Sensor, monitoring method and system for sleep apnea monitoring

Abstract

The invention relates to the technical field of human body monitoring, and discloses a sensor for sleep apnea monitoring, a monitoring method and a system.The sensor comprises a packaging structure prepared from a polymer material, a first friction layer, a second friction layer, a third friction layer, a fourth friction layer and a space structure layer, and the first friction layer, the second friction layer, the third friction layer, the fourth friction layer and the space structure layer are arranged in the packaging structure; the first friction layer is woven on the upper surface of the space structure layer, the second friction layer is woven on the lower surface of the space structure layer, the third friction layer and the first friction layer are separated by the space structure layer surface layer, and the second friction layer and the fourth friction layer are separated by the space structure layer surface layer. When the space structure layer deforms due to external pressure, the first friction layer, the second friction layer, the third friction layer and the fourth friction layer convert external pressure vibration into electric signals to be output.

Description

Technical Field

[0001] The present invention relates to the technical field of human body monitoring, and in particular to a sensor, a monitoring method and a system for monitoring sleep apnea. Background Art

[0002] In recent years, with the continuous progress of social science and technology, people are paying more and more attention to their sleep quality. According to calculations, about one-third of a person's life is spent in sleep. Sleep is a necessary process for life. The body's metabolism and all other physiological activities will be reduced, and the body is in a state of protective inhibition to avoid functional failure due to excessive consumption of nerve cells. At the same time, sleep can restore the normal physiological function of fatigued nerves. Sleep is also necessary to maintain normal human immune function. However, surveys show that a considerable number of people suffer from sleep disorders or sleep-related diseases. About 30% of adults have sleep disorders, and 30% of them have serious conditions, which even affect the functions of organs such as cardiovascular muscles and endanger health. Sleep disorders are an important cause of the formation and development of many diseases. Not only may they cause a decline in various physiological functions of the human body, but they will also lead to a series of systemic lesions. Diseases can cause severe hypoxemia and sleep disorders. Sleep disorders are closely related to the occurrence of diseases such as hypertension, arrhythmia, cardiovascular and cerebrovascular diseases and respiratory failure. A few patients may even die suddenly at night. Therefore, sleep disorders are increasingly receiving widespread attention from the international medical community. In daily life, sleep monitoring of patients with sleep breathing disorders has gradually become an essential basic examination in self-health monitoring. By analyzing physiological signals during sleep, the presence of certain potential diseases can be detected, which has important clinical significance.

[0003] At present, a variety of sleep detection methods have been put into clinical use. Among them, polysomnography can monitor more than 10 human physiological signals, including electromyography, electrocardiography, eye movement, abdominal and thoracic respiratory tension chart, nasal and oral ventilation, body position and movement, blood oxygen saturation, etc., in addition to electroencephalogram, and is considered to be the gold standard for diagnosing sleep apnea syndrome. Although the use of polysomnography can obtain very complete data for the diagnosis of sleep disorders, the patient needs to stay in the hospital for observation during the test, and professional medical staff are required to monitor the patient at any time. Since it cannot meet the needs of patients to monitor at home at any time, it is only used in medical institutions to diagnose a small number of patients with more serious conditions. Not only that, patients are also subject to the limitations and constraints of wired connections and binding of many instruments, which are complicated to operate and cause inconvenience to patients' sleep. Therefore, it is of great value to develop a low-cost sleep apnea monitoring system that can diagnose sleep disorders early, allow patients to use it anytime and anywhere without restraint, and can monitor in real time and continuously.

