An automatic sleep apnea syndrome detection device

Abstract

The application belongs to the technical field of medical monitoring, and specifically discloses a sleep apnea syndrome automatic detection device, which comprises a signal acquisition module, a data preprocessing module, a feature extraction module and a fusion decision module. The signal acquisition module is used for acquiring RR interval signals and respiratory wave signals. The data preprocessing module is used for removing interference areas through preprocessing to obtain preprocessed RR interval signals and respiratory wave signals. The feature extraction module is used for determining sleep apnea events characterized by the RR interval signals and the respiratory wave signals through statistical analysis based on the preprocessed RR interval signals and the respiratory wave signals. The fusion decision module is used for analyzing the correlation between sleep apnea events characterized by the RR interval signals and the respiratory wave signals and determining a sleep apnea detection result. Through the application, sleep apnea syndrome can be stably and efficiently detected while ensuring detection accuracy.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of medical monitoring, more specifically, relates to an automatic detection device for sleep apnea syndrome. BACKGROUND

[0002] Obstructive Sleep Apnea Syndrome (OSAS) is a common sleep disorder characterized by repeated upper airway collapse during sleep, leading to airflow obstruction and even complete respiratory arrest. OSAS not only affects the quality of sleep, but also increases the risk of other diseases.

[0003] In clinical diagnosis, OSAS is usually diagnosed by polysomnography (PSG), which can record brain waves, eye movements, muscle electrical activity, electrocardiogram, respiratory flow, blood oxygen saturation and other physiological signals, and comprehensively evaluate the severity of OSAS. However, PSG equipment is expensive, the detection process is complex, and patients need to be hospitalized for monitoring, which is not suitable for large-scale screening. Therefore, how to find a convenient, efficient and low-cost screening method, and recommend moderate and severe suspected OSAS patients to the respiratory department for PSG diagnosis, has become the focus of current research.

[0004] Currently, the initial screening detection schemes for OSAS mainly include the following: (1) portable sleep monitoring devices, (2) devices based on respiratory rate monitoring, and (3) devices based on dynamic electrocardiogram recorders with electrocardiogram and respiratory wave double parameters.

[0005] (1) Portable sleep monitoring devices;

[0006] Such devices usually monitor blood oxygen saturation, respiratory flow, heart rate and other parameters, and are more convenient than PSG, and some devices can be used in a home environment. However, due to limited monitoring signals, low detection accuracy, and susceptibility to individual physiological differences and environmental factors, there is a high misdiagnosis rate, and it is difficult to achieve large-scale screening.

[0007] (2) Devices based on respiratory rate monitoring;

[0008] This method uses chest impedance, air flow and other signals to detect apnea events. However, this method is susceptible to chest conditions and body position changes, resulting in certain limitations in detection accuracy, especially in identifying short-term apnea.

[0009] (3) Devices based on dynamic electrocardiogram recorders with electrocardiogram and respiratory wave double parameters;

[0010] The device calculates the RR interval change through the electrocardiosignal and records the respiratory wave data in combination with the thoracic impedance signal, and the detection accuracy of sleep apnea can be improved by using the double-parameter comprehensive analysis. Compared with single-parameter detection, this method belongs to non-invasive screening, and dynamic electrocardiogram recording has been widely used in cardiovascular disease detection, so it is suitable for large-scale screening. However, the final diagnosis of the method still depends on manual annotation, that is, the doctor needs to manually mark the start and end points of the sleep apnea event according to the heart rate grid and the respiratory wave, and then determine the mild, moderate or severe suspected patient according to the proportion of the onset time.

[0011] The electrocardiosignal (ECG) is the manifestation of the electrical activity of the heart on the body surface, which reflects the excitation state and excitation propagation process of each part of the heart. The R wave is the first positive wave in the QRS complex, representing the potential change of the depolarization (excitation) of the ventricular muscle cells, marking the beginning of ventricular contraction. The RR interval refers to the time interval between two adjacent R waves, reflecting a complete cardiac cycle (the time from ventricular contraction to the next contraction).

[0012] Currently, dynamic electrocardiogram recorders based on electrocardio and respiratory wave double parameters are mainly used for initial screening of OSAS patients, but the existing manual annotation method has the following defects.

