CN111166294A - Automatic sleep apnea detection method and device based on inter-heartbeat period - Google Patents
Automatic sleep apnea detection method and device based on inter-heartbeat period Download PDFInfo
- Publication number
- CN111166294A CN111166294A CN202010077427.7A CN202010077427A CN111166294A CN 111166294 A CN111166294 A CN 111166294A CN 202010077427 A CN202010077427 A CN 202010077427A CN 111166294 A CN111166294 A CN 111166294A
- Authority
- CN
- China
- Prior art keywords
- rate variability
- heart rate
- heartbeat interval
- sleep apnea
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4806—Sleep evaluation
- A61B5/4818—Sleep apnoea
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/0205—Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
- A61B5/02405—Determining heart rate variability
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
- A61B5/0245—Detecting, measuring or recording pulse rate or heart rate by using sensing means generating electric signals, i.e. ECG signals
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
Abstract
The invention provides a sleep apnea automatic detection method and device based on an inter-heartbeat period, which are used for solving the problems of inaccurate sleep apnea detection and low precision in the prior art. The sleep apnea automatic detection method comprises the steps of firstly collecting heartbeat interval information during sleep, then automatically extracting features from the heartbeat interval information through a residual error neural network, further extracting heart rate variability features, and then fusing the automatically extracted features and the heart rate variability feature set, so that whether apnea occurs or not is judged. According to the invention, by extracting the heartbeat interval characteristics in the ECG signal and deeply analyzing the characteristics in the residual error neural network, the heart rate variability characteristics are combined, and all weights in the residual error neural network can be adjusted in a fine way clinically, so that the flexibility, the accuracy and the precision of sleep detection are improved; meanwhile, only single-guide electrocardio information is needed, the acquisition process is simple and convenient, and the method has considerable universality.
Description
Technical Field
The invention belongs to the field of medical data monitoring, and particularly relates to a sleep apnea automatic detection method and device based on heartbeat intervals.
Background
Sleep Apnea (SA) is a common respiratory condition that leads to Sleep disorders. In rapid eye movement sleep, the airway is completely blocked by the genioglossus muscle, tendons and adipose tissue for at least 10 seconds, or a hypopnea event occurs for 10 seconds with a reduction in airflow of more than 50% and oxygen desaturation of more than 3%, all known as apnea. According to the survey, sleep apnea has affected people worldwide for over 30 years, and the number of patients has reached 6% of the worldwide adults in 2008. An increasing number of apnea sufferers are at risk for cardiovascular disease as well as clinical depression due to lack of timely diagnosis and intervention. The risk of apnea has been studied in a number of experiments and results have shown a strong correlation between apnea and several endogenous physiological phenomena and diseases.
Sleep apnea is long-term, harmful, but can be treated after diagnosis. Therapies such as Positive Airway Pressure (PAP) therapy and Palatopharyngoplasty (PPP) therapy are effective in early diagnosis. Therefore, a timely diagnosis is crucial to the therapeutic process of apnea. Traditionally, apnea syndrome severity has been diagnosed by Polysomnogram (PSG) signals, but when collecting PSG signals to detect apnea, the patient is required to sleep overnight with an invasive device, also requiring medical expertise and a clinical environment. Such conventional apnea detection procedures are overly complex and expensive, and therefore new methods are needed that can be performed quickly and accurately on non-invasive devices.
One effective solution to detect sleep apnea in the prior art is to use a single lead Electrocardiogram (ECG) signal to detect apnea, typically by manually designing various functions. For example, by extracting features from the ECG signal, classification can be performed in various classifiers such as proximity algorithms (kNN), Support Vector Machines (SVM), etc., or automatic feature extraction and classification can be performed through a deep neural network. But ECG signals sampled around 100Hz limit the depth of the neural network, thereby affecting the accuracy and precision of sleep breath detection. Therefore, the existing sleep respiration detection method has certain defects in the aspects of user experience, flexibility and accuracy.
Disclosure of Invention
The embodiment of the invention provides a sleep apnea automatic detection method and device based on a heartbeat interval.
In order to achieve the above object, the embodiments of the present invention adopt the following technical solutions.
In a first aspect, an embodiment of the present invention provides a method for automatically detecting sleep apnea based on an inter-heartbeat period, where the method includes the following steps:
step S1, collecting a human body Electrocardiogram (ECG) signal during sleeping, extracting heartbeat interval information of adjacent electrocardios according to R waves in five waves of an electrocardio cycle PQRST, and generating a heartbeat interval time sequence;
step S2, automatically extracting the characteristics of the heartbeat interval time sequence based on a preset residual error neural network to obtain a first heart rate variability characteristic;
step S3, extracting heart rate variability features based on the heartbeat interval time sequence to obtain a second heart rate variability feature;
and step S4, performing feature fusion on the first heart rate variability feature and the second heart rate variability feature set automatically extracted by the residual error neural network, inputting the first heart rate variability feature and the second heart rate variability feature set into a classifier, and judging whether apnea occurs.
Optionally, in step S1, a heartbeat interval time sequence is generated, and further, the extracted heartbeat interval information is resampled with a frequency of 2Hz by using a linear interpolation value to generate the heartbeat interval time sequence.
