TWI750531B - Detection device and detection method for apnea based on chest respiratory signal - Google Patents

Abstract

A detection device and a detection method for an apnea based on a chest respiratory signal are provided. The detection device includes a processor, a storage medium, and a transceiver. The transceiver obtains a first chest respiratory signal and a first unipolar ECG corresponding to the chest respiratory signal. The processor accesses and executes a plurality of modules in the storage medium, wherein the plurality of modules include a training module and a detection module. The training module uses the first chest respiratory signal and the first unipolar ECG as training data to train a machine learning model. The detection module obtains, via the transceiver, a second chest respiratory signal and a second unipolar ECG of a subject, and determines, according to the machine learning model, the second chest respiratory signal, and the second unipolar ECG, whether at least one apnea event is happened on the subject.

Description

Apnea detection device and detection method based on chest breathing signal

The present disclosure relates to a detection device and a detection method, and in particular, to a detection device and a detection method for apnea (Apnea) based on chest respiratory signals.

Obstructive sleep apnea (OSA) is a common sleep disorder in which the pharynx collapses during sleep resulting in complete or partial upper airway obstruction, resulting in apnea or reduced symptoms. At present, the main basis for the diagnosis of obstructive sleep apnea is polysomnography (PSG). For PSG examination, subjects must go to a sleep laboratory or sleep center to sleep for one night, under the supervision of a nursing staff, with electrode patches on the head, canthus, chin, heart, and legs, and on the chest. Put an induction belt on your abdomen, a blood oxygen meter on your fingers, a breathing sensor on your nose and mouth, and a sphygmomanometer on your arm to record sleep physiological data throughout the night. However, not all patients have the time to spend the night in a sleep lab or sleep center. Accordingly, how to propose a simplified method for diagnosing sleep disorders is one of the goals of those in the art.

The present disclosure provides a detection device and detection method for apnea based on chest breathing signals, which can use the subject's real-time unipolar electrocardiogram (ECG) and chest breathing signals to determine whether a subject is An apnea event occurred.

The apparatus for detecting apnea based on the chest breathing signal of the present disclosure includes a processor, a storage medium, and a transceiver. The transceiver obtains a first chest breathing signal and a first unipolar lead electrocardiogram corresponding to the first chest breathing signal. The storage medium stores multiple modules. The processor is coupled to the storage medium and the transceiver, and accesses and executes a plurality of modules, wherein the plurality of modules includes a training module and a detection module. The training module uses the first chest breathing signal and the first unipolar lead electrocardiogram as training data to train the machine learning model. The detection module obtains the subject's second chest respiration signal and the second unipolar lead ECG through the transceiver, and determines whether the subject is At least one apnea event occurred.

In an embodiment of the present disclosure, the first chest respiration signal includes a first data set corresponding to an apnea event and a second data set corresponding to a non-apneic event, and the first unipolar lead electrocardiogram includes a corresponding A third data set for apnea events and a fourth data set corresponding to non-apneic events.

In an embodiment of the present disclosure, the above-mentioned detection module determines the severity of the symptoms of obstructive sleep apnea of the subject according to the number of occurrences of at least one apnea event.

In an embodiment of the present disclosure, the detection module determines that the severity is high in response to the number of occurrences of the at least one apnea event being greater than the first threshold, and in response to the occurrence number of the at least one apnea event being less than or equal to the first threshold A threshold value but greater than a second threshold value is judged as medium severity, and the severity is judged as low in response to the number of occurrences of at least one apnea event being less than or equal to the second threshold value.

In an embodiment of the present disclosure, the training module measures a plurality of RR intervals according to the first unipolar lead electrocardiogram, and generates the second training data according to at least one of the plurality of RR intervals and the first chest breathing signal, And the support vector machine model is trained according to the second training data.

In an embodiment of the present disclosure, the above-mentioned detection module determines whether the subject has at least one apnea event according to the support vector machine model, the machine learning model, the second chest breathing signal, and the second unipolar lead electrocardiogram.

In an embodiment of the present disclosure, the above-mentioned training module extracts a plurality of features from a plurality of RR intervals or the first chest breathing signal, and generates the second training data according to the plurality of features.