[0013] (1) Time-consuming and labor-intensive, difficult to promote. Doctors need to manually analyze dynamic electrocardiogram and respiratory wave data, and the annotation time for each patient is 5-10 minutes, which seriously affects the screening efficiency and makes it difficult to promote to large-scale application.

[0014] (2) Large subjective error and low annotation consistency. Manual annotation depends on the experience of doctors, and there may be differences between different doctors, and it is easily affected by the interference area in the signal, resulting in low stability of the screening result.

[0015] (3) Difficult to detect in real time. Manual screening usually requires post-data review and analysis, which cannot realize real-time monitoring, limiting the application of this method in telemedicine and long-term monitoring.

[0016] The efficiency of manual annotation is low and the stability of the result is low. How to stably and efficiently detect sleep apnea syndrome while ensuring detection accuracy is a technical problem to be solved in the field. SUMMARY

[0017] In view of the defects of the prior art, the purpose of the present application is to stably and efficiently detect sleep apnea syndrome while ensuring detection accuracy.

[0018] To achieve the above-mentioned purpose, in a first aspect, the present application provides an automatic detection device for sleep apnea syndrome, which comprises a signal acquisition module, a data preprocessing module, a feature extraction module and a fusion decision module.

[0019] a signal acquisition module, configured to acquire the RR interval signal and the respiratory wave signal;

[0020] a data preprocessing module, configured to remove interference areas in the RR interval signal and the respiratory wave signal through preprocessing, to obtain preprocessed RR interval signal and respiratory wave signal;

[0021] a feature extraction module, configured to determine sleep apnea events represented by the RR interval signal and sleep apnea events represented by the respiratory wave signal through statistical analysis based on the preprocessed RR interval signal and the respiratory wave signal;

[0022] a fusion decision module, configured to analyze the correlation between the sleep apnea events represented by the RR interval signal and the sleep apnea events represented by the respiratory wave signal, and determine a sleep apnea detection result.

[0023] It can be understood that, by removing interference areas in the RR interval signal and the respiratory wave signal through preprocessing, interference segments such as arrhythmia (e.g., atrial fibrillation, premature beat) and motion artifact can be automatically excluded, and interference can be excluded, thereby significantly improving the stability of detection; by automatically acquiring the RR interval signal and the respiratory wave signal through the signal acquisition module, automatically performing preprocessing through the data preprocessing module, automatically identifying sleep apnea events (i.e., automatically performing sleep apnea-related behavior feature recognition) through the feature extraction module, and automatically generating a sleep apnea detection result through the fusion decision module, full automation from signal acquisition to result generation can be achieved, without the need for manual annotation, thereby significantly improving the efficiency of detection; by analyzing the correlation between the sleep apnea events represented by the RR interval signal and the sleep apnea events represented by the respiratory wave signal, heart rate grid and respiratory wave data can be effectively fused, the misjudgment rate is reduced compared to a single physiological signal detection method, and the detection accuracy is ensured.

[0024] Therefore, through cooperation between the data preprocessing module, the feature extraction module, and the fusion decision module, sleep apnea syndrome can be stably and efficiently detected while ensuring detection accuracy.

[0025] In a possible implementation, the data preprocessing module includes an RR interval signal preprocessing submodule and a respiratory wave signal preprocessing submodule.

[0026] The RR interval signal preprocessing submodule is configured to:

[0027] identify motion artifact interference areas in the RR interval signal through a deep learning model;

[0028] identify abnormal RR intervals in the RR interval signal through statistical analysis;

[0029] remove motion artifact interference regions in the RR interval signal and abnormal RR intervals;

[0030] a respiration wave signal preprocessing submodule, configured to:

[0031] identify motion artifact interference regions in the respiration wave signal through the deep learning model;

[0032] remove motion artifact interference regions in the respiration wave signal.

[0033] In a possible implementation, the abnormal RR interval is an abnormal RR interval caused by atrial fibrillation, an abnormal RR interval caused by premature beat, an abnormal RR interval caused by bigeminy, or an abnormal RR interval caused by trigeminy.

[0034] In a possible implementation, the feature extraction module comprises an RR interval signal feature extraction submodule and a respiration wave signal feature extraction submodule.

[0035] The RR interval signal feature extraction submodule is configured to remove high-frequency noise in the (RR interval) signal, retain sleep apnea features, and determine sleep apnea events represented by the RR interval signal based on the sleep apnea features in the RR interval signal.