Optionally, in step S2, automatically extracting features of the time series of heartbeat periods, further including: and training the residual error neural network by using a Convolutional Neural Network (CNN) and a Back Propagation (BP) algorithm to finish automatic feature extraction.
Optionally, the convolutional neural network is composed of 33 layers of one-dimensional convolutional layers, and is combined with a batch normalization layer, a dropout layer and a ReLu function layer to form 16 residual blocks.
Optionally, in the ReLu function layer, the one-dimensional convolutional layer is a core of the CNN, an input vector of the activation function h is X, an output is an input of a next layer, and an output Y of the convolutional layer of the l-th layer is Y(l) convComprises the following steps:
in the formula (1), the reaction mixture is,
is a convolution calculation, B is a bias matrix, and W is a weight vector having a fixed size.
Optionally, an average pooling layer is employed between the convolutional layers.
Optionally, a short-circuit connection is established in the 16 residual blocks, the input of the residual block is added to the output of the residual block, and projection is performed in the short-circuit connection so that the input size and the output size are matched, thereby completing automatic extraction of the neural network features.
Optionally, the step S3 includes rate variability features including: mean values of the intervals of the heart beats, standard deviations of the intervals of the heart beats, skewness of the intervals of the heart beats, kurtosis of the intervals of the heart beats, NN50 metric values, pNN50 index, standard deviations of the differences between adjacent intervals of the heart beats, square root of the mean squared difference of the differences between adjacent intervals of the heart beats, alan factor, very low frequencies of the intervals of the heart beats, high frequencies of the intervals of the heart beats, and a ratio of low frequencies LF and high frequencies HF of the intervals of the heart beats.
Optionally, the determining in step S4 whether apnea occurs further includes:
the first heart rate variability features automatically extracted by the residual error neural network and the second heart rate variability features manually extracted are fused into a feature vector and input into a layer of fully-connected neural network classifier; the feature vector passes through the classifier to obtain an unnormalized discrimination value x, and the discrimination value is converted into a pseudo probability that the patient has sleep apnea in the current time period through an output function sigmoid (x); when the discrimination value x is not negative, the pseudo probability is more than or equal to 50 percent, the occurrence of sleep apnea of the patient in the current time period is judged, and the label value is predicted
Equal to 1; when the discrimination value x is negative, the pseudo probability is less than 50%, the patient is judged not to have sleep apnea in the current time period, and the label value is predicted
Equal to 0.
In a second aspect, an embodiment of the present invention provides an automatic sleep apnea detecting apparatus based on an inter-heartbeat period, where the automatic sleep apnea detecting apparatus includes: the device comprises a heartbeat interval information acquisition module, a depth feature extraction module, a heart rate variability feature extraction module and a sleep apnea judgment module; wherein the content of the first and second substances,
the heartbeat interval information acquisition module is simultaneously connected with the depth characteristic extraction module and the heart rate variability characteristic extraction module and is used for acquiring a human electrocardiogram ECG signal during sleeping, extracting heartbeat interval information of adjacent electrocardios according to R waves in five waves of an electrocardio period PQRST, generating a heartbeat interval time sequence and sending the heartbeat interval time sequence to the depth characteristic extraction module and the heart rate variability characteristic extraction module;
the depth feature extraction module is used for automatically extracting features according to the generated heartbeat interval time sequence and based on a preset residual error neural network, and sending the automatically extracted features to the sleep apnea judgment module;
the heart rate variability feature extraction module is used for extracting heart rate variability features according to the generated heartbeat interval time sequence and sending the heart rate variability features to the sleep apnea judgment module;
the sleep apnea judging module is simultaneously connected with the depth feature extracting module and the heart rate variability feature extracting module and used for judging whether sleep apnea occurs or not according to the received automatic extracted features and the heart rate variability features.
The invention has the following beneficial effects: according to the sleep apnea automatic detection method and device based on the heartbeat interval, provided by the embodiment of the invention, the heartbeat interval information during sleep is collected firstly, then the characteristics of the heartbeat interval information are automatically extracted through a residual error neural network, the heart rate variability characteristics are further extracted, and then the automatically extracted characteristics and the heart rate variability characteristic set are fused, so that whether apnea occurs or not is judged, the flexibility is high, and the accuracy and the detection precision of sleep apnea detection are improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flowchart illustrating an embodiment of a method for sleep apnea detection based on inter-heartbeat intervals;
FIG. 2 is a flowchart illustrating a residual neural network calculation in the method for automatic sleep apnea detection according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an automatic sleep apnea detecting device based on inter-heartbeat intervals according to an embodiment of the present invention.
Detailed Description
The technical problems, aspects and advantages of the invention will be explained in detail below with reference to exemplary embodiments. The following exemplary embodiments are merely illustrative of the present invention and are not to be construed as limiting the invention. It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
According to the embodiment of the invention, the sleep apnea is detected by extracting the characteristic of Heartbeat Interval (HI) in an electrocardiogram ECG signal and deeply analyzing the characteristic in a Residual neural Network (RN). The residual neural network, which combines the Heart Rate Variability (HRV), is a depth algorithm applied to wearable devices or smart phones by using network compression technology, and all weights in the residual neural network can be finely adjusted, so that the residual neural network has great flexibility compared with a manually designed function.
