least includes a heart rate; step S2, performing pre-processing and peak extraction on the physiological signal; step S3, on the basis of the pre-processed physiological signal, performing extraction, which involves: on the basis of the processed signal, using a multi-domain feature extraction algorithm to perform multi-domain heart rate variability feature extraction; step S4, performing multi-domain feature set construction on extracted multi-domain heart rate variability features and performing rhythm disorder risk category classification; and step S5, using a random forest algorithm to construct a sleep rhythm state assessment model, in order to detect a potential sleep rhythm disorder risk. Compared with the prior art, the present invention has the advantages of improving the accuracy, and also solving the problem of the workload required for frequent physiological signal collection, etc.

Description

A method and device for detecting sleep rhythm disorders based on long-range signals and multi-feature extraction Technical Field

[0001] This invention relates to the field of medical signal processing technology, and in particular to a method and apparatus for detecting sleep rhythm disorders based on long-range signals and multi-feature extraction. Background Technology

[0002] Human physiological activities follow certain periodic patterns, such as sleep, wakefulness, heart rate, mood, and cognition, exhibiting regular changes throughout the day and night. This circadian rhythm is the body's adaptive response to its environment, with sleep and wakefulness being the most fundamental rhythmic features among all physiological functional fluctuations. However, with societal development, the accelerated pace of life, and increased competitive pressure, prolonged stress can lead to sleep rhythm disorders. These disorders manifest as a mismatch between sleep and wakefulness times and the natural circadian rhythm, or as non-physiological sleep phenomena such as difficulty falling asleep, insufficient sleep, or disrupted sleep phases. Sleep rhythm disorders can seriously affect people's physical and mental health, work capacity, and quality of life, and are closely related to the onset of many acute and chronic diseases. Prolonged sleep rhythm disorders can lead to emotional instability, physical and mental fatigue, weakened immunity, and decreased attention and behavioral abilities. In high-risk industries, this can cause cognitive impairment such as impaired judgment and operational errors, potentially leading to accidental injuries or fatalities.

[0003] Currently, most existing medical devices and wearable smart devices for predicting the risk of sleep rhythm disorders only quantify sleep rhythms. Risk prediction often requires manual estimation and analysis, lacking efficient data processing and analysis capabilities. This prevents them from providing real-time and more accurate early warnings of rhythm disorder risks. Furthermore, existing methods fail to incorporate individual sleep rhythm types for personalized rhythm disorder assessments. In addition, the massive amounts of physiological signal data collected continuously pose significant challenges to signal processing, necessitating a method for predicting the risk of rhythm disorders under long-term data acquisition conditions. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for detecting sleep rhythm disorders based on long-range signals and multi-feature extraction.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for detecting sleep rhythm disorders based on long-range signals and multi-feature extraction, comprising:

[0007] Step S1: Obtain the physiological signals of the target user, wherein the physiological signals include at least heart rate;

[0008] Step S2: Preprocess and extract peak values ​​from the physiological signals;

[0009] Step S3: Extracting physiological signals based on preprocessed signals. Multi-domain heart rate variability features are extracted using a multi-domain feature extraction algorithm based on the preprocessed signals.

[0010] Step S4: Construct a multi-domain feature set and classify rhythm disorder risk categories based on the extracted multi-domain heart rate variability features;

[0011] Step S5: Construct a sleep rhythm state assessment model using the random forest algorithm to detect potential risks of sleep rhythm disorders.

[0012] The physiological signals are collected via wearable electronic devices.

[0013] The preprocessing process in step S2 involves using a bandpass filter to remove high-frequency and low-frequency noise from the acquired physiological signals.

[0014] The multi-domain heart rate variability features include HRV time-domain features, HRV frequency-domain features, and HRV nonlinear features. The HRV nonlinear features include heart rate asymmetry features, Pompeii diagram features, and various entropy complexities.

[0015] The Pompeii diagram features include the semi-major axis SD1, the semi-minor axis SD2, and the ratio of the major axis to the semi-minor axis SD1 / SD2;

[0016] The various entropy complexities include approximate entropy ApEn, sample entropy SampE, fuzzy entropy FuzzyEn, fuzzy metric entropy FuzzyMEn, and distribution entropy DistEn.

[0017] The heart rate asymmetry features include: Porta index, Guzik index, slope index, and area index.

[0018] Step S4 includes:

[0019] The extracted multi-domain heart rate variability features and individual MEQ types are combined to construct a multi-domain feature set D, and the feature set is preprocessed, including handling missing values ​​and normalizing features.