[0036] The respiration wave signal feature extraction submodule is configured to remove high-frequency noise in the (respiration wave) signal, retain sleep apnea features, and determine sleep apnea events represented by the respiration wave signal based on the sleep apnea features in the respiration wave signal.

[0037] In a possible implementation, the above removing high-frequency noise in the signal and retaining sleep apnea features comprises:

[0038] The signal is decomposed by 5 orders using db4 wavelet transform to obtain D1, D2, D3, D4, D5, and A5, D1 is a detail coefficient obtained by 1st layer decomposition, D2 is a detail coefficient obtained by 2nd layer decomposition, D3 is a detail coefficient obtained by 3rd layer decomposition, D4 is a detail coefficient obtained by 4th layer decomposition, D5 is a detail coefficient obtained by 5th layer decomposition, and A5 is an approximation coefficient obtained by 5th layer decomposition;

[0039] Based on D3, D4, and D5, the signal is reconstructed to retain sleep apnea features.

[0040] In a possible implementation, the above determining sleep apnea events represented by the RR interval signal based on the sleep apnea features in the RR interval signal comprises:

[0041] determining the peak and the trough in the RR interval signal;

[0042] determining the peak interval between two adjacent peaks and the trough interval between two adjacent troughs in the RR interval signal based on the peak and the trough in the RR interval signal;

[0043] determining the start and end points of the sleep apnea event based on the coefficient of variation threshold, the coefficient of variation of the peak interval and the coefficient of variation of the trough interval.

[0044] In a possible implementation, the determination of the start and end points of the sleep apnea event based on the coefficient of variation threshold, the coefficient of variation of the peak interval and the coefficient of variation of the trough interval includes:

[0045] If the coefficient of variation of the peak interval continuously exceeds the coefficient of variation threshold (for example, 0.3) for three adjacent peaks, and the duration is greater than a preset time threshold (for example, 60 seconds), the start point of the sleep apnea event is marked; or if the coefficient of variation of the trough interval continuously exceeds the coefficient of variation threshold (for example, 0.3) for three adjacent troughs, and the duration is greater than the time threshold (for example, 60 seconds), the start point of the sleep apnea event is marked.

[0046] After the start point of the sleep apnea event is marked, if the coefficients of variation of the peak interval and the trough interval continuously decrease to be less than a baseline value, and the duration is greater than the time threshold (for example, 60 seconds), the end point of the sleep apnea event is marked, and the baseline value is less than the coefficient of variation threshold.

[0047] It can be understood that experiments show that when the coefficient of variation of the peak interval or the trough interval continuously exceeds a preset coefficient of variation threshold (for example, 0.3) and the duration exceeds a preset time threshold (for example, 60 seconds), the start of a sleep apnea event is marked; and after the start point is determined, the coefficients of variation of the peak interval and the trough interval are continuously monitored, and when the coefficients of variation continuously decrease to be less than a preset baseline value (the baseline value is less than the coefficient of variation threshold) and the duration exceeds the time threshold, the end of the sleep apnea event is marked.

[0048] In a possible implementation, the determination of the sleep apnea event based on the sleep apnea feature in the respiratory wave signal includes:

[0049] calculating the respiratory wave amplitude change rate by using a sliding window;

[0050] determining the start and end points of the sleep apnea event based on the respiratory wave amplitude change rate and the baseline amplitude.

[0051] In a possible implementation, the determining the start and end points of the sleep apnea event based on the respiratory wave amplitude change rate and the baseline amplitude comprises:

[0052] If the respiratory wave amplitude change rate continuously decreases to be lower than the first preset percentage (e.g., 50%) of the baseline amplitude and the duration is greater than the time threshold (e.g., 60 seconds), the start point of the sleep apnea event is marked.

[0053] If the respiratory wave amplitude change rate continuously increases to be higher than the second preset percentage (e.g., 80%) of the baseline amplitude and the duration is greater than the time threshold (e.g., 60 seconds) after the start point of the sleep apnea event is marked, the end point of the sleep apnea event is marked, and the first preset percentage is less than the second preset percentage.