In order to facilitate understanding of the embodiments of the present invention, the following description will be further explained by taking several specific examples as examples with reference to the drawings, and the embodiments do not limit the technical solutions of the present invention.
First embodiment
The present embodiment provides a method for automatically detecting sleep apnea based on inter-heartbeat, and fig. 1 is a schematic flow chart of the method for automatically detecting sleep apnea. As shown in fig. 1, the method for automatically detecting sleep apnea based on heartbeat interval comprises the following steps:
and step S1, collecting a human body Electrocardiogram (ECG) signal during sleeping, extracting heartbeat interval information of adjacent electrocardios according to R waves in five waves of an electrocardio cycle PQRST, and generating a heartbeat interval time sequence.
In the step, the Electrocardiogram (ECG) signals of the human body during sleeping can be acquired by simple and comfortable equipment such as mobile equipment. Preferably, the sampling frequency is 100 Hz. And generating a heartbeat interval time sequence, and further resampling the extracted heartbeat interval information by adopting a 2Hz frequency by using a linear interpolation value to generate the heartbeat interval time sequence.
And step S2, automatically extracting the features of the heartbeat interval time sequence based on a preset residual error neural network to obtain a first heart rate variability feature.
In this step, the automatic feature extraction is performed on the heart beat period time sequence, which further includes: and training the residual error neural network by using a Convolutional Neural Network (CNN) and a Back Propagation (BP) algorithm to complete automatic feature extraction.
Fig. 2 is a flowchart of the residual neural network calculation according to the present embodiment. And obtaining a first heart rate variability characteristic through depth characteristic extraction. As shown in fig. 2, after a heartbeat interval time sequence with a duration of 3 minutes is input, firstly, extracting 128 feature maps (feature maps) through 5 one-dimensional convolution layers with a size of 1 × 20, then adding a Batch Normalization (BN) layer for normalization and activating a ReLu function layer, and then performing average pooling through an average pooling layer (averaging) with a size of 1 × 2; next, 16 identical residual blocks are connected in sequence, each residual block is added with the self after the transmitted feature maps pass through 2 one-dimensional convolution layers with the size of 1 x 3, 2 BN layers and 2 Relu function layers, and the short-circuit connection is used for adding the input and the output of the residual blocks; finally, the obtained 512 feature maps are the first heart rate variability feature extracted automatically after passing through a convolution layer with the size of 1 × 3 and an averaging po _ ling layer.
The one-dimensional convolutional layer is the core of the CNN, in the convolutional layer, a weight vector W with a fixed size is multiplied by input data item by item, that is, convolution operation is performed, and the convolution result and a bias term B form the input of an activation function h. The output of the activation function h will be the input to the next layer. If the input vector is X, the output Y of the first convolutional layer is:
in the formula (1), the reaction mixture is,
is a convolution calculation, B is a bias matrix, and W is a weight vector having a fixed size.
In this embodiment, all activation functions h in the convolutional layer are ReLU functions to speed up training and reduce the final error rate in the neural network.
The ReLU function for input x is:
hRelU(x)＝max(0,x) (2)
there are two types of pooling layers in CNN: an average pooling layer and a maximum pooling layer. The max-pooling layer is used to reduce mean shift errors due to poor weight initialization, or using Xavier initialization in the convolutional layer, may also reduce mean shift errors. Averaging the pooling layers can reduce the standard deviation error in the convolutional layer output. Noise caused by measurement errors and other physiological mechanisms may result in standard deviation errors of the inter-beat intervals. In addition, beat interval correction may also be performed using an averaging filter. In this embodiment, an average pooling layer is used between convolutional layers to reduce the standard deviation error of heartbeat intervals caused by measurement errors and other physiological mechanisms in convolutional layer output.
A batch normalization layer (BN layer) for reducing internal covariate shifts in the residual neural network, normalizing the standard deviation Var of input X and the expected E:
in equation (3), E (X) and Var (X) are the expectation and deviation of input X, X is the input sample, and E is a smoothing term to avoid zeros, which is typically set to a negligible very small positive number. A bulk normalization layer (BN layer) in a neural network may implement the function of data whitening.
In 16 residual blocks of a residual neural network, each residual block is composed of two one-dimensional convolution layers, two BN layers and two Relu function layers, and its output Y is defined by an input X:
Y＝F(X,{W}) (4)
in equation (4), F represents the mapping of the block and the set { W } represents all the parameters in the block, resulting in the output Y of the residual blockrComprises the following steps:
Yr＝Y+X＝F(X,{W})+X (5)
when the dimension of Y is greater than X, X is projected to maintain the same dimension of both matrices.
Calculating the ith parameter W by BP algorithmiGradient G ofiThe method comprises the following steps:
when G isiEqual to zero time, the value of,
possibly zero; when the gradient equals zero, the convolutional layer stops training.
Gradients in residual blocks
Is that
In the 16 residual blocks, a short-circuit connection is established, the input of the residual block is added to its output, and the projection is performed in the short-circuit connection so that the input size matches the output size.
Step S3, performing heart rate variability feature extraction based on the heartbeat interval time sequence to obtain a second heart rate variability feature.
In this step, the heart rate variability features are small variations of the heart rate between successive heartbeat cycles, and also small variations between heartbeat intervals, and the heart rate variability information can be obtained by performing linear and nonlinear data feature extraction on the heartbeat intervals.