[0020] The individuals' MEQ type is a sleep rhythm type, including: early bird type (preferring to sleep early), owl type (preferring to sleep late), and intermediate type.

[0021] The method for constructing the sleep rhythm state assessment model includes,

[0022] The Gini index is chosen as the splitting criterion for constructing the decision tree. The Gini index (D,a) for each feature in the dataset is calculated as follows:

[0023] Where, p k Let n represent the probability of different rhythm disorder risk categories in the dataset, where n is the number of rhythm disorder risk categories, and Gini(D) represents the probability of each risk category. v ) is the Gini coefficient for each feature, where a represents the feature category, and D v The feature set segmented based on feature a;

[0024] Starting from the root node, calculate the Gini index of all features, select the feature with the smallest Gini index and its corresponding value as the optimal feature and optimal threshold; based on the selected optimal feature and optimal threshold, divide the dataset into two subsets, and then recursively repeat the above steps for each subset, that is, select another optimal feature and its threshold for further division, until the stopping conditions are met, wherein the stopping conditions include: all samples belong to the same class, there are no available features for further division, the preset maximum tree depth is reached, or the number of samples in the subset is less than the preset threshold.

[0025] The method for constructing the sleep rhythm state assessment model also includes:

[0026] The feature set is randomly divided according to a preset ratio to obtain a training set and a test set. The test set is input into the trained sleep rhythm state assessment model for result prediction. The sleep rhythm state assessment model is constructed by ensemble decision tree, and the model accuracy is evaluated using accuracy, confusion matrix and 10-fold cross-validation.

[0027] A sleep rhythm disorder detection device based on long-range signals and multi-feature extraction includes a memory, a processor, and a program stored in the memory, characterized in that the processor executes the program to implement the method described above.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] 1. Monitoring and analyzing the sleep rhythm of target users based on long-range signals helps to capture potential rhythm disorders in real time during long-term sleep states, reducing the workload of frequently collecting physiological signals.

[0030] 2. This method extracts multiple features from long-range signals and analyzes the risk of rhythm disturbances from multiple perspectives. It combines multi-domain heart rate variability features and individual MEQ types. In particular, the introduction of heart rate asymmetry features, Pompeii diagram features, and various entropy complexities into the multi-domain heart rate variability features greatly improves the accuracy of rhythm disturbance risk prediction.

[0031] 3. When applied to daily life and work, it can help predict potential rhythm disorders in individuals. On the one hand, it can effectively detect potential rhythm disorders and reduce physical and mental health problems caused by rhythm disorders. On the other hand, it can optimize the work performance of workers in production operations and improve safety, reducing accidents caused by fatigue and lack of concentration caused by rhythm disorders. Attached Figure Description

[0032] Figure 1 is a schematic diagram of the main steps of the method of the present invention;

[0033] Figure 2 is a scatter plot of the RP interval. Detailed Implementation

[0034] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0035] A sleep rhythm disorder detection method based on long-range signals and multi-feature extraction, as shown in Figure 1, includes:

[0036] Step S1: Obtain the physiological signals of the target user, wherein the physiological signals include at least heart rate;

[0037] Physiological signals are collected through wearable electronic devices. Generally, wearable electronic devices can be non-invasive electronic devices such as ECG patches and pulse sensors, which can collect the physiological data of the target user in real time. Specifically, the physiological signals are mainly heart rate, and can also be combined with pulse data. Of course, in some embodiments, the pulse sensor can be a blood oxygen pulse sensor.

[0038] Step S2: Preprocess and extract peak values ​​from physiological signals;

[0039] The preprocessing process involves using a bandpass filter to remove high-frequency and low-frequency noise from the acquired physiological signals. Specifically, this process includes setting a bandpass filter with a frequency range of 0.5 Hz to 4 Hz to remove low-frequency and high-frequency noise from the acquired physiological signals.

[0040] Furthermore, in this embodiment, the Pan-Tompkins algorithm is used to extract peak features from the processed data;

[0041] Step S3: Extraction of preprocessed physiological signals. Based on the preprocessed signals, multi-domain heart rate variability features are extracted using a multi-domain feature extraction algorithm.

[0042] Multidomain heart rate variability features include HRV time-domain features, HRV frequency-domain features, and HRV nonlinear features. HRV nonlinear features include heart rate asymmetry features, Pompeii diagram features, and various entropy complexities.