[0054] It can be understood that two preset percentages are set, one is used to identify the decrease of the respiratory wave amplitude change rate (e.g., 50%), and the other is used to identify the increase of the respiratory wave amplitude change rate (e.g., 80%), and the two percentages satisfy the first preset percentage is less than the second preset percentage. Experiments show that if the respiratory wave amplitude change rate continuously decreases to be lower than the first preset percentage of the baseline amplitude, and the duration of this low change rate state exceeds the preset time threshold (e.g., 60 seconds), it marks the start of a sleep apnea event; and once the start point is marked, the amplitude change rate is continuously monitored, and when it continuously increases to be higher than the second preset percentage of the baseline amplitude, and the duration of this high change rate state exceeds the time threshold, it marks the end of a sleep apnea event.

[0055] In a possible implementation, the sleep apnea detection result comprises a first determination result, a second determination result and a third determination result, and the fusion decision module is specifically configured to:

[0056] If the occurrence number of the sleep apnea event represented by the RR interval signal is greater than the first preset event number (e.g., 3 times) and the occurrence number of the sleep apnea event represented by the respiratory wave signal is greater than the first preset event number in the same target time period, the first determination result is generated, which is used to indicate that the sleep apnea event in the target time period is of high credibility.

[0057] If the number of sleep apnea events represented by the RR interval signal is greater than the first preset event number, or the number of sleep apnea events represented by the respiratory wave signal is greater than the first preset event number, the number of sleep apnea events in a preset time period before the target time period is greater than the second preset event number (for example, 2 times), and the number of sleep apnea events in a preset time period after the target time period is greater than the second preset event number, a second determination result is generated, indicating that the sleep apnea events in the target time period are of high credibility, and the first preset event number is greater than the second preset event number.

[0058] Events other than the sleep apnea events of high credibility are determined to be of low credibility, and a third determination result is generated, indicating the sleep apnea events of low credibility.

[0059] It can be understood that experiments show that, for the RR interval signal and the respiratory wave signal in a specific target time period, if the number of sleep apnea events detected by each of the two signals is greater than a certain event number (for example, 3 times), it indicates that the sleep apnea events in the time period are of high credibility; if the number of sleep apnea events detected by either signal is greater than the first preset event number, and the number of sleep apnea events in a preset time period before and after the target time period is greater than the second preset event number (for example, 2 times), it also indicates that the sleep apnea events in the time period are of high credibility.

[0060] Through this multi-signal fusion and time period comparison method, the accuracy and reliability of sleep apnea event detection can be significantly improved.

[0061] Overall, compared with the prior art, the above technical solutions conceived by the present application have the following beneficial effects:

[0062] (1) The deep learning technology and statistical analysis are used to calibrate the interference region of the RR interval and respiratory wave data, automatically exclude interference segments such as arrhythmia (such as atrial fibrillation, premature beat) and motion artifact, realize interference exclusion, and improve the stability of analysis.

[0063] (2) The onset region of sleep apnea is automatically calibrated, without manual annotation, and the detection efficiency is improved.

[0064] (3) The RR interval signal and the respiratory wave signal are fused and analyzed to improve the detection accuracy, and compared with the single physiological signal detection method, the false positive rate is reduced.

[0065] (4) The calculation overhead is small, which is suitable for dynamic electrocardiogram for OSAS mass screening, real-time monitoring, remote medical treatment, and long-term monitoring equipment. BRIEF DESCRIPTION OF DRAWINGS

[0066] Figure 1 is a structural schematic diagram of a sleep apnea syndrome automatic detection device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0067] In order to make the objects, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0068] The terms "first" and "second" and the like in the specification and claims of the present application are used to distinguish different objects, rather than to describe a specific order of the objects. For example, the first preset percentage and the second preset percentage are used to distinguish different preset percentages, rather than to describe a specific order of the preset percentages.

[0069] In the embodiments of the present application, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration, in the embodiments of the present application, any embodiment or design scheme described as "exemplary" or "for example" should not be construed as being more preferred or having more advantages than other embodiments or design schemes. Rather, the words "exemplary" or "for example" are used to present the relevant concept in a specific manner.

[0070] In the description of the embodiments of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more, for example, a plurality of processing units means two or more processing units, and the like; a plurality of elements means two or more elements, and the like.

[0071] The embodiments of the present application are described below with reference to the drawings in the embodiments of the present application.