[0004] The waveform amplitude, shape and period of the human pulse wave contain important physiological information reflecting the condition of the heart and blood vessels. Being familiar with and mastering the relationship between pulse waves and sleep syndrome monitoring, extracting physiological and pathological information of the human body from pulse waves, and using it as a basis for clinical diagnosis and treatment, has always been valued by people. In recent years, many domestic and foreign scholars have devoted themselves to the study of human pulse signals. Photoelectric volumetric pulse wave is a commonly used method for detecting pulse signals, but because it is a finger-clip or wristband device worn on the body of the person being tested, it will bring inconvenience to the tester and affect the comfort during sleep. In addition, if there is sweat on the fingers, it will affect the accuracy of the measurement results of the photoelectric sensor. Summary of the invention

[0005] In view of the above-mentioned prior art, the present invention provides a sensor, a monitoring method and a system for sleep apnea monitoring, which mainly solve the technical problems existing in the above-mentioned background technology.

[0006] To achieve the above object, the technical solution of the embodiment of the present invention is implemented as follows: The first aspect of the present invention discloses a sensor for sleep apnea monitoring, the sensor comprising a packaging structure made of a polymer material and a first friction layer, a second friction layer, a third friction layer, a fourth friction layer, and a spatial structure layer arranged in the packaging structure, the first friction layer is woven on the upper surface of the spatial structure layer, the second friction layer is woven on the lower surface of the spatial structure layer, the third friction layer and the first friction layer, and the second friction layer and the fourth friction layer are separated by the surface layer of the spatial structure layer, when the spatial structure layer is deformed due to external pressure, the first friction layer, the second friction layer, the third friction layer, and the fourth friction layer convert the external pressure vibration into an electrical signal for output.

[0007] Optionally, the first friction layer, the second friction layer, and the third friction layer are made of dielectric materials with different electrical polarities.

[0008] Optionally, the spatial structure layer is composed of a plurality of three-dimensional airbags, the first friction layer is woven on the upper surface of the three-dimensional airbag, and the second friction layer is woven on the lower surface of the three-dimensional airbag.

[0009] Optionally, the first friction layer and the fourth friction layer are both formed by twisting and sewing conductive nanofibers embedded with Mxene and elastic polymer nanofibers.

[0010] Optionally, the second friction layer is made of a fiber material that easily loses electrons, or a fiber material plated with electrodes.

[0011] Optionally, the material of the third friction layer includes composite elastic conductive nanofiber fabric and polymer fiber fabric.

[0012] Optionally, when the material of the third friction layer is a composite elastic conductive nanofiber fabric, a wire is led out from the first friction layer and the second friction layer and coupled to serve as a first signal output end, and a wire is led out from the third friction layer and the fourth friction layer and coupled to serve as a second signal output end, and the first signal output end and the second signal output end simultaneously output electrical signals to the outside.

[0013] Optionally, when the material of the third friction layer is polymer fiber fabric, only the wires are led out from the first friction layer and the second friction layer and coupled to serve as the first signal output end, and the electrical signal is outputted externally from the first signal output end.

[0014] A second aspect of the present invention discloses a monitoring system for sleep apnea, the monitoring system comprising a sensor for sleep apnea monitoring as described in any of the foregoing items, the system further comprising a low-pass filtering module, a signal amplifying module, an analog / digital conversion module, a microcontroller module, and a wireless data transmission module, the signal output end of the sensor being connected to the low-pass filtering module, the low-pass filtering module being connected to the signal amplifying module, the analog / digital conversion module, and the microcontroller module in sequence, and the microcontroller module being connected to a mobile terminal signal via the wireless data transmission module.