In an embodiment of the present disclosure, the above-mentioned features are associated with at least one of the following: an average value of RR intervals, a second or third sequence correlation coefficient of RR intervals, and the number of RR interval pairs, wherein RR interval pairs Including two adjacent RR intervals, and the time interval between the two RR intervals exceeds 50 ms, the standard deviation of the two adjacent RR intervals, the normalized very low frequency range power of the RR intervals, the normalized chest respiration signal The normalized very low frequency range power, the normalized low frequency range power of the chest breathing signal, and the normalized high frequency range power of the chest breathing signal.

In an embodiment of the present disclosure, the above-mentioned detection device further includes a wearable device. A wearable device is worn on the subject and is communicatively connected to the transceiver, wherein the wearable device includes a monopolar lead electrode and an accelerometer. A unipolar lead electrode was attached to the subject's chest to measure a second unipolar lead ECG. An accelerometer was placed on the subject's chest to measure the second chest respiration signal.

The method for detecting apnea based on the chest breathing signal of the present disclosure includes: obtaining a first chest breathing signal and a first unipolar lead electrocardiogram corresponding to the first chest breathing signal; The polar lead ECG is used as training data to train the machine learning model; the subject's second chest breathing signal and the second unipolar lead ECG are obtained; and the second chest breathing signal and the second unipolar lead are obtained according to the machine learning model The electrocardiogram determines whether the subject has at least one apnea event.

Based on the above, the present disclosure can determine whether a subject has an apnea event and the severity of OSA symptoms of the subject based on the subject's chest breathing signal and unipolar lead signal.

In order to make the content of the present disclosure more comprehensible, the following specific embodiments are given as examples on which the present disclosure can indeed be implemented. Additionally, where possible, elements/components/steps using the same reference numerals in the drawings and embodiments represent the same or similar parts.

FIG. 1 is a schematic diagram of an apnea detection apparatus 100 based on a chest breathing signal according to an embodiment of the present disclosure, wherein the detection apparatus 100 can determine the subject's breathing according to the chest breathing signal and the unipolar lead electrocardiogram of the subject. Severity of OSA symptoms. The detection device 100 includes a processor 110 , a storage medium 120 and a transceiver 130 . In one embodiment, the detection device 100 further includes a wearable device 140 .

The processor 110 is, for example, a central processing unit (CPU), or other programmable general-purpose or special-purpose micro control unit (micro control unit, MCU), microprocessor (microprocessor), digital signal processing digital signal processor (DSP), programmable controller, application specific integrated circuit (ASIC), graphics processor (graphics processing unit, GPU), arithmetic logic unit (arithmetic logic unit, ALU) , complex programmable logic device (complex programmable logic device, CPLD), field programmable logic gate array (field programmable gate array, FPGA) or other similar elements or a combination of the above elements. The processor 110 may be coupled to the storage medium 120 and the transceiver 130 , and access and execute a plurality of modules and various application programs stored in the storage medium 120 .

The storage medium 120 is, for example, any type of fixed or removable random access memory (random access memory, RAM), read-only memory (ROM), and flash memory (flash memory). , a hard disk drive (HDD), a solid state drive (SSD), or similar components or a combination of the above components for storing a plurality of modules or various application programs executable by the processor 110 . In this embodiment, the storage medium 120 can store a plurality of modules including the training module 121 and the detection module 122, the functions of which will be described later.

The transceiver 130 transmits and receives signals in a wireless or wired manner. Transceiver 130 may also perform operations such as low noise amplification, impedance matching, frequency mixing, up or down frequency conversion, filtering, amplification, and the like. The transceiver 130 can be used to receive the first chest breathing signal as training data and the first monopolar lead electrocardiogram corresponding to the first chest breathing signal, or receive the second chest breathing signal measured from the subject and corresponding to the second chest breathing signal. Second unipolar lead ECG of chest respiratory signals. For example, the transceiver 130 can use, for example, global system for mobile communication (GSM), personal handy-phone system (PHS), code division multiple access (CDMA) systems, wideband code division multiple access (WCDMA) systems, long term evolution (LTE) systems, worldwide interoperability for microwave access (WiMAX) systems, wireless fidelity (wireless) fidelity, Wi-Fi) system or Bluetooth (Bluetooth) and other communication technologies to receive the first unipolar lead electrocardiogram and the first chest respiration signal as training data, or receive information from the wearable device 140 worn on the subject. The measured second unipolar lead electrocardiogram and the second chest respiration signal, wherein the first unipolar lead electrocardiogram and the first chest respiration signal are measured at the same time period, and the second unipolar lead electrocardiogram and the second The chest respiration signal is measured at the same time period.

The wearable device 140 can be worn on the body of the subject, and is communicatively connected to the transceiver 130 through a communication technology such as Bluetooth (but not limited thereto). The wearable device 140 is used to measure the second chest breathing signal and the second unipolar lead electrocardiogram of the subject. Specifically, the wearable device 140 may include a monopolar lead electrode 141 and an accelerometer 142 . The unipolar lead electrode 141 can be attached to the subject's chest to measure the second unipolar lead electrocardiogram. The accelerometer 142 can be disposed on the subject's chest to measure the subject's chest rise and fall, thereby generating a corresponding second chest breathing signal. The second unipolar lead ECG and the second chest respiration signal are measured at the same time period. Therefore, when the second unipolar lead ECG generates a pulse wave corresponding to the apnea event at a specific time, the second chest respiration signal should also generate a pulse wave corresponding to the apnea event at the specific time.

The first chest breathing signal as training data may include a first data set corresponding to apnea events and a second data set corresponding to non-apneic events. In this embodiment, the first data set may include a plurality of one-minute chest breathing signals, and at least one apnea event occurs during each chest breathing signal. The second set of data may include a plurality of chest respiration signals of one minute in length, and no apnea event occurs during the measurement of each chest respiration signal. On the other hand, the first unipolar lead electrocardiogram as training data may include a third data set corresponding to apnea events and a fourth data set corresponding to non-apneic events. In this embodiment, the third data set may include a plurality of one-minute electrocardiograms, and at least one apnea event occurs during each electrocardiogram measurement. The fourth data set may include a plurality of one-minute-long electrocardiograms with no apneic events occurring during each electrocardiogram measurement.

The training module 121 can use the first data set, the second data set, the third data set and the fourth data set as training data to train the machine learning model, wherein the trained machine learning model can be used to The second chest respiration signal and the second unipolar lead electrocardiogram determine whether the subject has an apnea event. It should be noted that the training data in this embodiment may be time-domain signals that have not been wavelet-transformed, rather than time-frequency signals that have been wavelet-transformed. Since the dimension of the time-domain signal is lower than that of the time-frequency signal, the training module 121 that uses the time-domain signal instead of the time-frequency signal to train the machine learning model will spend less computing power.

In this embodiment, the training data may include multiple data pairs corresponding to different time periods, and each data pair includes the first chest respiration signal and the first unipolar lead electrocardiogram corresponding to the same time period. In the training data, the first chest respiration signal and the first unipolar lead ECG corresponding to the same data pair are arranged in adjacent queues. FIG. 2A is a schematic diagram illustrating training data of a convolutional neural network model 200 (shown in FIG. 2B ) according to an embodiment of the present disclosure. As shown in FIG. 2A , the training data may include a first unipolar lead electrocardiogram 21 , a first thoracic respiration signal 22 , a first unipolar lead electrocardiogram 23 and a first thoracic respiration signal 24 . The first unipolar lead electrocardiogram 21 corresponding to the first specific period is arranged in the first queue, and the first chest respiration signal 22 corresponding to the first specific period is arranged in the second queue adjacent to the first queue. A queue in which the first unipolar lead electrocardiogram 21 and the first chest respiration signal 22 correspond to the same data pair. Next, the first unipolar lead electrocardiogram 23 corresponding to the second specific period (ie: the period immediately following the first specific period) is arranged in the third queue, and the first chest respiration corresponding to the second specific period The signals 24 are arranged in a fourth queue adjacent to the third queue, wherein the first unipolar lead electrocardiogram 23 and the first chest respiration signal 24 correspond to the same data pair. The training module 121 can train the machine learning model according to the training data shown in FIG. 2A .

In one embodiment, the first thoracic respiration signal as training data further includes a fifth data set representing the thoracic respiration signal disturbed by noise, and the first unipolar lead electrocardiogram as training data further includes a fifth data set representing the thoracic respiration signal disturbed by noise The sixth data collection of the disturbed electrocardiogram. The training module 121 can use the first data set, the second data set, the third data set, the fourth data set, the fifth data set and the sixth data set as training data to train the machine learning model. The machine learning model trained by the first, second, third, fourth, fifth and sixth data sets not only The signal and the second unipolar lead electrocardiogram can determine whether the subject has an apnea event, and can also judge the second chest breathing signal and the first chest respiration signal received by the transceiver 130 according to the second chest breathing signal and the second unipolar lead electrocardiogram. Two unipolar lead ECG data may be disturbed by noise. Therefore, when judging the severity of the OSA symptoms of the subject, the detection module 122 can first filter out the data interfered with by the noise.

After training the machine learning model, the detection module 122 can determine whether the subject has an apnea event according to the machine learning model and the second chest breathing signal and the second unipolar lead electrocardiogram measured from the subject. In one embodiment, the detection module 122 may further determine the severity of the OSA symptoms of the subject according to the number of apnea events. Specifically, the detection module 122 can define the number of occurrences of apnea events of the subject per hour as an apnea index (Apnea Index, AI). The detection module 122 may determine that the severity of the OSA symptoms of the subject is high in response to the AI being greater than a first threshold (eg, the first threshold is 30). The detection module 122 may determine that the severity of the OSA symptoms of the subject is moderate in response to the AI being less than or equal to the first threshold but greater than a second threshold (eg, the second threshold is 15). The detection module 122 may determine that the severity of the symptoms of OSA in the subject is low (or determine that the subject does not have symptoms of OSA) in response to the AI being less than or equal to the second threshold.

In one embodiment, if the detection module 122 determines that the subject has severe OSA symptoms, the detection module 122 may issue a warning through the transceiver 130 to remind the subject's family, doctors or people around that the subject should be affected. condition of the examinee. For example, the detection module 122 can transmit an alert message to the mobile devices of the subject's family through the transceiver 130 when the subject has severe OSA symptoms. Compared with the traditional method of using an electroencephalogram to measure the subject's electroencephalography (EEG) to diagnose the OSA symptoms of the subject, the present disclosure only needs to use a single unit of the wearable device 140 with a simple structure and a low price. The pole lead electrode 141 and the accelerometer 142 can then monitor the severity of the subject's OSA symptoms in real time.

The above-mentioned machine learning model is, for example, a convolutional neural network (CNN) model. The biggest difference between the convolutional neural network and the traditional multi-layer perceptual network is that the convolutional neural network has more convolutional layers and pooling layers, which give the convolutional neural network the ability to extract the features of the input signal. The design of convolutional layers has several characteristics. The first feature is local awareness. In traditional neural networks, each neuron is connected to each sampling point, so a large number of weights are required, which makes training the network extremely difficult. In the convolutional neural network, the number of weights of each neuron is the same as the size of the convolution kernel, so it is equivalent to that each neuron is only connected to the corresponding part of the sampling points, which can greatly reduce the number of weights . A smaller number of weights reduces the risk of overfitting. The second feature is the weight sharing mechanism. The convolutional neural network trains and updates the optimal convolution kernel weight through the back-propagation error algorithm, but during the convolution process, the weight of the convolution kernel does not change. The third feature is multiple convolution kernels. If only one convolution kernel is used, only some features of the signal can be extracted. If multiple convolution kernels are used, multiple features of the input signal can be extracted. The greater the number of convolutional layers, the more features a convolutional neural network can extract.

After the input signal is subjected to nonlinear transformation by convolutional layers and activation functions, a feature map can be generated. The most important function of the activation function is to introduce the nonlinearity of the neural network, because if the activation function is not added, the convolutional layer and the fully connected layer are only simple linear operations, and there is still no solution to the linear inseparable problem.

In order to reduce the dimension of the features extracted by the convolution operation and increase the speed of the learning process, the convolutional layer is followed by a pooling layer. Pooling layers are a way to compress feature maps and preserve important information. The sampling method adopted by the pooling layer may include max pooling or mean pooling. The maximum pooling method selects the maximum value in the pooling window as the sampling value. The average pooling method is to add all the values in the pooling window and take the average as the sampling value. The feature map after pooling still retains the maximum likelihood of local range alignment. In other words, the pooled information can focus more on whether there are matching features in the feature map, rather than focusing on the location of these features. Therefore, compared with the traditional neural network, the convolutional neural network can better determine whether a feature is included in the feature map without considering the location of the feature. Therefore, even if the feature of the input signal is shifted, the convolutional neural network can recognize the feature. After the pooling layer, the fully connected layer will integrate the highly abstracted features after multiple convolutions and pooling. Then the output layer outputs a corresponding probability for each classification, where the sum of the probability of all classifications is 1.

FIG. 2B is a schematic diagram of a convolutional neural network model 200 according to an embodiment of the present disclosure, wherein the convolutional neural network model 200 is an aspect of the machine learning model trained by the training module 121 . The convolutional neural network model 200 may include an input layer 220 , a convolutional layer 231 , a pooling layer 232 , a convolutional layer 241 , a pooling layer 242 , a fully connected layer 251 , a fully connected layer 252 , and an output layer 260 . The input data 210 of the convolutional neural network model 200 shown in FIG. 2B is, for example, the second chest respiration signal and the second unipolar lead electrocardiogram measured from the subject, and the output data of the convolutional neural network model 200 270 represents the judgment result of whether an apnea event occurs.

的卷積核。經過卷積層231的輸入資料210會轉變為128個尺寸為

的特徵圖。接著，池化層232使用尺寸為

的滑動視窗對卷積層231輸出的特徵圖進行取樣以產生128個尺寸為

的特徵圖。卷積層241包括64個尺寸16

的卷積核。卷積層241進一步地對池化層232輸出的特徵圖進行卷積運算以產生64個尺寸為

的特徵圖。接著，池化層242使用尺寸為

的滑動視窗對卷積層241輸出的特徵圖進行取樣以產生64個尺寸為

的特徵圖。而後，池化層242輸出的特徵圖依序地輸入至具有128個神經元的全連接層251以及具有64個神經元的全連接層252。輸出層260可根據Softmax激活函數來計算全連接層252的輸出的對應於暫停呼吸事件的第一機率以及對應於非暫停呼吸事件的第二機率。若第一機率大於第二機率，則偵測模組122可判斷輸入資料210對應於至少一暫停呼吸事件。反之，若第一機率小於或等於第二機率，則偵測模組122可判斷輸入資料210對應於非暫停呼吸事件。In this embodiment, the input data 210 includes the second chest respiration signal for 1 minute and the second unipolar lead electrocardiogram for 1 minute, and the input data 210 includes 12,000 sampling points sampled at a sampling frequency of 100 Hz, wherein the first The sampling points for the second chest respiration signal and the second unipolar lead ECG were 6,000 points, respectively. The convolutional layer 231 includes 128 layers of size

the convolution kernel. The input data 210 passing through the convolutional layer 231 is transformed into 128 pieces of size

feature map. Next, the pooling layer 232 uses a size of

The sliding windows of the convolutional layer 231 output feature maps to produce 128 features of size

feature map. The convolutional layer 241 includes 64 dimensions 16

the convolution kernel. The convolutional layer 241 further performs a convolution operation on the feature map output by the pooling layer 232 to generate 64 dimensions of

feature map. Next, the pooling layer 242 uses a size of

The sliding window samples the feature map output by the convolutional layer 241 to produce 64 dimensions of

feature map. Then, the feature maps output by the pooling layer 242 are sequentially input to the fully connected layer 251 with 128 neurons and the fully connected layer 252 with 64 neurons. The output layer 260 may calculate the output of the fully connected layer 252 according to the Softmax activation function, a first probability corresponding to a breathing pause event and a second probability corresponding to a non-pausing breathing event. If the first probability is greater than the second probability, the detection module 122 may determine that the input data 210 corresponds to at least one apnea event. Conversely, if the first probability is less than or equal to the second probability, the detection module 122 may determine that the input data 210 corresponds to a non-apneic breathing event.

It should be noted that the activation function used by the output layer 260 may be, for example, a softmax function, a sigmoid function, a hyperbolic tangent function or a rectified linear unit (ReLU) function, but the present disclosure is not limited thereto.

In one embodiment, the training module 121 can further generate a support vector machine (SVM) model. The detection module 122 can determine whether the subject has an apnea event according to the support vector machine model, the machine learning model, and the second chest breathing signal and the second unipolar lead electrocardiogram measured from the subject. For example, if at least one of the support vector machine model and the machine learning model determines that the subject has an apnea event, the detection module 122 may determine in response to at least one of the support vector machine model and the machine learning model. The subject has an apnea event and a judgment result representing the occurrence of the subject's apnea event is output.

The training module 121 can train the aforementioned support vector machine model according to the first chest breathing signal and the first monopolar lead electrocardiogram. The training module 121 can measure a plurality of RR intervals according to the first unipolar lead electrocardiogram. Then, the training module 121 can generate second training data for training the support vector machine model according to the first chest breathing signal and at least one of the plurality of RR intervals, and then train the aforementioned training data according to the second training data Support vector machine model. Specifically, the training module 121 can extract a plurality of features from the first chest breathing signal or a plurality of RR intervals, and generate the second training data according to the features. The plurality of features are, for example, related to the average value of the RR interval, the second or third sequence correlation coefficient of the RR interval, the number of RR interval pairs (the RR interval pair includes two adjacent RR intervals, and the interval between the two RR intervals is time interval of more than 50 ms), the standard deviation of two adjacent RR intervals, the normalized very low frequency power (VLFP) of the RR interval, the normalized very low frequency power of the chest respiration signal, The normalized low frequency power (LFP) of the chest breathing signal or the normalized high frequency power (HFP) of the chest breathing signal, but the present disclosure is not limited thereto. The above-mentioned LFP is approximately between 0.04-0.15 Hz and the HFP is approximately between 0.15-0.4 Hz.

FIG. 3 is a flowchart illustrating a method for detecting apnea based on a chest breathing signal according to an embodiment of the present disclosure, wherein the detecting method is implemented by, for example, the detecting device 100 shown in FIG. 1 . In step S301, a first chest breathing signal and a first unipolar lead electrocardiogram corresponding to the first chest breathing signal are obtained. In step S302, the first chest respiration signal and the first unipolar lead electrocardiogram are used as training data to train the machine learning model. In step S303, a second chest respiration signal and a second unipolar lead electrocardiogram of the subject are obtained. In step S304, it is determined whether the subject has at least one apnea event according to the machine learning model, the second chest breathing signal and the second unipolar lead electrocardiogram.

In conclusion, the present disclosure can determine whether a subject has an apnea event and the severity of OSA symptoms of the subject based on the subject's chest breathing signal and unipolar lead signal. The use of chest breathing signals can significantly improve the accuracy of detection of apnea events. The chest respiration signal and unipolar lead ECG do not need to undergo wavelet transformation, and can also be used as the training data of the machine learning model in the form of time domain signals, thereby reducing the computing power required for the training and use of the machine learning model. The wearable device can simply use the accelerometer to measure the chest breathing signal to accurately assess the OSA symptoms of the subject using the chest breathing signal, without the need to configure the more expensive EEG to measure the EEG as a machine learning model training data. Therefore, the wearable device of the present disclosure has the advantages of simple structure and low price.

100: Detection device 110: Processor 120: Storage Media 121: Training Module 122: Detection module 130: Transceiver 140: Wearables 141: Unipolar lead electrode 142: Accelerometer 21, 23: First unipolar lead ECG 22, 24: First chest breathing signal 210: Enter data 220: Input layer 231, 241: Convolutional layer 232, 242: Pooling layer 251, 252: fully connected layer 260: output layer 270: output data S301, S302, S303, S304: steps

FIG. 1 is a schematic diagram illustrating a device for detecting apnea based on a chest breathing signal according to an embodiment of the present disclosure. 2A is a schematic diagram illustrating training data of a convolutional neural network model according to an embodiment of the present disclosure. FIG. 2B is a schematic diagram of a convolutional neural network model according to an embodiment of the present disclosure. 3 is a flowchart illustrating a method for detecting apnea based on a chest breathing signal according to an embodiment of the present disclosure.

100: Detection device

110: Processor

120: Storage Media

121: Training Module

122: Detection module

130: Transceiver

140: Wearables

141: Unipolar lead electrode

142: Accelerometer

Claims (9)

A device for detecting apnea based on a chest breathing signal, comprising: a transceiver for obtaining a first chest breathing signal and a first unipolar lead electrocardiogram corresponding to the first chest breathing signal, wherein the first unipolar lead a lead electrocardiogram and the first chest respiration signal are measured at the same first time period; a storage medium storing a plurality of modules; and a processor coupled to the storage medium and the transceiver and accessing and Execute the plurality of modules, wherein the plurality of modules include: a training module that uses the first chest respiration signal and the first unipolar lead electrocardiogram as training data to train a machine learning model, wherein the The training data includes time-domain signals without wavelet conversion; and a detection module, which obtains the subject's second chest respiration signal and second unipolar lead electrocardiogram through the transceiver, wherein the second single The polar lead ECG and the second chest respiration signal are measured in the same second time period, and it is determined according to the machine learning model, the second chest respiration signal and the second unipolar lead Whether the subject has at least one apnea event, wherein the first chest breathing signal includes a first data set corresponding to an apnea event and a second data set corresponding to a non-apneic event, and the first monopolar The lead electrocardiogram includes a third data set corresponding to the apnea event and a fourth data set corresponding to the non-apneic event. The detection device according to claim 1, wherein the detection module determines the severity of the symptoms of obstructive sleep apnea of the subject according to the number of occurrences of the at least one apnea event. The detection device of claim 2, wherein the detection module determines that the severity is high in response to the number of occurrences of the at least one apnea event being greater than a first threshold, and in response to The severity is determined to be medium in response to the occurrence of the at least one apnea event being less than or equal to the first threshold but greater than the second threshold, and in response to the occurrence of the at least one apnea event being less than or equal to the second threshold to judge the severity to be low. The detection device of claim 1, wherein the training measures a plurality of RR intervals according to the first unipolar lead electrocardiogram, and according to the plurality of RR intervals and the first chest respiration signal At least one of them generates second training data, and trains a support vector machine model according to the second training data. The detection device of claim 4, wherein the detection module is based on the support vector machine model, the machine learning model, the second chest breathing signal and the second monopolar conductor The electrocardiogram is used to determine whether the subject has the at least one apnea event. The detection device of claim 4, wherein the training module extracts a plurality of features from the plurality of RR intervals or the first chest breathing signal, and generates the plurality of features according to the plurality of features Second training material. The detection device of claim 6, wherein the plurality of features are associated with at least one of the following: RR interval average value, second or third sequence correlation coefficient of RR interval, number of RR interval pairs, wherein the RR interval pair includes two adjacent RR intervals, and the time interval between the two RR intervals Over 50 ms, the standard deviation of two adjacent RR intervals, the normalized VLF power of the RR interval, the normalized VLF power of the chest respiration signal, the normalized low frequency range power of the chest respiration signal, and the chest Normalized high frequency range power of the respiration signal. The detection device according to claim 1, further comprising: a wearable device, which is worn on the subject and is communicatively connected to the transceiver, wherein the wearable device comprises: a single a polar lead electrode attached to the subject's chest to measure the second monopolar lead electrocardiogram; and an accelerometer disposed on the subject's chest to measure the second chest respiration signal . A method for detecting apnea based on a chest breathing signal, comprising: obtaining a first chest breathing signal and a first unipolar lead electrocardiogram corresponding to the first chest breathing signal, wherein the first unipolar lead electrocardiogram and the first chest respiration signal is measured in the same first time period; the first chest respiration signal and the first monopolar lead electrocardiogram are used as training data to train a machine learning model, wherein the training data Including the time domain signal without wavelet conversion; obtaining the subject's second chest respiration signal and the second unipolar lead electrocardiogram, wherein the second unipolar lead electrocardiogram and the second chest respiration signal are in the Mutually The same second time period is measured; and according to the machine learning model, the second chest respiration signal and the second unipolar lead electrocardiogram to determine whether the subject has at least one apnea event, wherein the The first chest breathing signal includes a first set of data corresponding to an apnea event and a second set of data corresponding to a non-apneic event, and the first unipolar lead ECG includes a third set of data corresponding to the apnea event A data set and a fourth data set corresponding to the non-apneic event.

Images