[0072] Figure 1 is a structural schematic diagram of a sleep apnea syndrome automatic detection device provided by an embodiment of the present application, as shown in the figure, the device comprises a signal acquisition module 10, a data preprocessing module 20, a feature extraction module 30 and a fusion decision module 40. Figure 1

[0073] The signal acquisition module 10 acquires dynamic electrocardiogram signals (specifically, RR interval signals) and respiratory wave signals, which are used for analyzing sleep apnea events.

[0074] The data preprocessing module 20 performs preprocessing such as denoising, interference region calibration and rejection, standardization and the like on the collected original signals, so as to improve the data quality.

[0075] ​Feature extraction module 30: Extract key features from pre-processed signals, such as RR interval fluctuation, respiratory amplitude drop / recovery trend, etc. Use time series pattern matching algorithm to identify periodic changes in RR interval grid (RR interval signal) and respiratory wave graph (respiratory wave signal).

[0076] Fusion decision module 40: Use RR interval signal and respiratory wave signal fusion analysis to determine high-confidence sleep apnea events.

[0077] (1) The following exemplary description of the signal acquisition module 10.

[0078] The signal acquisition module can collect electrocardiogram (ECG, Electrocardiogram) signals through a dynamic electrocardiogram recorder, calculate RR intervals; record respiratory waves through thoracic impedance measurement; synchronize timestamps to ensure time alignment between different signals, and improve the accuracy of data analysis using statistical methods to exclude abnormal RR intervals (such as atrial fibrillation, premature beats, bigeminy, trigeminy, etc.).

[0079] (2) The following exemplary description of the data preprocessing module 20. The data preprocessing module 20 includes: RR interval signal preprocessing submodule and respiratory wave signal preprocessing submodule.

[0080] (2-1) RR interval signal preprocessing submodule: use deep learning technology and statistical methods to preprocess RR interval signal.

[0081] Specifically, use CNN deep learning to calibrate interference areas and exclude motion artifact interference segments such as turning over. And use statistical methods to exclude abnormal RR intervals such as atrial fibrillation, premature beats, bigeminy, trigeminy, etc.

[0082] In dynamic electrocardiogram (ECG), motion artifacts such as turning over are non-cardiac rhythm related interference, which can cause signal distortion and inaccurate extraction of RR intervals, affecting subsequent analysis. Therefore, such motion artifact interference needs to be calibrated and excluded.

[0083] Even if the motion artifact interference area is calibrated, there may still be abnormal intervals in the RR interval due to arrhythmia (such as atrial fibrillation, premature beats, bigeminy, trigeminy). Therefore, abnormal intervals need to be excluded.

[0084] Exemplarily, the specific way to exclude motion artifact interference segments can be to input the RR interval signal (a time series segment of 256 points, which is one-dimensional data) into a pre-trained CNN model.

[0085] The CNN architecture includes an input layer, a convolutional layer, a pooling layer, and a fully connected layer. The input layer accepts standardized RR interval signals. The convolutional layer extracts local features of the RR interval signals (such as waveform morphology). The pooling layer reduces the feature dimension and enhances the generalization ability of the model. The fully connected layer classifies the signal segment as "interference" or "non-interference".

[0086] The CNN model outputs the calibration results: the interference segment positions output by the model are labeled and recorded, and excluded from processing.

[0087] For example, the specific way to exclude abnormal RR intervals can be to exclude outliers that exceed 3 times the standard deviation (SD); based on the coefficient of variation (CV), exclude intervals with high volatility and poor stability; through time series pattern analysis, exclude abnormal RR intervals such as atrial fibrillation, premature beats, and double and triple rhythms.

[0088] The coefficient of variation is the ratio of the standard deviation to the mean, usually expressed as a percentage.

[0089] Atrial fibrillation: the distribution of RR intervals is highly irregular (CV > 0.3). In atrial fibrillation, RR intervals are usually irregular, and the coefficient of variation can be used to exclude abnormal intervals caused by atrial fibrillation.

[0090] Premature beat: a short interval followed by a long interval, showing a typical compensatory pause pattern. Premature beats usually manifest as a sudden shortening of RR intervals (i.e., coupling interval), followed by a longer compensatory interval (RR interval extension), which can be used to identify and exclude abnormal intervals caused by atrial fibrillation.