[0015] The third aspect of the present invention discloses a method for monitoring sleep apnea, which is applied to the monitoring system as described above, and comprises collecting pulse signals through the sensor, and extracting features from the pulse signals by a deep learning network for evaluating sleep breathing conditions, wherein the specific steps of extracting features from the pulse signals by the deep learning network for evaluating sleep breathing conditions include: S1. Pass the pulse signal to the first Conv1D convolution layer in the deep learning network to extract local time features. The expression of the Conv1D convolution layer is: Where k is the size of the convolution kernel, is the weight of the convolution kernel, b is the bias, t is the current time step, i is the offset of the convolution kernel, Represents the input pulse signal; S2. Normalize the output of the aforementioned Conv1D convolutional layer. The normalization formula is: in is the mean of the data, is the variance of the data, To prevent the denominator from being zero; S3, perform maximum pooling on the normalized results, and randomly discard neurons during the maximum pooling process to prevent data overfitting; S4, repeat steps S1 to S3, but use different kernel sizes, output channels and pooling windows to further extract and compress features, and finally obtain the extracted feature data ,in is the number of time steps, is the characteristic dimension; S5, using a double-layer BiLSTM to further extract time series features, which outputs new feature data , is the new feature dimension; S6. Combine the original pulse signal and the characteristic data Combine and obtain the combined result , where n is the number of classes; S7. Import to Softmax classifier and convert the output of fully connected layer into class probability; S8. Output the maximum probability corresponding condition as the result label to the mobile terminal, and transmit the class probability result to the expert database for comparison and correction to complete the recognition.

[0016] The beneficial effects of the present invention are as follows: the sensor provided by the present application can capture external pressure vibrations of different degrees in layers while satisfying the structure that the contact area can be automatically adjusted according to the external pressure, and can automatically adjust the contact area between the two layers according to the different pressures. The third friction layer and the first friction layer, and the second friction layer and the fourth friction layer are separated by the surface layer of the space structure layer, and a certain gap is provided while ensuring that each friction sensor works independently, and finally a nested, multi-electrode structure is formed. In the nested double-electrode structure, the multi-layer structure design with adjustable contact area ensures that the fabric sensor responds to vibrations of a wide range of external working pressures through elastic structures of different strengths and different spaces, and has high energy conversion efficiency while ensuring high sensitivity and a wide working frequency range. It is suitable for detecting weak human physiological signals of different frequencies, such as pulse signals and breathing signals, and well solves the problem of weak or interrupted acquisition signals that may be caused by the activities of the user during sleep. In addition, without the need for an external power supply to power the sensor, the vibration caused by the pulse of different parts of the human body and people with different degrees of sleep apnea syndrome can be converted into electrical energy. The mobile terminal uploads the pulse data to the cloud. Through the deep learning network algorithm, it can automatically and accurately detect multiple feature points of the pulse signal, calculate the relevant parameter values, and interact with the expert database in real time to perform self-correction and evolution with high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the structure of the flexible fabric sensor implemented in the present invention.

[0018] Figure 2 It is a schematic diagram of the planed structure of the flexible fabric sensor implemented in the present invention.

[0019] Figure 3 It is a schematic diagram of the preparation of conductive nanofibers embedded with Mxene according to an embodiment of the present invention.

[0020] Figure 4 This is a process for manufacturing a conductive composite friction layer formed by twisting conductive nanofibers embedded in Mxene and elastic polymer nanofibers in an embodiment of the present invention.

[0021] Figure 5 This is a schematic diagram of the flexible fabric sensor of the present invention being sewn on different positions of wearable clothing and worn by a user.

[0022] Figure 6 It is a two-dimensional schematic diagram of the principle of the sensor signal transmission process in the dual-electrode mode of this embodiment.

[0023] Figure 7 It is a two-dimensional schematic diagram of the principle of the sensor signal transmission process in the single electrode mode of this embodiment.

[0024] Figure 8 This is a structural flow chart of the system for detecting and diagnosing sleep apnea syndrome in this embodiment. DETAILED DESCRIPTION

[0025] The technical solution of the present invention is further elaborated in detail below in conjunction with the drawings and specific embodiments of the specification. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. In the following description, the expression "some embodiments" is related to a subset of all possible embodiments, but it should be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0026] In the following description, a large number of specific details are provided to provide a more thorough understanding of the present invention. However, it is apparent to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some technical features well known in the art are not described.