In the residual neural network calculation process shown in fig. 2, a heartbeat interval time sequence with a duration of 3 minutes is input, and artificial feature extraction is performed for each minute, so that the same feature can be calculated to obtain three results, and finally a second heart rate variability feature is obtained as shown in fig. 2, where the second heart rate variability feature includes: mean values of the intervals of the heart beats, standard deviations of the intervals of the heart beats, skewness of the intervals of the heart beats, kurtosis of the intervals of the heart beats, NN50 metric values, pNN50 index, standard deviations of the differences between adjacent intervals of the heart beats, square root of the mean squared difference of the differences between adjacent intervals of the heart beats, alan factor, very low frequencies of the intervals of the heart beats, high frequencies of the intervals of the heart beats, and a ratio of low frequencies LF and high frequencies HF of the intervals of the heart beats. All of the second heart rate variability features form a feature vector.
Said second heart rate variability feature extraction based on a heartbeat interval time sequence further comprising the steps of:
step S301, calculating the average value of heartbeat intervals (Mean of RR intervals, Mean RR);
step S302, calculating Standard deviation of inter-beat intervals (SDRR);
step S303, calculating the Skewness of the heartbeat interval (Skewness of RR intervals, Skewness RR);
step S304, calculating the peak of the heartbeat interval (Kurtosis of RR intervals, Kurtosis RR);
step S305, calculating an NN50 metric (variant 1), wherein the NN50 metric (variant 1) is the number of interval pairs of adjacent heartbeat interval pairs, the duration of the first heartbeat interval of each adjacent heartbeat interval pair is greater than that of the second heartbeat interval by at least 50ms (number of patient ad jacent normal to normal intervals differential by more than 50ms NN 50);
step S306, calculating an NN50 metric value (variant 2), the NN50 metric value (variant 2) being the number of pairs of adjacent heartbeat intervals for which the second heartbeat interval exceeds the first heartbeat interval by more than 50 ms;
step S307, calculating the pNN50 index of variant 1, dividing each NN50 index by the total number of heartbeat intervals (Percentof NN50in the total number of RR intervals, PNN 50);
step S308, calculating pNN50 index for variant 2, dividing each NN50 index by total number of heartbeat intervals;
step S309, calculating Standard deviation of Difference between adjacent heartbeat intervals (SDSD);
step S310, calculating The square root of The square average of The difference between adjacent heartbeat intervals (The root mean square of difference between adjacencies NNintervals, R MSSD);
step S311, calculate Allan factor A (T) to evaluate on time scales 5,10,15,20 and 25S
In the formula (8), Ni(T) is the number of QRS detection points in a window of length T, extending from iT to (i +1) T, E is the desired operator;
step S312, calculating Very Low Frequencies (VLFs) of the heartbeat interval;
step S313, calculating a Low Frequency (LF) of the heartbeat interval;
step S314, calculating the High Frequency (HF) of the heartbeat interval;
in step S315, the ratio of LF to HF of the heartbeat interval is calculated.
And step S4, performing feature fusion on the first heart rate variability feature and the second heart rate variability feature set automatically extracted by the residual error neural network, inputting the first heart rate variability feature and the second heart rate variability feature set into a classifier, and judging whether apnea occurs.
The step is to fuse the first heart rate variability feature automatically extracted by the residual error neural network and the second heart rate variability feature manually extracted into a feature vector and input the feature vector into a classifier. The feature vector passes through a layer of fully-connected neural network classifier to obtain an unnormalized discrimination value x, the discrimination value is converted into the pseudo probability that the patient has sleep apnea in the current time period through an output function sigmoid (x), the value range is between 0 and 100 percent, and the sigmoid function of the input x is expressed as follows:
when training, when the true label is y and the predicted label is
The binary cross entropy loss function is:
predicting labels based on pseudo-probabilities
The calculation method comprises the following steps:
when the discrimination value x is not negative, the pseudo probability is more than or equal to 50 percent, the occurrence of sleep apnea of the patient in the current time period is judged, and the label value is predicted
Equal to 1; when the discrimination value x is negative, the pseudo probability is less than 50%, the patient is judged not to have sleep apnea in the current time period, and the label value is predicted
Equal to 0.
According to the technical scheme, the sleep apnea automatic detection method based on the heartbeat interval comprises the steps of firstly collecting heartbeat interval information during sleep, then automatically extracting a first heart rate variability feature from the heartbeat interval information through a residual error neural network, further extracting a second heart rate variability feature, and then fusing the first heart rate variability feature and the second heart rate variability feature set, so that whether apnea occurs or not is judged. The automatic sleep apnea detection method only needs single-lead electrocardiographic information, is simple and convenient in acquisition process, accurately detects whether sleep apnea occurs or not, and improves flexibility, accuracy and precision of sleep detection.