[0091] Double and triple rhythms: RR intervals show a periodic fixed pattern. The RR interval characteristics of double rhythm have a fixed alternating pattern: a normal RR interval followed by a premature beat, forming a periodic pattern of "normal-premature-normal-premature", which can be used to identify and exclude abnormal intervals caused by double rhythm. The RR interval characteristics of triple rhythm also have a fixed alternating pattern: two normal RR intervals followed by a premature beat (e.g., "normal-normal-premature"), which can be used to identify and exclude abnormal intervals caused by triple rhythm.

[0092] The processing result of excluding abnormal RR intervals: all abnormal intervals are excluded, and the data in the interference area is filled by interpolation processing, only retaining the RR intervals that truly reflect the regulation of the autonomic nervous system.

[0093] (2-2) Respiratory wave signal preprocessing submodule: using deep learning technology to preprocess the respiratory wave signal.

[0094] Specifically, the CNN model is used to identify and exclude interference regions, ensuring the accuracy of the data. Respiratory wave signals are collected from body surface impedance and are susceptible to motion artifacts such as turning over. Interference features include sudden changes in signal amplitude or flat segments.

[0095] As an example, the respiratory wave signal (a time series segment with a length of 512 points) is input into a pre-trained CNN model.

[0096] The CNN architecture includes an input layer, a convolutional layer, a pooling layer, and a fully connected layer. The input layer accepts standardized respiratory wave signals. The convolutional layer extracts local interference features (such as amplitude mutations). The pooling layer reduces the feature dimension and reduces the computational complexity. The fully connected layer classifies the interference regions.

[0097] The CNN model outputs a labeling result: labels the interference regions to ensure the signal quality for subsequent analysis.

[0098] (3) The feature extraction module 30 is described as an example. The feature extraction module 30 includes: an RR interval signal feature extraction submodule and a respiratory wave signal feature extraction submodule.

[0099] (3-1) The RR interval signal feature extraction submodule is used to: through RR interval grid analysis, detect the periodic changes of heart rate, and identify sleep apnea events.

[0100] The processing logic of the RR interval signal feature extraction submodule includes the following links: (3-1-1) signal time-frequency decomposition, (3-1-2) signal reconstruction, (3-1-3) peak and trough detection, (3-1-4) calculation of peak-to-peak distance and trough-to-trough distance, (3-1-5) determination of sleep apnea events.

[0101] (3-1-1) Signal time-frequency decomposition: db4 wavelet transform is used for 5-order decomposition.

[0102] ;

[0103] Among them, A5 (low-frequency component): represents the long-period trend, reflecting the overall heart rate change trend. D5-D3 (medium-frequency component): related to the rhythm changes of the autonomic nervous system. D2-D1 (high-frequency component): may contain noise or short-term changes.

[0104] (3-1-2) Signal reconstruction: select D5-D3 for signal reconstruction, retaining the main sleep apnea features.

[0105] ;

[0106] Remove high-frequency noise and retain sleep apnea-related heart rate change information.

[0107] (3-1-3) Peak-trough detection: use first derivative + second derivative method to detect heart rate fluctuation pattern.

[0108] ;

[0109] Set threshold for peak detection: local maxima, as follows:

[0110] and ;

[0111] Set threshold for trough detection: local minima, as follows:

[0112] and .

[0113] (3-1-4) Calculate peak-to-peak distance and trough-to-trough distance: detect periodic heart rate variation.

[0114] Calculate the interval between consecutive peaks (P-P) ) and the interval between troughs (T-T) ):

[0115] ;

[0116] (3-1-5) Determine sleep apnea event;

[0117] Calculate the coefficient of variation (CV) of the peak-to-peak distance (or trough-to-trough distance):

[0118] ;

[0119] wherein, denotes the standard deviation of the peak-to-peak distance (or trough-to-trough distance), denotes the mean of the peak-to-peak distance (or trough-to-trough distance). Exemplarily, a sliding window (e.g. the sliding window size can be set to 3 minutes long) can be set, and for the peak-to-peak distance (or trough-to-trough distance) within the sliding window, the standard deviation and the mean are calculated, and then the standard deviation is divided by the mean to obtain the coefficient of variation.

[0120] When three consecutive peaks / troughs are detected, and the coefficient of variation (CV) continues to exceed the set threshold (e.g. 0.3) for more than 60 seconds, it is marked as the starting point of sleep apnea.

[0121] When the coefficient of variation of the peak-to-peak distance and the trough-to-trough distance returns to the baseline value (CV < 0.1) and maintains for more than 60 seconds, it is marked as the termination point of sleep apnea.