[0027] It should be understood that the present invention can be implemented in different forms and should not be interpreted as being limited to the embodiments proposed herein. On the contrary, providing these embodiments will make the disclosure thorough and complete, and the scope of the present invention will be fully conveyed to those skilled in the art. And the purpose of the terms used herein is only to describe specific embodiments and is not intended to be a limitation of the present invention. When used herein, the singular forms of "one", "one" and "said / the" are also intended to include plural forms, unless the context clearly indicates another way. It should also be understood that the terms "compose" and / or "include" when used in this specification determine the presence of the features, integers, steps, operations, elements and / or parts, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, parts and / or groups. When used herein, the term "and / or" includes any and all combinations of the relevant listed items.

[0028] It should also be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "inside", "outside", "left", "right" and similar expressions used herein are for illustrative purposes only and are not intended to be the only implementation method.

[0029] In order to fully understand the present invention, a detailed structure will be proposed in the following description to illustrate the technical solution proposed by the present invention. The optional embodiments of the present invention are described in detail as follows, but in addition to these detailed descriptions, the present invention may also have other implementations.

[0030] Please refer to the attached Figures 1 to 7 In a first aspect, the present invention discloses a sensor for monitoring sleep apnea, the sensor comprising a packaging structure made of a polymer material and a first friction layer, a second friction layer, a third friction layer, a fourth friction layer, and a space structure layer arranged in the packaging structure, the first friction layer is woven on the upper surface of the space structure layer, the second friction layer is woven on the lower surface of the space structure layer, the third friction layer and the first friction layer, and the second friction layer and the fourth friction layer are separated by the surface layer of the space structure layer, when the space structure layer is deformed due to external pressure, the first friction layer, the second friction layer, the third friction layer, and the fourth friction layer convert the external pressure vibration into an electrical signal for output.

[0031] The packaging structure is usually made of a waterproof, dustproof and breathable expanded polytetrafluoroethylene fabric. The first friction layer and the second friction layer are respectively woven on the upper surface and the lower surface of the space structure layer by 3D weaving technology, so as to ensure that a gap is generated between the first friction layer and the second friction layer. When the sensor is driven by external pressure, the contact area between the two layers can be automatically adjusted according to the pressure. The third friction layer and the first friction layer, and the second friction layer and the fourth friction layer are separated by the surface layer of the space structure layer, so as to provide a certain gap while ensuring that each friction sensor works independently, and finally form a nested, multi-electrode structure. In the nested double-electrode structure, through the elastic structure of different strengths and different spaces, the multi-layer structure design with adjustable contact area ensures that the fabric sensor responds to vibrations in a wide range of external working pressures, has high energy conversion efficiency, and ensures high sensitivity and a wide working frequency range. It is suitable for detecting weak human physiological signals of different frequencies, such as pulse signals and breathing signals, and well solves the problem of weak or interrupted acquisition signals that may be caused by the user's activities during sleep.

[0032] The working principle of the sensor disclosed in the present application is as follows: when the external pressure vibration is transmitted to the third friction layer, the third friction layer with an arbitrary sewing pattern is driven to immediately deform and contact the first friction layer, and then the first friction layer is pushed to squeeze the spatial structure layer until it contacts the second friction layer. Due to the different affinity of adjacent friction material layers for electrons, the positive and negative charges generated by friction are transferred to each other. Due to electrostatic induction, the charges generated in the contact area will reach equilibrium, so there is no potential difference between adjacent friction material layers; once the external pressure is released, the third friction layer and the first friction layer will immediately bounce open due to the elasticity of the spatial structure layer, thereby generating a gap between the first friction layer and the second friction layer, resulting in a voltage drop between the two layers, and a potential difference will also be generated between the first friction layer and the second friction layer, and between the third friction layer and the first friction layer. The different potentials cause free electrons to flow in the external circuit until it can completely offset the positive charge, and finally realize the conversion of the external pressure vibration to an electrical signal by the double-stage linkage friction electric pressure sensor.

[0033] In addition, a variety of patterns can be sewn on the third material layer and the packaging structure made of polymer materials to meet the aesthetic design pattern requirements of wearable clothing or apparel for different ages and occasions. The number of sewing layers can be single or double according to the needs of different scenes, and the shape can be regular and symmetrical or irregular. It can be sewn with any wearable clothing, or it can be sewn as a logo on the surface of wearable accessories. It can not only accurately monitor the human pulse, but also add color to the clothing without affecting the wearing comfort.

