Apnea recognition method, apparatus and system

By incorporating deformable elements into a pillow to acquire pressure and snoring signals, and then using a deep learning model for comprehensive analysis, the problem of low accuracy in snoring monitoring is solved, enabling efficient identification and accurate judgment of sleep apnea events.

WO2026237984A1PCT designated stage Publication Date: 2026-11-19ZHANGZHOU SOLEX SMART HOME CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2026/096616
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-13
Filing Date
2026-05-13
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Existing methods for monitoring sleep apnea events through snoring are not very accurate, are easily affected by external interference, and can impact sleep quality.

Method used

Pressure and snoring signals are acquired using deformable elements in the pillow, and then identified and judged using a deep learning model. The combination of pressure and snoring signals is used to determine whether a sleep apnea event has occurred. The deep learning model is used to filter and normalize the signals to reduce noise interference.

Benefits of technology

It improves the accuracy of sleep apnea event detection, reduces the false negative rate, and does not affect the user's sleep quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2026096616_19112026_PF_FP_ABST
    Figure CN2026096616_19112026_PF_FP_ABST
Patent Text Reader

Abstract

An apnea recognition method, apparatus and system, relating to the technical field of respiration monitoring. The apnea recognition method comprises the following steps: acquiring a pressure signal of a deformation element in a pillow, and simultaneously acquiring a snore signal; and analyzing the pressure signal and the snore signal to determine whether an apnea event has occurred. Also disclosed in the present invention is an apnea recognition apparatus, comprising a pressure acquisition module, a snore acquisition module and a recognition module, wherein the pressure acquisition module is used for acquiring a pressure signal of a deformation element in a pillow, the snore acquisition module is used for acquiring a snore signal, and the recognition module is used for analyzing the pressure signal and the snore signal to determine whether an apnea event has occurred. Also disclosed in the present invention is an apnea recognition system, comprising a pillow, a sound collection member, a pressure collection member and the described apnea apparatus. The apnea recognition method, apparatus and system have high accuracy for monitoring apnea events.
Need to check novelty before this filing date? Find Prior Art

Description

Sleep apnea identification methods, devices and systems Technical Field

[0001] This invention relates to the field of respiratory monitoring technology, and in particular to a method, device and system for identifying sleep apnea. Background Technology

[0002] Among snorers, about 20% frequently experience breathing difficulties while sleeping, a condition medically termed sleep apnea syndrome. Sleep apnea syndrome (SAS) refers to a clinical syndrome caused by various factors leading to pauses in breathing and / or hypoventilation during sleep, resulting in recurrent episodes of intermittent hypoxemia and hypercapnia, thereby causing a series of pathophysiological changes in the body.

[0003] Because sleep apnea occurs during sleep and is not easily detected, if long-term sleep apnea goes undetected and untreated, it can lead to a series of diseases. Therefore, detecting sleep apnea is crucial for human health.

[0004] Existing methods for detecting sleep apnea are divided into direct and indirect monitoring methods. Direct monitoring obtains respiratory information by monitoring nasal breath flow, while indirect monitoring primarily obtains respiratory information through snoring, heart sounds, and lung sounds. Monitoring nasal breath flow or monitoring heart and lung sounds requires the monitor to wear corresponding devices, which can easily affect the sleep quality of the monitored individual and has therefore been gradually abandoned. Currently, snoring monitoring is mainly used, but snoring monitoring is easily affected by other external information, resulting in low accuracy. Summary of the Invention

[0005] The purpose of this invention is to provide a method, device, and system for identifying sleep apnea, in order to solve the problem of low accuracy in monitoring sleep apnea events through snoring.

[0006] To achieve this objective, the present invention adopts the following technical solution:

[0007] In a first aspect, a method for identifying sleep apnea includes the following steps:

[0008] The pressure signal of the deformable element (11) in the pillow (1) is acquired, and the snoring signal is acquired at the same time;

[0009] The pressure signal and the snoring signal are identified and judged to determine whether a sleep apnea event has occurred.

[0010] Optionally, when identifying and judging the pressure signal and the snoring signal, the pressure signal and the snoring signal are input into a deep learning model, and the deep learning model determines whether a sleep apnea event has occurred based on the confidence level of the sleep apnea category.