Second embodiment
The embodiment provides an automatic sleep apnea detecting device based on an inter-heartbeat period, and fig. 3 is a schematic structural diagram of the automatic sleep apnea detecting device. As shown in fig. 3, the automatic sleep apnea detecting apparatus includes: the device comprises a heartbeat interval information acquisition module, a depth feature extraction module, a heart rate variability feature extraction module and a sleep apnea judgment module; wherein the content of the first and second substances,
the heartbeat interval information acquisition module is simultaneously connected with the depth characteristic extraction module and the heart rate variability characteristic extraction module and is used for acquiring a human electrocardiogram ECG signal during sleeping, extracting heartbeat interval information of adjacent electrocardios according to R waves in five waves of an electrocardio cycle PQRST, generating a heartbeat interval time sequence and sending the heartbeat interval time sequence to the automatic characteristic extraction module and the heart rate variability characteristic extraction module;
the depth feature extraction module is used for automatically extracting features according to the generated heartbeat interval time sequence and based on a preset residual error neural network to obtain a first heart rate variability feature, and sending the automatically extracted features to the sleep apnea judgment module;
the heart rate variability feature extraction module is used for extracting heart rate variability features based on the heartbeat interval time sequence, obtaining second heart rate variability features and sending the heart rate variability features to the sleep apnea judgment module;
the sleep apnea judging module is simultaneously connected with the depth feature extracting module and the heart rate variability feature extracting module and used for judging whether sleep apnea occurs or not according to the received automatic extracted features and the heart rate variability features.
The device for automatically detecting sleep apnea based on cardiac intervals in this embodiment is a technical solution corresponding to the method for automatically detecting sleep apnea based on cardiac intervals in the first embodiment, the same or similar parts in the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the device or system embodiments, since they are substantially similar to the method embodiments, they are described simply, and related parts refer to part of the description of the method embodiments and are not repeated herein.
According to the technical scheme, the sleep apnea automatic detection device based on the heartbeat interval is suitable for wide audience groups, has universality, provides accurate and rapid disease auxiliary diagnosis services for patients, and can timely discover the endogenous disease hidden danger of cardiovascular and cerebrovascular diseases.
While the foregoing is directed to the preferred embodiment of the present invention, it is understood that the invention is not limited to the exemplary embodiments disclosed, but is made merely for the purpose of providing those skilled in the relevant art with a comprehensive understanding of the specific details of the invention. It will be apparent to those skilled in the art that various modifications and adaptations of the present invention can be made without departing from the principles of the invention and the scope of the invention is to be determined by the claims.
Claims (10)
1. A method for automatically detecting sleep apnea based on a heartbeat interval, the method comprising the steps of:
step S1, collecting a human body Electrocardiogram (ECG) signal during sleeping, extracting heartbeat interval information of adjacent electrocardios according to R waves in five waves of an electrocardio cycle PQRST, and generating a heartbeat interval time sequence;
step S2, automatically extracting the characteristics of the heartbeat interval time sequence based on a preset residual error neural network to obtain a first heart rate variability characteristic;
step S3, extracting heart rate variability features based on the heartbeat interval time sequence to obtain a second heart rate variability feature;
and step S4, performing feature fusion on the first heart rate variability feature and the second heart rate variability feature set automatically extracted by the residual error neural network, inputting the first heart rate variability feature and the second heart rate variability feature set into a classifier, and judging whether apnea occurs.
2. The method according to claim 1, wherein the step S1 generates a heartbeat interval time series, and further wherein the extracted heartbeat interval information is resampled with a frequency of 2Hz using linear interpolation values to generate the heartbeat interval time series.
3. The method for automatically detecting sleep apnea of claim 1, wherein in step S2, the automatic feature extraction is performed on the heart beat time sequence, and further comprising: and training the residual error neural network by using a Convolutional Neural Network (CNN) and a Back Propagation (BP) algorithm to finish automatic feature extraction.
4. The method according to claim 3, wherein the convolutional neural network comprises 33 layers of one-dimensional convolutional layers, and is combined with the batch normalization layer, the dropout layer and the ReLu function layer to form 16 residual blocks.
5. The method of claim 4, wherein the ReLu function layer is a core of CNN, and the one-dimensional convolution layer is a core of CNNThe input vector of the activation function h is X, the output is the input of the next layer, the output Y of the convolution layer of the first layer(l) convComprises the following steps:
in the formula (1), the reaction mixture is,
is a convolution calculation, B is a bias matrix, and W is a weight vector having a fixed size.
6. The method of claim 4, wherein an average pooling layer is employed between the convolutional layers.
7. The method of claim 4, wherein a short-circuit connection is established among the 16 residual blocks, the input of the residual block is added to the output of the residual block, and projection is performed in the short-circuit connection to match the input size with the output size, thereby completing automatic extraction of neural network features.
8. The method of claim 1, wherein the step S3 center rate variability features comprise: mean values of the intervals of the heart beats, standard deviations of the intervals of the heart beats, skewness of the intervals of the heart beats, kurtosis of the intervals of the heart beats, NN50 metric values, pNN50 index, standard deviations of the differences between adjacent intervals of the heart beats, square root of the mean squared difference of the differences between adjacent intervals of the heart beats, alan factor, very low frequencies of the intervals of the heart beats, high frequencies of the intervals of the heart beats, and a ratio of low frequencies LF and high frequencies HF of the intervals of the heart beats.