[0034] In an optional embodiment, the first friction layer, the second friction layer, and the third friction layer are made of dielectric materials with different electrical polarities, thereby increasing the output power.

[0035] The first friction layer and the fourth friction layer are both formed by twisting and sewing conductive nanofibers embedded in Mxene and elastic polymer nanofibers. Specifically, the sewing yarn is a double-layer structure, the outer layer is composed of insulating fiber filaments, and the inner layer is composed of conductive nanofiber filaments embedded in Mxene. The conductive nanofiber filaments can be made of a variety of materials including silver nanowires, carbon nanotubes, graphene nanofibers, carbon fiber filaments, etc. In the present application, graphene nanofiber filaments are preferred. The insulating fiber filaments can be made of one of materials such as polyethylene terephthalate, polyamide fibers, polyurethane fibers, etc. In the present invention, polyethylene terephthalate is preferred.

[0036] Furthermore, the first friction layer can be sewn on the upper surface of the space structure layer by sewing methods such as flat stitching, shrinkage stitching, hidden stitching, etc., and the second friction layer is also sewn on the upper and lower surfaces of the space structure layer by the above sewing method. The second friction layer is made of a fiber material that easily loses electrons, or a fiber material plated with an electrode, and the electrode can be one of a silver electrode, a gold electrode, and a copper electrode.

[0037] The material of the third friction layer includes composite elastic conductive nanofiber fabric and polymer fiber fabric. When it is composite elastic conductive nanofiber fabric, it is formed by twisting multiple polymer fiber filaments and composite conductive fiber filaments. The number of polymer fibers is 1 to 6, preferably 3 to 5, and the number of composite conductive filaments is 1 to 6, preferably 1 to 2.

[0038] Furthermore, the sewn third friction layer can be multi-layer or single-layer, with each layer having a thickness of 500μm~3mm, preferably 1~2mm, and sewn into a single-layer pattern. If there are too many sewn layers, the weight of the sensor will increase, and the slight external pressure will hardly drive the third friction layer to displace and deform and contact the first friction layer.

[0039] The third friction layer can be a regular pattern or an irregular pattern with any logo. The first friction layer is selected to have an optimal single layer thickness of 1-2 mm, and when it is sewn with a mixed yarn made by twisting three elastic polymer fibers and one elastic conductive fiber, the displacement deformation of the sensor sensing the external pressure can be optimized, which can ensure that the device has a wide working pressure range and high sensitivity.

[0040] Since the first friction layer, the second friction layer, the third friction layer, and the fourth friction layer are tightly sewn together, if there is no surface layer of the spatial structure layer made of polymer elastic yarn, the third friction layer and the first friction layer, the second friction layer and the fourth friction layer will be in close contact without gaps, which is not conducive to the contact and separation of the third friction layer and the first friction layer, the second friction layer and the fourth friction layer, and at the same time, it is not very sensitive to the pressure response of different sizes and frequencies from the outside. If the sewing thickness of the surface layer is too thick, on the basis of the original spatial structure gap, when the third friction layer and the first friction layer or the second friction layer and the fourth friction layer are deformed, the contact area will be reduced or even unable to contact due to the long distance.

[0041] In an optional embodiment, the spatial structure layer is composed of a plurality of three-dimensional airbags, the first friction layer is woven on the upper surface of the three-dimensional airbags, and the second friction layer is woven on the lower surface of the three-dimensional airbags.

[0042] When the material of the third friction layer is a composite elastic conductive nanofiber fabric, a wire is led out from the composite nanoconductive fibers in the first friction layer and the conductive fibers in the second friction layer and coupled as a first signal output end, and a wire is led out from the composite nanoconductive fibers in the third friction layer and the conductive fibers in the fourth friction layer and coupled as a second signal output end, the first signal output end and the second signal output end form a dual-electrode structure and output electrical signals to the outside at the same time.