[0122] (3-2) Respiratory wave signal feature extraction sub-module is used for: analyzing respiratory wave amplitude variation, assisting in identifying sleep apnea event.

[0123] The processing logic of the respiration wave signal feature extraction submodule includes the following aspects: (3-2-1) signal time-frequency analysis, (3-2-2) amplitude calculation, (3-2-3) sleep apnea starting point determination, and (3-2-4) sleep apnea termination point determination.

[0124] (3-2-1) Signal time-frequency analysis: db4 wavelet transform is used for 5-order decomposition, and D5-D3 is selected for signal reconstruction to retain the main sleep apnea features.

[0125] (3-2-2) Amplitude calculation: a 32-point sliding window (which can select 32 points) is used to calculate the respiration wave amplitude change rate (ACR):

[0126] .

[0127] (3-2-3) Sleep apnea starting point determination:

[0128] When ACR continuously falls below 50% of the baseline amplitude for more than 60 seconds, the sleep apnea starting point is determined.

[0129] (3-2-4) Sleep apnea termination point determination:

[0130] When ACR rises to more than 80% of the baseline amplitude and maintains for more than 60 seconds, the sleep apnea termination point is determined.

[0131] (4) The fusion decision module 40 is exemplarily described as follows.

[0132] Rule matching method:

[0133] If both the RR interval rastergram and the respiration wave meet the sleep apnea characteristics in the same time period, it is determined as a high-confidence sleep apnea event:

[0134] Condition A: There are three or more consecutive sleep apnea OSAS events in the RR interval rastergram;

[0135] Condition B: There are three or more consecutive sleep apnea OSAS events in the respiration wave;

[0136] If conditions A and B are met, it is determined as a high-confidence sleep apnea event.

[0137] In addition, if three or more sleep apnea events are detected in a time period by one signal (RR interval rastergram or respiration wave), but three or more sleep apnea events are not detected by the other signal (respiration wave or RR interval rastergram), secondary screening is performed:

[0138] Case 1: Three or more sleep apnea events are detected by the RR interval grid but three or more sleep apnea events are not detected by the breath wave; check the data of the previous and subsequent 3 minutes of this period, if there are two sleep apnea OSAS events (events detected by RR interval grid analysis or events detected by breath wave signal analysis), it is still determined as a high-confidence sleep apnea event;

[0139] Case 2: Three or more sleep apnea events are detected by the breath wave but three or more sleep apnea events are not detected by the RR interval grid; check the data of the previous and subsequent 3 minutes, if there are two sleep apnea OSAS events (events detected by RR interval grid analysis or events detected by breath wave signal analysis), it is still determined as a high-confidence sleep apnea event.

[0140] (5) The deployment mode of the sleep apnea syndrome automatic detection device is exemplarily illustrated as follows.

[0141] Exemplarily, the sleep apnea syndrome automatic detection device can be divided into one terminal device end (such as a wearable device), and the signal acquisition module, the data preprocessing module, the feature extraction module and the fusion decision module are deployed on the terminal device end.

[0142] Exemplarily, the sleep apnea syndrome automatic detection device can also be two terminal device ends (such as a wearable device and a smart phone), and the two terminal device ends are in communication connection, the signal acquisition module is deployed on the first terminal device end (such as a wearable device), and the data preprocessing module, the feature extraction module and the fusion decision module are deployed on the second terminal device (such as a smart phone).

[0143] Exemplarily, the sleep apnea syndrome automatic detection device can be divided into a local device end (such as a wearable device) and a cloud server, and the local device end and the cloud server are in communication connection, the signal acquisition module is deployed on the local device end, and the data preprocessing module, the feature extraction module and the fusion decision module are deployed on the cloud server.

[0144] The application adopts a scheme combining dynamic electrocardiogram signal RR interval and breath wave to realize automatic screening and risk grading of sleep apnea syndrome. High-confidence sleep apnea events can be detected. The system can automatically identify four levels of normal, mild, moderate and severe suspicious according to the duration of sleep apnea events (the longer the time, the more serious), and combined with doctor's review, moderate and severe high-risk patients are recommended to the respiratory department for further diagnosis and treatment.

[0145] It can be understood that various digital numbers involved in the embodiments of the present application are only distinguished for convenience of description, and are not used to limit the scope of the embodiments of the present application.