[0043] HRV temporal features include the standard deviation (SDNN) of normal sinus RR intervals, the root mean square (RMSSD) of the difference between adjacent RR intervals, and the percentage of adjacent RR intervals with a difference greater than 50 ms out of all RR intervals. 50 The specific calculation method is as follows:

[0044] Where: N is the total number of normal heartbeats in the entire recording, and meanRR is the average RR interval of N heartbeats; RR i , RRi+1 is the length of two adjacent sinus cycles; NN 50 N represents the number of consecutive RR intervals with a difference greater than 50ms in the entire record, where N is the total number of RR intervals.

[0045] HRV frequency domain characteristics include total power TP, low-frequency power LF, high-frequency power HF, and low-frequency / high-frequency ratio LF / HF, etc., and the specific calculation method is as follows;

[0046] An AR model power spectrum analysis was performed on the RR interval sequence. The area within the range of 0-0.4 Hz was taken as the total power (TP). The area within the range of 0.04-0.15 Hz was taken as the low-frequency power (LF), and the area within the range of 0.15-0.4 Hz was taken as the high-frequency power (HF).

[0047] The features of a Pompeii diagram include the semi-major axis SD1, the semi-minor axis SD2, and the ratio of the major axis to the semi-minor axis SD1 / SD2, which are calculated in the following way;

[0048] Where: X = {x i |i=1..N},Y={y i |i=1..N}, (xi,yi) are the coordinates of the points in the Pompeii scatter plot;

[0049] Various entropy complexities include approximate entropy ApEn, sample entropy SampE, fuzzy entropy FuzzyEn, fuzzy metric entropy FuzzyMEn, and distribution entropy DistEn;

[0050] Heart rate asymmetry characteristics include: Porta index, Guzik index, slope index, and area index.

[0051] For the RR interval (RR1, RR2, ..., RR... i ,RR i+1 ,…,RRn-1 ,RR n Figure 2 illustrates the algorithm for the heart rate asymmetry index implemented using the following formula, where LI is the identification line y = x, and P... i It is the i-th point, which can be represented as (RR) i ,RR i+1 ), ΔRR=RR i+1 -RR i D i Represents P i The distance to LI, expressed as the angle between the straight line between the origin and the x-axis, is θ. i The phase angle between the horizontal axis and LI is θ. LI :

[0052] The Porta index, abbreviated as PI, is measured proportionally to points below LI. The formula for calculating PI is:

[0053] Where: b and m represent the number of points below L1 and points not on L1, respectively;

[0054] The Guzik index, abbreviated as GI, is the ratio of the cumulative distance to a point, and it can be calculated as follows:

[0055] The slope index, abbreviated as SI, assesses the percentage of the cumulative phase angle above LI, and can be calculated as follows:

[0056] Where: R θi =θ LI -θ i

[0057] For the area index, abbreviated as AI, the area of ​​the i-th sector is represented as:

[0058] Where: r is the distance from Pi to the origin of the coordinate system.

[0059] Then, AI is defined as the ratio of Si to Li, and the formula for AI is:

[0060] Step S4: Construct a multi-domain feature set and classify rhythm disorder risk categories based on the extracted multi-domain heart rate variability features, including:

[0061] The extracted multi-domain heart rate variability features and individual MEQ types are combined to construct a multi-domain feature set D, and the feature set is preprocessed, including handling missing values ​​and normalizing features. In this embodiment, z-score normalization is used to normalize the data, and all data are within the range of [-1, 1].

[0062] An individual's MEQ type is a sleep rhythm type, including: early bird (preferring to sleep early), owl (preferring to sleep late), and intermediate type.

[0063] This example categorizes individuals’ rhythm disorder risk based on their PSQI scores: a PSQI score above 8 indicates a high risk of rhythm disorder, while a PSQI score below 8 indicates a low risk of rhythm disorder.

[0064] Step S5: Construct a sleep rhythm state assessment model using the random forest algorithm to detect potential risks of sleep rhythm disorders.

[0065] Methods for constructing sleep rhythm state assessment models include,

[0066] The Gini index is chosen as the splitting criterion for constructing the decision tree. The Gini index (D,a) for each feature in the dataset is calculated as follows:

[0067] Where, p k Let n represent the probability of different rhythm disorder risk categories in the dataset, where n is the number of rhythm disorder risk categories, and Gini(D) represents the probability of each risk category. v ) is the Gini coefficient for each feature, where a represents the feature category, and D v The feature set segmented based on feature a;

[0068] Starting from the root node, calculate the Gini index of all features, select the feature with the smallest Gini index and its corresponding value as the optimal feature and optimal threshold; based on the selected optimal feature and optimal threshold, divide the dataset into two subsets, and then recursively repeat the above steps for each subset, that is, select another optimal feature and its threshold for further division, until the stopping conditions are met. The stopping conditions include: all samples belong to the same class, there are no available features for further division, the preset maximum tree depth is reached, or the number of samples in the subset is less than the preset threshold.