[0043] In an optional embodiment, when the material of the third friction layer is polymer fiber fabric, only the wires are led out from the first friction layer and the second friction layer and coupled to serve as the first signal output end, and the electrical signal is outputted externally from the first signal output end.

[0044] See also Figure 8 In a second aspect, the present invention discloses a monitoring system for sleep apnea, the monitoring system comprising a sensor for sleep apnea monitoring as described in any of the above items, the system further comprising a low-pass filtering module, a signal amplifying module, an analog / digital conversion module, a microcontroller module, and a wireless data transmission module, the signal output end of the sensor is connected to the low-pass filtering module, the low-pass filtering module is connected to the signal amplifying module, the analog / digital conversion module, and the microcontroller module in sequence, and the microcontroller module is connected to a mobile terminal signal through the wireless data transmission module.

[0045] The electrical signal obtained by the sensor is first transmitted to the low-pass filter module, which is used to filter out high-frequency noise and retain complete information below 45Hz to ensure that the pulse waveform is not distorted. The filtered signal then enters the signal amplification module, which amplifies the signal to improve the signal-to-noise ratio and make subsequent processing more accurate. The amplified analog signal is converted into a digital signal through the analog / digital conversion module (ADC) for digital processing.

[0046] The digital signal is then transmitted to the microcontroller module, which is responsible for coordinating the operation of the entire system, including data acquisition, processing and transmission. The microcontroller module sends the data to a mobile terminal, such as a mobile phone, computer or tablet, through a wireless data transmission module (such as Bluetooth, ZigBee, etc.). The mobile terminal uploads the data to the cloud through the 5G network and uses the powerful computing power of the cloud to process the data.

[0047] In the cloud, the improved CNN+BILSTM deep learning model algorithm is used to analyze, identify and store pulse or respiratory data. These models can effectively identify the characteristics of sleep apnea and provide accurate analysis results. The processed data and analysis results are transmitted back to the mobile terminal via the 5G network, and users can view the pulse waveform, related measurement parameters, analysis results and expert diagnosis suggestions in real time on the mobile terminal.

[0048] Furthermore, since the sensor in the present invention is based on a combination of flexible materials and fabrics, it can be sewn together with wearable clothing and arranged on the wrist, arm, chest and other parts of the body where the pulse is easy to measure. The wrist acquisition device is a wristband structure fixed to the radial artery of the wrist, which collects pulse wave signals in real time through a flexible fabric sensor. The above device is a prior art in the field and is not specifically described in this embodiment.

[0049] See also Figure 8 In a third aspect, the present invention discloses a method for monitoring sleep apnea, which is applied to the monitoring system as described above. The monitoring method includes collecting pulse signals through the sensor, and extracting features from the pulse signals by a deep learning network for evaluating sleep breathing conditions, wherein the specific steps of extracting features from the pulse signals by the deep learning network for evaluating sleep breathing conditions include: S1, set the input pulse wave signal , whose shape is (N, 1), where N represents the time step or the number of sampling points. The pulse signal is passed to the first Conv1D convolution layer in the deep learning network for local time feature extraction. The output shape is (N - k + 1, C), where C is the number of output channels (number of convolution kernels). The expression of the Conv1D convolution layer is: Where k is the size of the convolution kernel, is the weight of the convolution kernel, b is the bias, t is the current time step, i is the offset of the convolution kernel, Represents the input pulse signal; S2. Normalize the output of the Conv1D convolutional layer. At this time, the input shape is (N - k + 1, C). The normalization formula is: in is the mean of the data, is the variance of the data, To prevent the denominator from being zero; S3. Perform maximum pooling on the normalized result to reduce the dimension by taking the maximum value in each pooling window. After pooling, the data output shape is ((N - k + 1) / p, C), where p is the pooling window size. In the maximum pooling process, neurons are randomly discarded to prevent data overfitting and improve generalization ability. At this time, the output shape remains unchanged ((N -k + 1) / p, C); In addition, the Conv1D layer carries the residual connection to the previous layer input to extract local temporal features. The residual connection maintains the output shape of (N - k + 1) / p - + 1, ); S4. Repeat steps S1 to S3, with the same Conv1D and MaxPooling layers, but using different kernel sizes, output channels, and pooling windows to further extract and compress features. The output shape is ((N - k + 1) / p - +1), ), execute S4 again to further extract features, and we will get an extracted feature data ,in is the number of time steps, is the characteristic dimension; S5, using a double-layer BiLSTM to further extract time series features, which outputs new feature data , To create a new feature dimension, the two-layer BiLSTM combines forward and backward data flows to capture bidirectional semantic dependencies, where the LSTM unit formula is: S6. Combine the original pulse signal and the characteristic data Combine and obtain the combined result , where n is the number of classes; S7. Import it into the Softmax classifier and convert the output of the fully connected layer into class probability. At this time, the output is the probability value of each class. The Softmax formula is: S8. Output the maximum probability corresponding condition as the result label to the mobile terminal, and transmit the class probability result to the expert database for comparison and correction to complete the recognition.