[0146] It should be understood that the above description is only the preferred embodiment of the application, and is not intended to limit the application. Any modification, equivalent replacement and improvement made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A sleep apnea syndrome automatic detection device characterized by comprising: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

2. The apparatus of claim 1, wherein the sleep apnea syndrome automatic detection apparatus further comprises: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

3. The apparatus of claim 2, wherein the means for determining comprises means for determining a sleep apnea hypopnea index. The method comprises the following steps:

4. The apparatus of claim 1, wherein the sleep apnea syndrome automatic detection apparatus further comprises: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the The RR interval signal feature extraction submodule is configured to remove high-frequency noise in the signal, retain sleep apnea features, and determine a sleep apnea event represented by the RR interval signal based on the sleep apnea features in the RR interval signal. The respiratory wave signal feature extraction submodule is configured to remove high-frequency noise in the signal, retain sleep apnea features, and determine a sleep apnea event represented by the respiratory wave signal based on the sleep apnea features in the respiratory wave signal.

5. The apparatus of claim 4, wherein the means for determining comprises means for determining a sleep apnea hypopnea index. The removing high-frequency noise in the signal and retaining sleep apnea features comprises: The signal is decomposed by five orders using db4 wavelet transform to obtain D1, D2, D3, D4, D5 and A5, D1 is a detail coefficient obtained by first layer decomposition, D2 is a detail coefficient obtained by second layer decomposition, D3 is a detail coefficient obtained by third layer decomposition, D4 is a detail coefficient obtained by fourth layer decomposition, D5 is a detail coefficient obtained by fifth layer decomposition, and A5 is an approximation coefficient obtained by fifth layer decomposition. The signal is reconstructed based on D3, D4 and D5 to retain sleep apnea features.

6. The apparatus of claim 4, wherein the means for determining comprises means for determining a sleep apnea hypopnea index. The determining a sleep apnea event represented by the RR interval signal based on the sleep apnea features in the RR interval signal comprises: determining peaks and troughs in the RR interval signal; determining a peak interval between two adjacent peaks and a trough interval between two adjacent troughs in the RR interval signal based on the peaks and troughs in the RR interval signal; determining a start and end point of a sleep apnea event based on a variation coefficient threshold, a variation coefficient of the peak interval and a variation coefficient of the trough interval.

7. The apparatus of claim 6, wherein the means for determining comprises means for determining a sleep apnea hypopnea index. The determining a start and end point of a sleep apnea event based on a variation coefficient threshold, a variation coefficient of the peak interval and a variation coefficient of the trough interval comprises: for three adjacent peaks, if the variation coefficient of the peak interval continuously exceeds the variation coefficient threshold and the duration is greater than a time threshold, a start point of the sleep apnea event is marked; or for three adjacent troughs, if the variation coefficient of the trough interval continuously exceeds the variation coefficient threshold and the duration is greater than a time threshold, a start point of the sleep apnea event is marked; after the start point of the sleep apnea event is marked, if the variation coefficients of the peak interval and the trough interval continuously are less than a baseline value and the duration is greater than a time threshold, an end point of the sleep apnea event is marked, and the baseline value is less than the variation coefficient threshold.

8. The apparatus of claim 4, wherein the means for determining comprises means for determining a sleep apnea hypopnea index (SAHI) of the patient. The determining a sleep apnea event represented by the respiratory wave signal based on the sleep apnea features in the respiratory wave signal comprises: calculating a respiratory wave amplitude change rate using a sliding window; determining a start and end point of a sleep apnea event based on the respiratory wave amplitude change rate and a baseline amplitude.

9. The apparatus of claim 8, wherein the means for determining comprises means for determining a sleep apnea hypopnea index. The determining a start and end point of a sleep apnea event based on the respiratory wave amplitude change rate and a baseline amplitude comprises: if the respiratory wave amplitude change rate continuously is lower than a first preset percentage of the baseline amplitude and the duration is greater than a time threshold, a start point of the sleep apnea event is marked; In calibrating the starting point of the sleep apnea event, if the respiratory wave amplitude change rate continues to be higher than the second preset percentage of the baseline amplitude, and the duration is greater than the time threshold, the end point of the sleep apnea event is calibrated, and the first preset percentage is less than the second preset percentage.

Images