[0011] Optionally, the deep learning model determines that a sleep apnea event has occurred when the confidence level of the sleep apnea category obtained by the deep learning model based on the input pressure signal is greater than a first threshold, or when the confidence level of the sleep apnea category obtained by the deep learning model based on the input snoring signal is greater than a second threshold.

[0012] Optionally, when the confidence level of the apnea category obtained by the deep learning model based on the input pressure signal is greater than a third threshold and the confidence level of the apnea category obtained by the deep learning model based on the input snoring signal in the same time period is greater than a fourth threshold, the deep learning model determines that an apnea event has occurred, wherein the third threshold is less than the first threshold and the fourth threshold is less than the second threshold.

[0013] Optionally, the first threshold is 0.9 and the second threshold is 0.8.

[0014] Optionally, the third threshold is 0.8 and the fourth threshold is 0.7.

[0015] Optionally, the pressure signal is filtered and normalized before being input into the deep learning model.

[0016] Optionally, before the snoring signal is input into the deep learning model, the snoring signal is converted into a spectrogram and normalized.

[0017] Optionally, the deep learning model includes two parallel sub-models, namely the MobileNet V2 model and the MNetV2 model; the sub-model for the pressure signal input is the MobileNet V2 model, and the sub-model for the snoring signal input is the MNetV2 model.

[0018] Secondly, a sleep apnea recognition device includes:

[0019] The pressure acquisition module is used to acquire the pressure signal output by the deformation element (11) inside the pillow (1);

[0020] The snoring acquisition module is used to acquire snoring signals;

[0021] The identification module identifies the pressure signal and the snoring signal to determine whether a sleep apnea event has occurred.

[0022] Optionally, it also includes a preprocessing module, which is used to filter and normalize the pressure signal, and to convert the snoring signal into a spectrogram and normalize it.

[0023] Thirdly, a sleep apnea recognition system includes:

[0024] A pillow (1), the pillow (1) including a deformable element (11);

[0025] A pressure acquisition device is connected to the deformable element (11), and the pressure acquisition device is used to acquire the pressure signal of the deformable element (11);

[0026] A sound acquisition device, wherein the sound acquisition device is used to acquire snoring signals;

[0027] In the aforementioned sleep apnea recognition device, the pressure acquisition module is used to acquire the pressure signal acquired by the pressure acquisition element, and the snoring acquisition module is used to acquire the snoring signal acquired by the sound acquisition element.

[0028] Optionally, the deformable element (11) is an airbag, and the sleep apnea recognition system further includes an air pump (2), with a silencer connected between the air pump (2) and the airbag.

[0029] Fourthly, an electronic device includes a processor and a memory, the memory storing an application program, the processor being configured to run the application program within the memory to perform the steps in the sleep apnea recognition method.

[0030] Fifthly, a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the sleep apnea recognition method.

[0031] The beneficial effects of the present invention are as follows: The sleep apnea identification method, device and system proposed in the present invention can use the pressure change of the deformable element in the pillow and the snoring to determine whether a sleep apnea event has occurred. Compared with a single monitoring variable, such as only monitoring snoring or only monitoring the pressure change of the deformable element, it reduces the false negative rate of sleep apnea events and improves the accuracy of sleep apnea event identification.

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Attached Figure Description

[0033] Figure 1 is a flowchart of the sleep apnea recognition method in an embodiment of the present invention;

[0034] Figure 2 is a schematic diagram of a pressure signal that does not exhibit sleep apnea characteristics in an embodiment of the present invention;

[0035] Figure 3 is a schematic diagram of a pressure signal exhibiting sleep apnea characteristics in an embodiment of the present invention;

[0036] Figure 4 is a schematic diagram of a snoring signal exhibiting sleep apnea characteristics in an embodiment of the present invention.