9. The method for automatically detecting sleep apnea of claim 1, wherein said determining in step S4 whether apnea occurs further comprises:
the first heart rate variability features automatically extracted by the residual error neural network and the second heart rate variability features manually extracted are fused into a feature vector and input into a layer of fully-connected neural network classifier; the feature vector passes through the classifier to obtain an unnormalized discrimination value x, and the discrimination value is converted into a pseudo probability that the patient has sleep apnea in the current time period through an output function sigmoid (x); when the discrimination value x is not negative, the pseudo probability is more than or equal to 50 percent, the occurrence of sleep apnea of the patient in the current time period is judged, and the label value is predicted
Equal to 1; when the discrimination value x is negative, the pseudo probability is less than 50%, the patient is judged not to have sleep apnea in the current time period, and the label value is predicted
Equal to 0.
10. An automatic sleep apnea detecting device based on a heartbeat interval, comprising: the device comprises a heartbeat interval information acquisition module, a depth feature extraction module, a heart rate variability feature extraction module and a sleep apnea judgment module; wherein the content of the first and second substances,
the heartbeat interval information acquisition module is simultaneously connected with the depth characteristic extraction module and the heart rate variability characteristic extraction module and is used for acquiring a human electrocardiogram ECG signal during sleeping, extracting heartbeat interval information of adjacent electrocardios according to R waves in five waves of an electrocardio cycle PQRST, generating a heartbeat interval time sequence and sending the heartbeat interval time sequence to the automatic characteristic extraction module and the heart rate variability characteristic extraction module;
the depth feature extraction module is used for automatically extracting features according to the generated heartbeat interval time sequence and based on a preset residual error neural network to obtain a first heart rate variability feature, and sending the automatically extracted features to the sleep apnea judgment module;
the heart rate variability feature extraction module is used for extracting heart rate variability features based on the heartbeat interval time sequence, obtaining second heart rate variability features and sending the heart rate variability features to the sleep apnea judgment module;
the sleep apnea judging module is simultaneously connected with the depth feature extracting module and the heart rate variability feature extracting module and used for judging whether sleep apnea occurs or not according to the received automatic extracted features and the heart rate variability features.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010077427.7A CN111166294B (en) | 2020-01-29 | 2020-01-29 | Automatic sleep apnea detection method and device based on inter-heartbeat period |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010077427.7A CN111166294B (en) | 2020-01-29 | 2020-01-29 | Automatic sleep apnea detection method and device based on inter-heartbeat period |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111166294A true CN111166294A (en) | 2020-05-19 |
| CN111166294B CN111166294B (en) | 2021-09-14 |
Family
ID=70620740
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010077427.7A Active CN111166294B (en) | 2020-01-29 | 2020-01-29 | Automatic sleep apnea detection method and device based on inter-heartbeat period |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111166294B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112426147A (en) * | 2020-10-21 | 2021-03-02 | 华南师范大学 | Sleep respiratory event detection model processing method, system and storage medium |
| CN113180688A (en) * | 2020-12-14 | 2021-07-30 | 上海交通大学 | Coronary heart disease electrocardiogram screening system and method based on residual error neural network |
| CN113243890A (en) * | 2021-05-10 | 2021-08-13 | 清华大学深圳国际研究生院 | Sleep apnea syndrome recognition device |
| CN113450901A (en) * | 2020-05-28 | 2021-09-28 | 孙羽 | Control method and device of respiratory support system and respiratory support system |
| CN114073495A (en) * | 2021-12-03 | 2022-02-22 | 中山大学·深圳 | Sleep apnea detection method and device based on information similarity |
| TWI837948B (en) | 2022-11-15 | 2024-04-01 | 國立勤益科技大學 | Method for fully detecting apnea and hypopnea events |
Citations (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6658287B1 (en) * | 1998-08-24 | 2003-12-02 | Georgia Tech Research Corporation | Method and apparatus for predicting the onset of seizures based on features derived from signals indicative of brain activity |
| US20050054940A1 (en) * | 2003-04-23 | 2005-03-10 | Almen Adam J. | Apparatus and method for monitoring heart rate variability |
| US20080051669A1 (en) * | 2006-08-28 | 2008-02-28 | Wolfgang Meyer | Diagnosis of sleep apnea |
| CN101816822A (en) * | 2010-05-27 | 2010-09-01 | 天津大学 | Setting method of functional electrical stimulation PID (Proportion Integration Differentiation) parameter double source characteristic fusion particle swarm |