[0050] Specifically, the sensor provided in the first aspect of the present application is used to collect pulse signals from any arterial part of the human body, such as the wrist, arm, etc., and the voltage and current signals output from the sensor are combined with the amplitude, frequency, area, rhythm and other features independently extracted by the improved CNN+BILSTM deep learning network algorithm. The classifier is arranged through the data sets marked by various evaluation standards in sleep breathing events to determine the sleep state data.

[0051] When judging sleep breathing events, the peaks and troughs are used for judgment. The specific steps are as follows: The decrease in the pulse wave peak value is generally greater than a (a<1) times the maximum peak value, and the increase in the valley value is greater than b (b<1) times the minimum valley value. Therefore, the maximum peak threshold can be determined as c (c<1) times the maximum peak value, and the valley threshold is d (d>1) times the minimum valley value, that is, the respiratory event occurs in the time period after the peak value is less than c times the maximum peak value and the valley value is greater than d times the minimum valley value. The possible segments of respiratory events that have been initially determined through threshold screening are further confirmed by waveform characteristics. The rising amplitude of the main wave amplitude H cannot be less than a certain threshold (such as e (e<1) times the maximum peak value), otherwise the current segment is excluded as the period of occurrence of respiratory events. Therefore, a sleep apnea syndrome model can be established based on the judgment of the above thresholds, peak values ​​and valley values.

[0052] Since apnea and hypopnea can cause micro-arousal, this has important protective significance for patients, because micro-arousal reactions can stimulate pharyngeal contraction, restore airway patency, improve the respiratory center or compensate for its defects, thereby restoring normal rhythmic breathing. Each respiratory event will cause a micro-arousal. With the occurrence of micro-arousal, the heart rate will rise rapidly, and breathing will gradually return to normal. Then the patient will quickly fall asleep, and the heart rate will drop rapidly again. Therefore, each respiratory event will inevitably produce a process in which the heart rate first rises rapidly and then drops rapidly. Multiple respiratory events will inevitably lead to periodic changes in heart rate. Therefore, attention should be paid to identifying the characteristic peaks and singularities of micro-arousals, as well as the waveform of the specific periodic changes in heart rate, which can be used to judge respiratory events and calibrate algorithms.

[0053] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. The protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A sensor for sleep apnea monitoring, characterized in that: The sensor includes a packaging structure made of polymer material and a first friction layer, a second friction layer, a third friction layer, a fourth friction layer, and a space structure layer arranged in the packaging structure, the first friction layer is woven on the upper surface of the space structure layer, the second friction layer is woven on the lower surface of the space structure layer, the third friction layer and the first friction layer, and the second friction layer and the fourth friction layer are separated by the surface layer of the space structure layer, and when the space structure layer is deformed due to external pressure, the first friction layer, the second friction layer, the third friction layer, and the fourth friction layer convert the external pressure vibration into an electrical signal for output.