[0037] Figure 5 is a functional block diagram of the sleep apnea recognition device in an embodiment of the present invention;

[0038] Figure 6 is a schematic diagram of the sleep apnea recognition system in an embodiment of the present invention;

[0039] Figure 7 is a schematic diagram of the internal structure of the pillow in an embodiment of the present invention;

[0040] Figure 8 is a schematic diagram of the hardware structure of the electronic device that performs the sleep apnea recognition method in an embodiment of the present invention;

[0041] In the diagram: 1. Pillow; 11. Deformable element; 2. Air pump; 3. Air tube. Detailed Implementation

[0042] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0043] In the description of this invention, unless otherwise explicitly specified and limited, the terms "connected," "linked," and "fixed" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0044] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" of the second feature includes the first feature directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature includes the first feature directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0045] In the description of this embodiment, the terms "upper," "lower," "left," and "right," etc., refer to the orientation or positional relationship shown in the accompanying drawings. They are used only for ease of description and simplification of operation, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention. In addition, the terms "first" and "second" are used only for distinction in description and have no special meaning.

[0046] In the description of this invention, it should be noted that the flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all contents and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0047] There are two types of sleep apnea syndrome: central sleep apnea (CSA) and obstructive sleep apnea (OSA) associated with airway obstruction. The most common type is central sleep apnea, which is associated with the failure of the body to automatically initiate and control the respiratory cycle at the appropriate time, typically causing the cessation of virtually all breathing efforts during sleep. This condition can develop after a heart attack and is often a contributing factor to heart failure and other cardiopulmonary diseases. Obstructive sleep apnea is generally characterized by repeated pauses in breathing during sleep due to upper airway obstruction or collapse. It is common in overweight individuals who snore or have an enlarged neck. When awake, muscle tone keeps the throat open; when asleep, the airway in the neck narrows and closes, and the body may exhibit breath-holding behaviors, during which breathing sounds / snoring are weak or even absent. When the patient wakes, the throat muscles open the airway, and snoring ceases.

[0048] Apnea typically lasts ≥10 seconds, manifesting as a decrease in respiratory flow of ≥90%. Due to airway narrowing or collapse, snoring is characterized by intermittent snoring or sudden interruption. Utilizing this characteristic, and referring to Figures 1, 6, and 7, an embodiment of the present invention proposes an apnea identification method. This method combines monitoring pressure values ​​and snoring sounds to detect whether an apnea event has occurred. The apnea identification method mainly includes the following steps:

[0049] Step 1: Acquire the pressure signal of the deformable element 11 in pillow 1, and simultaneously acquire the snoring signal. The deformable element 11 can be an air bladder, water bag, etc. The measurable pressure signal for an air bladder is an air pressure signal, and the measurable pressure signal for a water bag is a water pressure signal. When the user rests on pillow 1 and breathes normally, the pressure value of the deformable element 11 will change periodically with the breathing flow. Therefore, the change in pressure value can be used to determine the change in breathing flow, and thus the breathing status. It is understood that one or more deformable elements 11 can be installed in pillow 1. When multiple deformable elements 11 are installed, the pressure signal of the deformable element 11 with the highest pressure value is used as the reference.

[0050] Step 2: Identify and judge the pressure signal and snoring signal to determine whether there is sleep apnea.

[0051] Acquiring snoring signals alone can easily lead to misjudgments of non-apnea events. For example, a person waking up at night or not falling asleep might be mistaken for apnea. Conversely, acquiring only the pressure signal from the deformable element 11, in cases of obstructive apnea accompanied by respiratory effort, might result in pressure fluctuations resembling those during normal breathing, making it difficult to categorize as an apnea event and potentially leading to missed apnea cases. Therefore, this invention combines pressure and snoring signals to reduce the missed apnea event rate, thereby improving the accuracy of apnea event identification.

[0052] Specifically, when identifying and judging pressure signals and snoring signals, both are input into a deep learning model, which then identifies whether a sleep apnea event has occurred. In the pressure signal, a sleep apnea event is characterized by a halt in the sinusoidal waveform change. In the snoring signal, it is characterized by an interval between two snoring sounds exceeding a preset value, with the interval consisting of white noise. The current reasonable range for the preset value is 8-10 seconds, typically set to 10 seconds. This preset value may be adjusted in the future based on fluctuations in the amount of experimental data.

[0053] To reduce computational load and external noise interference, pressure and snoring signals need to be preprocessed before being input into the deep learning model.

[0054] The specific steps for preprocessing the pressure signal are as follows: The acquired pressure signal is plotted on a "time-pressure value" chart, and adjacent scatter points are connected to form a line graph as shown in Figure 2 or Figure 3 (the vertical axis represents the pressure value, the unit of the vertical axis is kPa*5000, the vertical axis represents the actual pressure magnified 5000 times, and the horizontal axis represents time); starting from time axis 0, a window is formed at fixed intervals of 60 seconds, and each time the corresponding part of the line graph is sequentially extracted and filtered to obtain a smooth curve segment to remove interference noise caused by the external environment. Then, the curve segment is normalized. At this point, the preprocessing of the pressure signal is complete, forming multiple segments of first respiratory feature images as shown in Figure 4, ordered by time. Each segment of the first respiratory feature image can be used as input features for the next stage of the deep learning model. These first respiratory feature images represent the respiratory features within the corresponding 60 seconds.

[0055] In one embodiment, the filtering of the above-mentioned line graph uses SG filtering, which, compared with other filtering methods, can better preserve waveform features (such as peaks, edges, etc.) and has good suppression of high-frequency noise, but it does not provide a smooth transition. The curve segment normalization is performed using OpenCV. OpenCV is mature and stable, supports Linux and ARM embedded systems, and also supports Python and C++, facilitating debugging. Finally, the first respiratory feature image formed by the pressure signal is normalized to a 640*480 image to retain sufficient features while reducing computational load. The benefits of normalization are: 1. Eliminating the influence of dimensionality differences; 2. Improving model convergence speed and stability. Neural networks are sensitive to feature scale; normalization can accelerate training and improve performance.

[0056] The specific steps for preprocessing the snoring signal are as follows: The acquired snoring signal, as shown in Figure 4, is segmented, extracting audio segments with snoring intervals greater than 10 seconds and less than 60 seconds. After frame segmentation, feature extraction and transformation are performed, for example using Discrete Fourier Transform (DFT / FFT), converting it into a spectrogram with a 5-second window. Then, OpenCV is used for normalization. At this point, the preprocessing of the snoring signal is complete, forming the second respiratory feature image. Each processed second respiratory feature image can be used as input features for the next stage of the deep learning model to determine whether apnea occurs within that time period. It can be understood that if no snoring signal is acquired, i.e., no snoring occurs, it indicates that the user is awake, and therefore there is no apnea event. Deep learning can be trained end-to-end, automatically learning effective features from raw data, making it particularly suitable for complex audio inputs.

[0057] Specifically, when using a deep learning model for analysis, the Softmax activation function is used to transform the first and second respiratory feature images into probability distributions, and the confidence scores for each apnea category are calculated. For ease of distinction, the confidence score for the apnea category in the first respiratory feature image is defined as the first confidence score, and the confidence score for the apnea category in the second respiratory feature image is defined as the second confidence score. If the first confidence score exceeds a preset first threshold, an apnea event is determined to have occurred. For the second respiratory feature image, if the second confidence score exceeds a preset second threshold, an apnea event is determined to have occurred.

[0058] Specifically, the deep learning model used to identify the first respiratory feature image can be a pre-trained MobileNet V2 classification network, with the model outputting a vector in the format [1,2]. In this embodiment, the format [1,2] indicates that the output is in a one-row, two-column format, with the two columns describing the probability of the corresponding binary classification task. For example, an output of [0.9, 0.1] indicates that the probability of a normal pressure signal is 90%, and the probability of an abnormal pressure signal is 10%. Compared to other deep learning models, MobileNet V2 is a lightweight classification network structure with lower computational cost, higher computational efficiency, and easier deployment. Compared to standard convolution, it reduces computation by approximately 8 to 9 times, has a smaller model file, and is faster. The model is very suitable for deployment on embedded devices, while also offering good compatibility and convenient deployment. The deep learning model used to identify second respiratory feature images can employ the MNetV2 model (the MNetV2 model is based on MobileNet V2, but by adding Global Average Pooling, Dropout, and Softmax layers, it further enhances the model's classification ability and adapts to multi-class classification tasks. The main change is in the classification head part, which is adapted to MobileNet V2 to focus more on classification tasks). The model outputs a vector in the format [1, 2]. In this embodiment, the format [1, 2] indicates that the output is in a one-row, two-column format, with the two columns describing the probabilities of the corresponding binary classification task. For example, an output of [0.6, 0.4] indicates that the probability of snoring is 60% and the probability of no snoring is 40%.

[0059] During the pre-training of the deep learning model, a first prediction model based on pressure signals and a second prediction model based on snoring signals were formed. By inputting standard test samples with pressure values ​​as the monitoring object into the first prediction model, and then analyzing the confidence level of the pressure signal under apnea events, setting the first threshold to 0.9 maximizes the accuracy in identifying apnea events. By inputting standard test samples with snoring signals as the object into the second prediction model, and analyzing the confidence level of the snoring signal under apnea events (i.e., the non-snoring segment), it was determined that setting the second threshold to 0.8 maximizes the accuracy in identifying apnea events.

[0060] It is worth emphasizing that when the first threshold is set to 0.9 and the second threshold is set to 0.8, although the accuracy of judging by pressure signal and snoring signal separately can reach the highest respectively, there are still some missed judgments for each.

[0061] Statistical analysis of the training process of the deep learning model shows that when the first confidence level of the deep learning model output is greater than the third threshold and the second confidence level for the same time period is greater than the fourth threshold, the deep learning model determines that a sleep apnea event has occurred. The third threshold is less than the first threshold and the fourth threshold is less than the second threshold, which further improves the accuracy of the determination.

[0062] Based on the pre-training results of the deep learning model, the accuracy is high when the third threshold is 0.8 and the fourth threshold is 0.7.

[0063] Referring to Figure 5, an embodiment of the present invention also proposes a sleep apnea detection device 4, including a pressure acquisition module 401, a snoring acquisition module 402, and a recognition module 404. The pressure acquisition module 401 is used to acquire the pressure signal of the deformable element 11 inside the pillow 1; the snoring acquisition module 402 is used to acquire the snoring signal; the recognition module 404 is connected to the snoring acquisition module 402 and the pressure acquisition module 401, and the recognition module 404 is used to analyze the pressure signal and the snoring signal and output the detection result of the sleep apnea event.

[0064] In accordance with the preprocessing of pressure signal and snoring signal in the sleep apnea recognition method, the sleep apnea recognition device 4 also includes a preprocessing module 403. The preprocessing module 403 is used to filter and normalize the pressure signal, and to convert the snoring signal into a spectrum and normalize it.

[0065] Other specific implementations of the apnea recognition device 4 are the same as those of the apnea recognition method. For example, the specific recognition method of the recognition module 404 for apnea events will not be described in detail in this embodiment.

[0066] Referring to Figures 6 and 7, an embodiment of the present invention also proposes a sleep apnea recognition system, including a pillow 1, a pressure acquisition element, a sound acquisition element, and a sleep apnea recognition device 4 as shown in Figure 5. The deformable element 11 is a component of the pillow 1, meaning the pillow 1 includes the deformable element 11. The sound acquisition element is placed around or on the pillow 1 and connected to the snoring acquisition module 402. The snoring acquisition module 402 uses a specific algorithm (e.g., by setting a sound decibel limit; when the ambient sound exceeds this limit, it is judged as snoring; this is prior art and will not be elaborated here) to determine and mark whether the sound acquired by the sound acquisition element is snoring, thereby obtaining the original snoring signal. This is prior art and will not be elaborated here. The pressure acquisition element is set inside the pillow 1 and connected to the pressure acquisition module 401. The recognition module 404 and the preprocessing module 403 can be set on a smart terminal such as a mobile phone or computer. Specifically, the pressure acquisition element can be either a pressure sensor or a gravity sensor, which is connected to or abuts against the deformable element 11. The sound acquisition element can be a microphone. To avoid the sound sensor interfering with the user's sleep, the sound sensor is placed inside, outside, or around the pillow 1 and is not worn on the user's body.

[0067] Based on the premise that the deformable element 11 is an airbag, the above-mentioned sleep apnea recognition system also includes an air pump 2. The air pump 2 is connected to the deformable element 11 through an air tube 3 to realize the inflation and deflation of the deformable element 11. Specifically, the air pump 2 includes an inflation pump and a deflation pump. The sleep apnea recognition system also includes a solenoid valve connecting the inflation pump, the deflation pump, and the deformable element 11. The solenoid valve is communicatively connected to the main control board in the controller to achieve selective on / off switching among the inflation pump, the deflation pump, and the deformable element 11 under the control of the main control board.

[0068] To prevent noise generated during inflation and deflation from affecting the collection of snoring signals, the sleep apnea recognition system also includes silencers. Silencers are installed between the suction pump and the deformable element 11, and between the inflation pump and the deformable element 11. Furthermore, to improve the silencer effect, at least two silencers are connected in series between the suction pump 2 and the deformable element 11, and between the inflation pump 2 and the deformable element 11.

[0069] Referring to Figure 8, which is a schematic diagram of the hardware structure of the electronic device 80 for the sleep apnea recognition method provided in this embodiment, the electronic device 80 includes:

[0070] One or more processors 801 and memory 802, with one processor 801 as an example in Figure 8.

[0071] The processor 801 and the memory 802 can be connected via a bus or other means. Figure 8 shows an example of a connection via a bus.

[0072] The memory 802, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the sleep apnea recognition method in the embodiments of this application. The processor 801 executes various functional applications of the terminal device or server and completes data processing by running the non-volatile software programs, instructions, and modules stored in the memory 802, thereby implementing the sleep apnea recognition method of the above-described method embodiments.

[0073] The memory 802 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the sleep apnea recognition device. Furthermore, the memory 802 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 802 may optionally include memory remotely located relative to the processor 801, and this remote memory may be connected to the sleep apnea recognition device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0074] The one or more modules are stored in the memory 802. When executed by the one or more processors 801, they execute the sleep apnea recognition method in any of the above method embodiments. For example, they execute the method steps in FIG1 described above to realize the functions of modules 401-404 in FIG5.

[0075] The electronic devices described in this application exist in various forms, including but not limited to:

[0076] (1) Mobile communication devices: These devices are characterized by their mobile communication capabilities and primarily aim to provide voice and data communication. These terminals include: smartphones (e.g., iPhones), multimedia phones, feature phones, and low-end phones, etc.

[0077] (2) Ultra-mobile personal computer devices: These devices fall under the category of personal computers, possessing computing and processing capabilities, and generally also have mobile internet access features. These terminals include PDAs, MIDs, and UMPCs, such as the iPad.

[0078] (3) Server: A device that provides computing services. The components of a server include a microcontroller, hard disk, memory, system bus, etc. Servers are similar to general computer architectures, but because they need to provide highly reliable services, they have higher requirements in terms of processing power, stability, reliability, security, scalability, and manageability.

[0079] (4) Other electronic devices with data interaction functions.

[0080] Accordingly, in the embodiments of this application, a computer-readable storage medium is also provided, which may be a readable signal medium or a readable storage medium. It stores a program product capable of implementing the sleep apnea recognition method described above. In some possible embodiments, various aspects of the present invention may also be implemented as a program product comprising program code, which, when run on a terminal device, causes the terminal device to perform the steps according to various exemplary embodiments of this disclosure as described in the "Exemplary Methods" section of this specification.

[0081] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software and a general-purpose hardware platform, or of course, using hardware. Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0082] The descriptions of the above embodiments of the electronic device and computer-readable storage medium are similar to those of the above method embodiments, and have similar beneficial effects. For technical details not disclosed in the embodiments of the electronic device and computer-readable storage medium of the present invention, please refer to the descriptions of the method embodiments in this paper for understanding.

[0083] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will be able to make various obvious changes, readjustments, and substitutions without departing from the scope of protection of the present invention. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention. Industrial applicability

[0084] This invention relates to the field of respiratory monitoring technology. Its technical solution possesses industrial applicability, enabling industrial manufacturing and use, and can produce positive technical effects. This invention acquires pressure signals and snoring signals simultaneously through a deformable element within a pillow. A deep learning model is used to comprehensively analyze both signals to determine whether a sleep apnea event has occurred. This solution can be implemented using only a pressure acquisition component (such as a pressure sensor) and a sound acquisition component (such as a microphone) built into the pillow, without affecting the user's sleep quality. The pressure signal is filtered and normalized, and the snoring signal is converted into a spectrum and normalized. The preprocessing algorithm is mature and stable. The deep learning model uses lightweight MobileNet V2 and MNetV2, which have low computational requirements and are easy to deploy in embedded devices or smart terminals (mobile phones, computers, etc.). The deformable element in the pillow can be an air bladder, which, together with an air pump and a muffler, allows for inflation and deflation regulation. The relevant components (air bladder, air pump, solenoid valve, pressure sensor, microphone, etc.) are all industrially mature components with mature manufacturing processes, convenient assembly, and suitability for mass production. The sleep apnea recognition system of this invention can be integrated into smart pillows, smart home devices, or standalone monitoring devices, and can be widely used in homes, hospitals, elderly care institutions, sleep laboratories, and other scenarios. It can improve the screening efficiency and recognition accuracy of sleep apnea syndrome, facilitating early intervention and treatment, and has good socio-economic benefits and industrial applicability. The overall solution is stable, reliable, and cost-controllable, meeting industrial implementation requirements and possessing good industrial applicability and market application value.

Claims

1. A method of apnea identification, the method comprising: The sleep apnea identification method includes the following steps: The pressure signal of the deformable element (11) in the pillow (1) is acquired, and the snoring signal is acquired at the same time; The pressure signal and the snoring signal are identified and judged to determine whether a sleep apnea event has occurred.

2. The apnea identification method of claim 1, wherein, When identifying and judging the pressure signal and the snoring signal, the pressure signal and the snoring signal are input into a deep learning model, and the deep learning model determines whether a sleep apnea event has occurred based on the confidence level of the sleep apnea category.

3. The apnea identification method of claim 2, wherein, If the confidence level of the apnea category obtained by the deep learning model based on the input pressure signal is greater than a first threshold, or if the confidence level of the apnea category obtained by the deep learning model based on the input snoring signal is greater than a second threshold, the deep learning model determines that an apnea event has occurred.

4. The apnea identification method of claim 3, wherein, When the confidence level of the apnea category obtained by the deep learning model based on the input pressure signal is greater than a third threshold and the confidence level of the apnea category obtained by the deep learning model based on the snoring signal of the same time period is greater than a fourth threshold, the deep learning model determines that an apnea event has occurred. The third threshold is less than the first threshold and the fourth threshold is less than the second threshold.

5. The apnea identification method of claim 3, wherein, The first threshold is 0.9, and the second threshold is 0.

8.

6. The apnea identification method of claim 4, wherein, The third threshold is 0.8, and the fourth threshold is 0.

7.

7. The apnea identification method of claim 2, wherein, Before the pressure signal is input into the deep learning model, the pressure signal is filtered and normalized.

8. The apnea identification method of claim 2, wherein, Before the snoring signal is input into the deep learning model, the snoring signal is converted into a spectrogram and normalized.

9. The apnea identification method of claim 2, wherein, The deep learning model includes two parallel sub-models, namely the MobileNet V2 model and the MNetV2 model; the sub-model for the pressure signal input is the MobileNet V2 model, and the sub-model for the snoring signal input is the MNetV2 model.

10. An apnea identification apparatus, characterized by, include: The pressure acquisition module is used to acquire the pressure signal output by the deformation element (11) inside the pillow (1); The snoring acquisition module is used to acquire snoring signals; The identification module identifies the pressure signal and the snoring signal to determine whether a sleep apnea event has occurred.

11. The apnea identification device of claim 10, wherein, It also includes a preprocessing module, which is used to filter and normalize the pressure signal, and to convert the snoring signal into a spectrogram and normalize it.

12. An apnea identification system characterized by, include: A pillow (1), the pillow (1) including a deformable element (11); A pressure acquisition device is connected to the deformable element (11), and the pressure acquisition device is used to acquire the pressure signal of the deformable element (11); A sound acquisition device, wherein the sound acquisition device is used to acquire snoring signals; The sleep apnea recognition device as described in claim 10 or 11, wherein the pressure acquisition module is used to acquire the pressure signal acquired by the pressure acquisition element, and the snoring acquisition module is used to acquire the snoring signal acquired by the sound acquisition element.

13. The apnea identification system of claim 11, wherein, The deformable element (11) is an airbag, and the sleep apnea recognition system also includes an air pump (2), with a silencer connected between the air pump (2) and the airbag.

14. An electronic device, comprising: comprising a processor and a memory, the memory storing an application, the processor configured to run the application in the memory to perform the steps in the apnea identification method of any one of claims 1-9.

15. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, which, when executed by a processor, causes the processor to implement the apnea identification method of any one of claims 1-9.