| US20130046151A1 (en) * | 2011-02-14 | 2013-02-21 | The Board Of Regents Of The University Of Texas System | System and method for real-time measurement of sleep quality |
| CN103949010A (en) * | 2014-05-05 | 2014-07-30 | 吉林大学 | Application of high-frequency current for treating OSA |
| CN104545888A (en) * | 2014-12-27 | 2015-04-29 | 迪姆软件(北京)有限公司 | Sleep apnea collecting and analyzing system based on dynamic electrocardiogram and respiratory wave collection |
| CN104636492A (en) * | 2015-03-04 | 2015-05-20 | 浪潮电子信息产业股份有限公司 | Method for classifying dynamic data on basis of fuzzy integral feature fusion |
| CN106343977A (en) * | 2016-09-21 | 2017-01-25 | 中国人民解放军国防科学技术大学 | UAV (unmanned aerial vehicle) operator state monitoring system based on multi-sensor fusion |
| CN106361277A (en) * | 2016-08-26 | 2017-02-01 | 中山大学 | Sleep apnea syndrome assessment method based on electrocardiogram signals |
| CN107126203A (en) * | 2016-02-28 | 2017-09-05 | 南京全时陪伴健康科技有限公司 | Human health information measures display system in real time |
| CN107438398A (en) * | 2015-01-06 | 2017-12-05 | 大卫·伯顿 | Portable wearable monitoring system |
| CN108182974A (en) * | 2017-12-27 | 2018-06-19 | 深圳和而泰数据资源与云技术有限公司 | Illness appraisal procedure, terminal device and computer-readable medium |
| CN108606798A (en) * | 2018-05-10 | 2018-10-02 | 东北大学 | Contactless atrial fibrillation intelligent checking system based on depth convolution residual error network |
| US20180310876A1 (en) * | 2015-05-19 | 2018-11-01 | Shanghai Megahealth Technologies Co., Ltd | Noncontact detection method of sleep stages and sleep-disordered breathing |
| CN109106350A (en) * | 2018-09-20 | 2019-01-01 | 中山大学 | The preliminary screening method of the Sleep Apnea-hypopnea Syndrome of low physiological stress |
| CN109394188A (en) * | 2018-11-27 | 2019-03-01 | 中山大学 | A kind of adnormal respiration detection method, device and equipment based on heart rate variability |
| CN109674474A (en) * | 2018-11-30 | 2019-04-26 | 深圳和而泰智能控制股份有限公司 | Sleep apnea recognition methods, equipment and computer-readable medium |
| EP3485806A1 (en) * | 2017-11-20 | 2019-05-22 | Kinpo Electronics, Inc. | Wearable device capable of detecting sleep apnea event and detection method thereof |
| US20190209776A1 (en) * | 2014-12-22 | 2019-07-11 | Medicus Engineering Aps | Closed-loop control of insulin infusion and system for measuring autonomic nervous system modulation |
| CN110151138A (en) * | 2019-05-29 | 2019-08-23 | 中山大学 | Sleep apnea segment detection method, equipment based on convolutional neural networks |
| CN110309758A (en) * | 2019-06-26 | 2019-10-08 | 广州视源电子科技股份有限公司 | Electrocardiosignal feature extracting method, device, computer equipment and storage medium |
-
2020
- 2020-01-29 CN CN202010077427.7A patent/CN111166294B/en active Active
Patent Citations (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6658287B1 (en) * | 1998-08-24 | 2003-12-02 | Georgia Tech Research Corporation | Method and apparatus for predicting the onset of seizures based on features derived from signals indicative of brain activity |
| US20050054940A1 (en) * | 2003-04-23 | 2005-03-10 | Almen Adam J. | Apparatus and method for monitoring heart rate variability |
| US20080051669A1 (en) * | 2006-08-28 | 2008-02-28 | Wolfgang Meyer | Diagnosis of sleep apnea |
| CN101816822A (en) * | 2010-05-27 | 2010-09-01 | 天津大学 | Setting method of functional electrical stimulation PID (Proportion Integration Differentiation) parameter double source characteristic fusion particle swarm |
| US20130046151A1 (en) * | 2011-02-14 | 2013-02-21 | The Board Of Regents Of The University Of Texas System | System and method for real-time measurement of sleep quality |
| CN103949010A (en) * | 2014-05-05 | 2014-07-30 | 吉林大学 | Application of high-frequency current for treating OSA |
| US20190209776A1 (en) * | 2014-12-22 | 2019-07-11 | Medicus Engineering Aps | Closed-loop control of insulin infusion and system for measuring autonomic nervous system modulation |
| CN104545888A (en) * | 2014-12-27 | 2015-04-29 | 迪姆软件(北京)有限公司 | Sleep apnea collecting and analyzing system based on dynamic electrocardiogram and respiratory wave collection |
| CN107438398A (en) * | 2015-01-06 | 2017-12-05 | 大卫·伯顿 | Portable wearable monitoring system |
| CN104636492A (en) * | 2015-03-04 | 2015-05-20 | 浪潮电子信息产业股份有限公司 | Method for classifying dynamic data on basis of fuzzy integral feature fusion |
| US20180310876A1 (en) * | 2015-05-19 | 2018-11-01 | Shanghai Megahealth Technologies Co., Ltd | Noncontact detection method of sleep stages and sleep-disordered breathing |
| CN107126203A (en) * | 2016-02-28 | 2017-09-05 | 南京全时陪伴健康科技有限公司 | Human health information measures display system in real time |
| CN106361277A (en) * | 2016-08-26 | 2017-02-01 | 中山大学 | Sleep apnea syndrome assessment method based on electrocardiogram signals |
| CN106343977A (en) * | 2016-09-21 | 2017-01-25 | 中国人民解放军国防科学技术大学 | UAV (unmanned aerial vehicle) operator state monitoring system based on multi-sensor fusion |
| EP3485806A1 (en) * | 2017-11-20 | 2019-05-22 | Kinpo Electronics, Inc. | Wearable device capable of detecting sleep apnea event and detection method thereof |