[0069] Methods for constructing sleep rhythm state assessment models also include:

[0070] The feature set is randomly divided according to a preset ratio to obtain a training set and a test set. The test set is input into the trained sleep rhythm state assessment model to predict the results. The sleep rhythm state assessment model is constructed through an ensemble decision tree, and the accuracy is evaluated using the accuracy, confusion matrix and ten-fold cross-validation.

[0071] In this example, the random forest model repeats the decision tree construction process 100 times to generate 100 decision trees; these 100 decision trees are then combined into a random forest. The mode of the prediction results from the 100 decision trees is used as the final output of the random forest prediction model.

[0072] In this example, the multi-domain feature set is divided into a training set and a test set in a 7:3 ratio. The test set is then substituted into the random forest model to predict the results. The accuracy of the model is evaluated using the accuracy, confusion matrix, and 10-fold cross-validation.

[0073] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A sleep rhythm disorder detection method based on long-range signal and multi-feature extraction, characterized in that, include: Step S1: Obtain the physiological signals of the target user, wherein the physiological signals include at least heart rate; Step S2: Preprocess and extract peak values ​​from the physiological signals; Step S3: Extracting physiological signals based on preprocessed signals. Multi-domain heart rate variability features are extracted using a multi-domain feature extraction algorithm based on the preprocessed signals. Step S4: Construct a multi-domain feature set and classify rhythm disorder risk categories based on the extracted multi-domain heart rate variability features; Step S5: Construct a sleep rhythm state assessment model using the random forest algorithm to detect potential risks of sleep rhythm disorders; The multi-domain heart rate variability features include HRV time-domain features, HRV frequency-domain features, and HRV nonlinear features. The HRV nonlinear features include heart rate asymmetry features, Pompeii diagram features, and various entropy complexities. The Pompeii diagram features include the semi-major axis SD1, the semi-minor axis SD2, and the ratio of the major axis to the semi-minor axis SD1 / SD2; The various entropy complexities include approximate entropy ApEn, sample entropy SampE, fuzzy entropy FuzzyEn, fuzzy metric entropy FuzzyMEn, and distribution entropy DistEn. The heart rate asymmetry features include the Porta index, Guzik index, slope index, and area index; Step S4 includes: The extracted multi-domain heart rate variability features and individual MEQ types are combined to construct a multi-domain feature set D, and the feature set is preprocessed, including handling missing values ​​and normalizing features. The individuals' MEQ type is a sleep rhythm type, including: early bird type (preferring to sleep early), owl type (preferring to sleep late), and intermediate type; The method for constructing the sleep rhythm state assessment model includes, The Gini index is selected as the splitting criterion to construct the decision tree, wherein the calculation method of the Gini index Gini(D, a) of each feature of the data set is as follows: where p k is the probability of each rhythm disorder risk category in the dataset, n is the number of categories of rhythm disorder risk, Gini(D v ) is the Gini coefficient for each feature, a represents a feature category, and D v is the feature set split based on feature a. Starting from the root node, calculate the Gini index of all features, and select the feature with the smallest Gini index and its corresponding value as the optimal feature and the optimal threshold. Based on the selected optimal feature and optimal threshold, the dataset is divided into two subsets. Then, the above steps are repeated recursively for each subset, that is, an optimal feature and its threshold are selected again for further division, until the stopping conditions are met. The stopping conditions include: all samples belong to the same class, there are no available features for further division, the preset maximum tree depth is reached, or the number of samples in the subset is less than the preset threshold.

2. The method of claim 1, wherein the method is characterized by, The physiological signals are collected via wearable electronic devices.

3. The method of claim 1, wherein the method is characterized by, The preprocessing process in step S2 involves using a bandpass filter to remove high-frequency and low-frequency noise from the acquired physiological signals.

4. The method of claim 1, wherein the method is characterized by, The method for constructing the sleep rhythm state assessment model also includes: The feature set is randomly divided according to a preset ratio to obtain a training set and a test set. The test set is input into the trained sleep rhythm state assessment model for result prediction. The sleep rhythm state assessment model is constructed by ensemble decision tree, and the model accuracy is evaluated using accuracy, confusion matrix and 10-fold cross-validation. 5.A device for detecting sleep rhythm disorder based on long-range signal and multi-feature extraction, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-4.

Images