2. A sensor for sleep apnea monitoring according to claim 1, characterized in that: The first friction layer, the second friction layer and the third friction layer are made of dielectric materials with different electrical polarities.

3. A sensor for sleep apnea monitoring according to claim 2, characterized in that: The spatial structure layer is composed of a plurality of three-dimensional airbags, the first friction layer is woven on the upper surface of the three-dimensional airbag, and the second friction layer is woven on the lower surface of the three-dimensional airbag.

4. A sensor for sleep apnea monitoring according to claim 3, characterized in that: The first friction layer and the fourth friction layer are both formed by twisting and sewing conductive nanofibers embedded in Mxene and elastic polymer nanofibers.

5. A sensor for sleep apnea monitoring according to claim 3, characterized in that: The second friction layer is made of a fiber material that easily loses electrons, or a fiber material plated with electrodes.

6. A sensor for sleep apnea monitoring according to claim 3, characterized in that: The material of the third friction layer includes composite elastic conductive nanofiber fabric and polymer fiber fabric.

7. A sensor for sleep apnea monitoring according to claim 6, characterized in that: When the material of the third friction layer is a composite elastic conductive nanofiber fabric, a wire is led out from the first friction layer and the second friction layer and coupled to serve as a first signal output end, and a wire is led out from the third friction layer and the fourth friction layer and coupled to serve as a second signal output end, and the first signal output end and the second signal output end simultaneously output electrical signals to the outside.

8. A sensor for sleep apnea monitoring according to claim 7, characterized in that: When the material of the third friction layer is polymer fiber fabric, only the wires are drawn out from the first friction layer and the second friction layer and coupled as the first signal output end, and the electrical signal is outputted externally from the first signal output end.

9. A monitoring system for sleep apnea, the monitoring system comprising the sensor for sleep apnea monitoring according to any one of claims 1 to 8, characterized in that: The system also includes a low-pass filtering module, a signal amplifying module, an analog / digital conversion module, a microcontroller module, and a wireless data transmission module. The signal output end of the sensor is connected to the low-pass filtering module, and the low-pass filtering module is connected to the signal amplifying module, the analog / digital conversion module, and the microcontroller module in sequence. The microcontroller module is connected to the mobile terminal signal through the wireless data transmission module.

10. A method for monitoring sleep apnea, characterized in that: The monitoring method is applied to the monitoring system according to claim 9, and the monitoring method includes collecting pulse signals through the sensor, and extracting features from the pulse signals by a deep learning network for evaluating sleep breathing conditions, wherein the specific steps of extracting features from the pulse signals by the deep learning network for evaluating sleep breathing conditions include: S1. Pass the pulse signal to the first Conv1D convolution layer in the deep learning network to extract local time features. The expression of the Conv1D convolution layer is: Where k is the size of the convolution kernel, is the weight of the convolution kernel, b is the bias, t is the current time step, i is the offset of the convolution kernel, Represents the input pulse signal; S2. Normalize the output of the aforementioned Conv1D convolutional layer. The normalization formula is: in is the mean of the data, is the variance of the data, To prevent the denominator from being zero; S3, perform maximum pooling on the normalized results, and randomly discard neurons during the maximum pooling process to prevent data overfitting; S4, repeat steps S1 to S3, but use different kernel sizes, output channels and pooling windows to further extract and compress features, and finally obtain the extracted feature data ,in is the number of time steps, is the characteristic dimension; S5, using a double-layer BiLSTM to further extract time series features, which outputs new feature data , is the new feature dimension; S6. Combine the original pulse signal and the characteristic data Combine and obtain the combined result , where n is the number of classes; S7, Import to Softmax classifier and convert the output of fully connected layer into class probability; S8. Output the maximum probability corresponding condition as the result label to the mobile terminal, and transmit the class probability result to the expert database for comparison and correction to complete the recognition.

Images