| CN108182974A (en) * | 2017-12-27 | 2018-06-19 | 深圳和而泰数据资源与云技术有限公司 | Illness appraisal procedure, terminal device and computer-readable medium |
| CN108606798A (en) * | 2018-05-10 | 2018-10-02 | 东北大学 | Contactless atrial fibrillation intelligent checking system based on depth convolution residual error network |
| CN109106350A (en) * | 2018-09-20 | 2019-01-01 | 中山大学 | The preliminary screening method of the Sleep Apnea-hypopnea Syndrome of low physiological stress |
| CN109394188A (en) * | 2018-11-27 | 2019-03-01 | 中山大学 | A kind of adnormal respiration detection method, device and equipment based on heart rate variability |
| CN109674474A (en) * | 2018-11-30 | 2019-04-26 | 深圳和而泰智能控制股份有限公司 | Sleep apnea recognition methods, equipment and computer-readable medium |
| CN110151138A (en) * | 2019-05-29 | 2019-08-23 | 中山大学 | Sleep apnea segment detection method, equipment based on convolutional neural networks |
| CN110309758A (en) * | 2019-06-26 | 2019-10-08 | 广州视源电子科技股份有限公司 | Electrocardiosignal feature extracting method, device, computer equipment and storage medium |
Non-Patent Citations (4)
| Title |
|---|
| GOKHAN MEMIS: "《Multimodal Classification of Obstructive Sleep Apnea using Feature Level Fusion》", 《IEEE》 * |
| LEI WANG ET AL: "《A RR interval based automated apnea detection approach using residual network》", 《COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE》 * |
| URTNASAN ERDENEBAYAR ET AL: "《Deep learning approaches for automatic detection of sleep apnea events from an electrocardiogram》", 《COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE》 * |
| 郑慧君: "《利用心率的去趋势波动分析提取睡眠呼吸暂停低通气综合症患者的睡眠结构信息》", 《中国生物医学工程学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113450901A (en) * | 2020-05-28 | 2021-09-28 | 孙羽 | Control method and device of respiratory support system and respiratory support system |
| CN112426147A (en) * | 2020-10-21 | 2021-03-02 | 华南师范大学 | Sleep respiratory event detection model processing method, system and storage medium |
| CN112426147B (en) * | 2020-10-21 | 2023-08-08 | 华南师范大学 | Sleep respiratory event detection model processing method, system and storage medium |
| CN113180688A (en) * | 2020-12-14 | 2021-07-30 | 上海交通大学 | Coronary heart disease electrocardiogram screening system and method based on residual error neural network |
| CN113243890A (en) * | 2021-05-10 | 2021-08-13 | 清华大学深圳国际研究生院 | Sleep apnea syndrome recognition device |
| CN114073495A (en) * | 2021-12-03 | 2022-02-22 | 中山大学·深圳 | Sleep apnea detection method and device based on information similarity |
| CN114073495B (en) * | 2021-12-03 | 2024-03-22 | 中山大学·深圳 | Sleep apnea detection method and device based on information similarity |
| TWI837948B (en) | 2022-11-15 | 2024-04-01 | 國立勤益科技大學 | Method for fully detecting apnea and hypopnea events |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111166294B (en) | 2021-09-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111166294B (en) | Automatic sleep apnea detection method and device based on inter-heartbeat period | |
| Miao et al. | Multi-sensor fusion approach for cuff-less blood pressure measurement | |
| CN109157202B (en) | Cardiovascular disease early warning system based on multi-physiological signal deep fusion | |
| de Chazal et al. | Automated detection of obstructive sleep apnoea at different time scales using the electrocardiogram | |
| US7025729B2 (en) | Apparatus for detecting sleep apnea using electrocardiogram signals | |
| CN108577830B (en) | User-oriented physical sign information dynamic monitoring method and system | |
| KR101656611B1 (en) | Method for obtaining oxygen desaturation index using unconstrained measurement of bio-signals | |
| US11931154B2 (en) | Systems and methods for classifying ECG data | |
| CN109414204A (en) | Method and apparatus for determining the respiration information for object | |
| US20100274147A1 (en) | Detecting ischemia with nonlinear heart rate variability measures | |
| CN107260142A (en) | Physical signs error is reduced using inertial frequency | |
| JP2018502632A (en) | System and method for slow wave sleep detection | |
| US20220167856A1 (en) | Lung function monitoring from heart signals | |
| WO2023087512A1 (en) | Depression objective screening system, method, and device | |
| CN106108844B (en) | A kind of method and apparatus of determining sleep stage | |
| Liu et al. | Cuffless blood pressure measurement using smartwatches: a large-scale validation study | |
| Jung et al. | Oxygen desaturation index estimation through unconstrained cardiac sympathetic activity assessment using three ballistocardiographic systems | |
| US20210204857A1 (en) | Method and device for cardiac monitoring | |
| Podaru et al. | Determination and comparison of heart rate variability and pulse rate variability | |
| CN106108845B (en) | A kind of method and apparatus of determining sleep stage | |
| CN113470805A (en) | Method for establishing blood pressure model | |
| EP4327743A1 (en) | Sleeping state estimation system | |
| TWI733378B (en) | Method of establishing blood pressure model | |
| CN210697620U (en) | Full-automatic wearable cardiovascular health diagnosis bracelet | |
| Hoppenbrouwer et al. | Airflow from nasal pulse oximetry in the screening of obstructive sleep apnea |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |