A sleep breathing monitoring method, wearable device and storage medium

Abstract

This invention relates to the field of smart wearable device technology, and discloses a sleep breathing monitoring method, wearable device, and storage medium. The method uses a vibration signal as a reference signal, performs noise reduction processing on the acquired audio signal to obtain a pure respiratory airflow signal, and classifies the current breathing state based on the amplitude of the vibration signal and the signal strength of the pure respiratory airflow signal. When the vibration signal amplitude is higher than a first threshold and the pure respiratory airflow signal strength is lower than a second threshold, it is determined to be obstructive sleep apnea; when both are lower than the corresponding thresholds, it is determined to be central sleep apnea. This invention utilizes the physical complementarity of vibration and audio signals in their signal sources to remove environmental noise and can accurately distinguish between obstructive and central sleep apnea through a two-dimensional determination matrix of dual-mode signals.

Description

Technical Field

[0001] This invention relates to the field of smart wearable device technology, and in particular to a sleep breathing monitoring method, wearable device, and storage medium. Background Technology

[0002] Sleep apnea syndrome is a common sleep disorder, divided into two types: obstructive sleep apnea (OSA) and central sleep apnea (CSA). OSA is caused by physical obstruction of the airway and is characterized by respiratory effort but no airflow; CSA is caused by the brain not issuing breathing commands and is characterized by neither respiratory effort nor airflow. Both can lead to serious consequences such as hypertension and heart disease in the long term. Clinically, diagnosis usually requires polysomnography, but this method requires multiple sensors connected in specialized institutions, is costly, and disrupts natural sleep, making it difficult to widely apply in everyday situations.

[0003] Currently, consumer-grade wearable devices mainly employ two approaches for sleep apnea monitoring. One approach is based on vibration sensors (bone conduction). This approach can collect bone vibration signals caused by respiratory movements and has strong resistance to environmental noise, but it has significant limitations: when the device has poor contact with the skin, the user lies on their side causing pressure on the sensor, or central sleep apnea occurs, the vibration signal will disappear or weaken. The device cannot accurately distinguish between central sleep apnea and poor sensor contact, which are two completely different states, leading to missed detections or false alarms.

[0004] Another approach is based on an air conduction microphone, which can directly collect breathing airflow sounds and snoring sounds, providing a wealth of information. However, it is highly susceptible to environmental noise interference, including snoring from sleeping partners, air conditioning sounds, and other ambient noises, resulting in a very high false alarm rate. Furthermore, continuous collection of ambient sounds also poses a risk of privacy leaks.

[0005] Both of the above solutions face a trade-off between power consumption and accuracy. Keeping the microphone continuously on to ensure noise cancellation leads to excessive power consumption, while turning it off to save power results in the loss of airflow information; existing solutions lack an effective dynamic balancing mechanism. Furthermore, because they cannot accurately distinguish between OSA and CSA, existing consumer-grade devices struggle to provide differentiated interventions for different types of sleep apnea, resulting in limited intervention effectiveness. Summary of the Invention

[0006] The purpose of this invention is to provide a sleep breathing monitoring method, wearable device, and storage medium, which solves the problem that existing single-sensor solutions cannot simultaneously achieve effective noise reduction and accurate type differentiation.

[0007] To address the aforementioned technical problems, this invention provides a sleep breathing monitoring method, comprising: Vibration signals are collected as reference signals, and noise reduction processing is performed on the collected audio signals to obtain pure breathing airflow signals; Based on the amplitude of the vibration signal and the signal intensity of the pure respiratory airflow signal, the current respiratory state is classified and determined. When the amplitude of the vibration signal is higher than a first threshold and the signal intensity of the pure respiratory airflow signal is lower than a second threshold, it is determined to be obstructive sleep apnea. When the amplitude of the vibration signal is lower than the first threshold and the signal intensity of the pure respiratory airflow signal is lower than the second threshold, it is determined to be central sleep apnea.

[0008] Optionally, the noise reduction process employs an adaptive filtering algorithm or a deep learning mask network, using the vibration signal as a reference input and the audio signal as the main input, to extract the user's breathing components related to the vibration signal and remove environmental noise components unrelated to the vibration signal.

[0009] Optionally, the classification determination further includes: when the amplitude of the vibration signal and the signal strength of the pure breathing airflow signal are both lower than the third threshold, it is determined that the wearable device has fallen off.

[0010] Optionally, the vibration signal is continuously collected. When the amplitude of the vibration signal is lower than a preset percentage of the baseline value within a preset number of consecutive breathing cycles, the audio signal is collected to perform the noise reduction processing and the classification determination. After the classification determination is completed, the acquisition of the audio signal is stopped.

[0011] Optionally, it also includes: when the acquisition of the audio signal is triggered, and normal breathing airflow is detected after the noise reduction processing, it is determined that the vibration signal is misjudged due to poor contact, the event is ignored, and the weight of the vibration signal in the classification judgment is calibrated.

[0012] Optionally, it also includes: determining the user's sleep state based on the collected motion data, and triggering the collection of the audio signal after confirming that the user has fallen asleep.

[0013] Optionally, it also includes performing differentiated interventions based on the classification results: When obstructive sleep apnea is detected, a sound wave or vibration signal of a specific frequency is output; When central sleep apnea is diagnosed, a progressively stronger tactile pulse or a sound with a specific rhythm is output. Optionally, if the intervention fails after a preset number of consecutive attempts or if the heart rate exceeds a preset abnormal threshold, the intervention will be stopped and an external terminal alarm will be triggered.

[0014] Optionally, it also includes: collecting physiological characteristic signals and classifying them based on the vibration signal, the pure respiratory airflow signal, and the physiological characteristic signals; Optionally, the weights of the vibration signal, the pure respiratory airflow signal, and the physiological characteristic signal in the classification determination can be dynamically adjusted by combining the collected posture data.

[0015] The present invention also provides a wearable device, including a vibration sensor, a microphone, a processor, and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the methods described above.

[0016] Optionally, the vibration sensor is used to collect bone conduction vibration signals, and the microphone is used to collect breathing airflow sounds and environmental noise; Optionally, it also includes an inertial measurement unit for monitoring the user's head posture, body movements, and wearing status; Optionally, it also includes a PPG sensor for collecting at least one of the user's heart rate, blood oxygen saturation, and heart rate variability; Optionally, it may also include an actuator, which includes at least one of a speaker and a linear motor, for outputting an intervention signal.

[0017] The present invention also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0018] The beneficial effects of this invention are as follows: This invention uses vibration signals as reference signals to denoise audio signals. By leveraging the physical complementarity between vibration and audio signals, environmental noise is removed, solving the industry problem of high false alarm rates caused by environmental noise interference such as snoring from a partner in a home sleep setting. At the same time, it constructs a classification and judgment matrix based on the amplitude of vibration signals and the intensity of pure respiratory airflow signals, which enables the differentiation and identification of obstructive sleep apnea and central sleep apnea on wearable devices.

[0019] Furthermore, by triggering audio signal acquisition only when a suspected abnormality is detected, a dynamic balance between high-precision monitoring and low-power operation is achieved; by using the characteristic that both mode signals are below the threshold at the same time, wear-wearing drop events are accurately identified, eliminating a major source of false alarms for wearable devices; and by implementing differentiated intervention strategies for obstructive and central sleep apnea based on classification results, the effectiveness of intervention and user comfort can be significantly improved. Attached Figure Description

[0020] Figure 1 This is a flowchart of a sleep breathing monitoring method according to an embodiment of the present invention; Figure 2This is a block diagram illustrating the principle of VPU-assisted air-conducting microphone adaptive noise reduction in an embodiment of the present invention. Figure 3 This is a schematic diagram of the low-power triggering mechanism in an embodiment of the present invention; Figure 4 This is a two-dimensional decision matrix diagram in an embodiment of the present invention; Figure 5 This is a schematic diagram of the OSA path in an embodiment of the present invention; Figure 6 This is a schematic diagram of the OSA intervention process in an embodiment of the present invention; Figure 7 This is a schematic diagram of the CSA path in an embodiment of the present invention; Figure 8 This is a schematic diagram of the CSA intervention process in an embodiment of the present invention; Figure 9 This is a schematic diagram of a wearable device structure according to an embodiment of the present invention.

[0021] In the diagram: 1. Vibration sensor; 2. Microphone; 3. PPG sensor; 4. Inertial measurement unit; 5. Speaker; 6. Processor; 7. Bluetooth communication module. Detailed Implementation

[0022] The present invention will now be described with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. It should be understood that those skilled in the art can modify the invention described herein while still achieving its advantageous effects. Therefore, the following description should be understood as being of broad knowledge to those skilled in the art and is not intended to limit the invention.

[0023] The serial numbers assigned to components in this document, such as "first," "second," etc., are merely used to distinguish the described objects and have no sequential or technical meaning. The terms "connection" and "linkage" used in this application, unless otherwise specified, include both direct and indirect connections (linkages). In the description of this invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, 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 invention.

[0024] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0025] Based on the teachings of this specification, those skilled in the art can form new technical solutions through cross-combination of different implementation methods without creating technical contradictions. Such variations should all be considered to fall within the protection scope of this invention.

[0026] The invention is described more specifically by way of example in the following paragraphs with reference to the accompanying drawings. The advantages and features of the invention will become clearer from the following description. It should be noted that the drawings are in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the invention.

[0027] Example 1 Please refer to Figure 1 This invention provides a sleep breathing monitoring method, comprising: S1: Collect vibration signals as reference signals, perform noise reduction processing on the collected audio signals, and obtain pure breathing airflow signals; S2: Based on the amplitude of the vibration signal and the signal intensity of the pure respiratory airflow signal, the current respiratory state is classified and determined. When the amplitude of the vibration signal is higher than the first threshold and the signal intensity of the pure respiratory airflow signal is lower than the second threshold, it is determined to be obstructive sleep apnea. When the amplitude of the vibration signal is lower than the first threshold and the signal intensity of the pure respiratory airflow signal is lower than the second threshold, it is determined to be central sleep apnea.

[0028] The sleep breathing monitoring method provided in this embodiment collects vibration signals and audio signals, uses vibration signals to reduce noise in the audio signals, and classifies and determines the breathing state based on the combined characteristics of the dual-mode signals, thereby achieving the distinction between obstructive sleep apnea and central sleep apnea in complex noise environments.

[0029] Specifically, in step S1, a vibration signal is acquired as a reference signal, and an audio signal is acquired as the main input signal.

[0030] In a specific example, the vibration signal is a bone conduction vibration signal, which is collected by a vibration sensor that is close to the ear canal or auricular bone. It can reflect the user's own breathing effort, heartbeat and bone conduction snoring. Due to the physical characteristics of the bone conduction path, environmental noise (such as partner snoring, air conditioner noise, etc.) will not be conducted to the vibration sensor. Therefore, the vibration signal naturally does not contain environmental noise components.

[0031] It should be noted that, in this embodiment, "pure" in pure breathing airflow signal means that environmental noise components unrelated to the vibration signal have been removed. The resulting signal mainly reflects the user's own breathing airflow characteristics, but does not exclude the inclusion of other physiological sound components from the user.

[0032] In one specific example, the audio signal is acquired by an air conduction microphone located at the headphone vent or the ear canal, which can directly capture the respiratory airflow sound and snoring spectrum in the ear canal, but also contains environmental noise components, and is represented as a mixed signal of user breathing sound and environmental noise.

[0033] The above two signal acquisition methods utilize the complementary characteristics of vibration signals and audio signals in terms of signal source. The user's own breathing / snoring will appear in both vibration and sound forms at the same time, and the two have time synchronization and energy correlation, while environmental noise only appears in the audio signal.

[0034] Furthermore, in step S1, the audio signal is subjected to noise reduction processing to obtain a pure breathing airflow signal.

[0035] In a specific example, using the vibration signal as the reference input and the audio signal as the main input, an adaptive filtering algorithm or a deep learning mask network is used for noise reduction processing to extract the user's breathing components related to the vibration signal and remove environmental noise components unrelated to the vibration signal to obtain a pure breathing airflow signal.

[0036] In one specific example, the adaptive filtering algorithm employs the Normalized Least Mean Square (NLMS) algorithm. Using the audio signal as the main input, it extracts the user's breathing components related to the vibration signal and removes irrelevant environmental noise components.

[0037] Scenario: The user is sleeping, and his / her partner is snoring next to him / her (ambient noise 60dB).

[0038] The mathematical form is expressed as: The mixed signal acquired by the air conduction microphone is: S_mic=S_user_breath+S_partner_snore+S_noise; The pure signal collected by the vibration sensor is S_vpu=S_user_vibration, which does not contain environmental noise components such as partner snoring.

[0039] The cross-correlation function between S_vpu and S_mic was calculated, revealing a high correlation between S_user_breath and S_vpu, while S_partner_snore remained uncorrelated. Based on this, an adaptive filter was constructed to generate the estimated environmental noise signal. Outputs a pure breathing airflow signal .

[0040] Please refer to Figure 2 The noisy mixed signal S_mic collected by the air conduction microphone is directly input into the adder; the clean reference signal S_vpu collected by the vibration sensor enters the adaptive filter, which estimates the ambient noise signal based on this signal. The inverted value is then input into the adder; the adder will then combine S_mic with... The summation outputs a clean breathing airflow signal S_clean; the error signal e(n) is fed back from the adder to the adaptive filter, driving the filter to iteratively update its weights, making... By continuously approximating real-world environmental noise, we can ultimately achieve precise separation of environmental noise.

[0041] In another specific example, the noise reduction process employs a deep learning mask network, using the time-frequency characteristics of the vibration signal as a conditional input to generate a frequency domain mask for the audio signal. The mask is then applied to the spectrum of the audio signal, preserving the user's breathing-related frequency bands, suppressing environmental noise frequency bands, and outputting a clean breathing airflow signal.

[0042] In this embodiment, the noise reduction processing can solve the technical problem of high false alarm rate caused by environmental noise interference in traditional air conduction microphone solutions. Even in high-noise environments such as when a partner snores or when the air conditioner is on, it can accurately extract the user's unique breathing airflow sound. At the same time, since there is no need to continuously record ambient sound, it effectively protects the user's privacy.

[0043] Furthermore, in step S1, a low-power triggering mechanism can be further incorporated to optimize power consumption performance.

[0044] For details, please refer to Figure 3 The vibration sensor operates continuously in low-power mode, monitoring the amplitude changes of the vibration signal in real time and comparing them with a dynamic baseline. When the amplitude of the vibration signal is lower than a preset proportion of the baseline value within a preset number of consecutive breathing cycles, the acquisition of the audio signal is triggered, and the noise reduction processing is performed. After the subsequent classification and determination are completed, the acquisition of the audio signal is immediately stopped, the air conduction microphone is turned off, the system resumes low-power operation mode, and waits for the next trigger.

[0045] In a specific example, the preset breathing cycle is 3, and the preset ratio is 50%.

[0046] In this embodiment, a low-power vibration sensor is used as a continuous trigger, which only wakes up the acquisition of audio signals when a suspected abnormality is detected, thus resolving the contradiction between power consumption and accuracy.

[0047] Furthermore, in a specific example, the user's sleep state can be determined by combining the collected motion data. The audio signal acquisition and subsequent monitoring process are only triggered after confirming the user is asleep, avoiding invalid monitoring during the user's waking activity phase, further reducing overall power consumption and minimizing false alarms. The specific implementation of the sleep determination in this embodiment is a common practice among those skilled in the art and will not be elaborated further.

[0048] Furthermore, if normal breathing airflow is detected after the noise reduction process when the audio signal is triggered, it indicates that the decrease in vibration signal amplitude is not caused by apnea, but by a misjudgment caused by poor contact of the vibration sensor. In this case, the event is ignored, and the weight of the vibration signal in the subsequent classification judgment is automatically calibrated to reduce the trigger sensitivity of the vibration sensor and avoid repeated false triggering.

[0049] Furthermore, in step S2, the current respiratory state is classified and determined.

[0050] In this embodiment, based on the amplitude of the vibration signal and the signal strength of the pure breathing airflow signal, please refer to... Figure 4 A two-dimensional judgment matrix is ​​constructed to classify and determine the current respiratory state.

[0051] It should be noted that, Figure 4 The two-dimensional decision matrix is ​​represented by coordinates, with the positive direction indicating presence and the negative direction indicating absence.

[0052] For details, please refer to Figure 5 When the amplitude of the vibration signal is higher than the first threshold and the signal strength of the pure breathing airflow signal is lower than the second threshold, it is determined to be obstructive sleep apnea (OSA).

[0053] This state indicates that the user is making a breathing effort, the vibration signal amplitude is normal, but the airway is physically blocked, preventing airflow. The airflow sound is weak or absent, which is consistent with the pathological characteristics of obstructive sleep apnea.

[0054] Please refer to Figure 7When the amplitude of the vibration signal is lower than a first threshold and the signal intensity of the pure respiratory airflow signal is lower than a second threshold, it is determined to be central apnea (CSA). This state indicates that the user has no respiratory effort, the vibration signal amplitude is low, there is no airflow, and the airflow sound is absent, which is consistent with the pathological characteristics of central apnea, that is, the brain does not issue a breathing command, resulting in complete cessation of breathing.

[0055] Furthermore, the two-dimensional decision matrix also includes the following decision logic: When the amplitude of the vibration signal is higher than the first threshold and the signal strength of the pure breathing airflow signal is higher than the second threshold, it is determined to be normal breathing, and the baseline is continuously monitored and updated.

[0056] When the amplitude of the vibration signal is lower than the first threshold and the signal strength of the pure breathing airflow signal is higher than the second threshold, it is determined to be a sensor malfunction or artifact, no alarm is triggered, the abnormal event is recorded and continuously monitored.

[0057] When both the amplitude of the vibration signal and the signal strength of the pure breathing airflow signal are below the third threshold, it is determined that the device is worn abnormally, monitoring is suspended and the breathing apnea alarm is not triggered.

[0058] In a single VPU (Vibration Pickup Unit) solution, the absence of vibration signal during central apnea is identical to the absence of signal when the earphone is detached, leading to numerous false alarms. This invention accurately identifies detachment events by utilizing the simultaneous absence of dual-mode signals, thus eliminating this source of false alarms.

[0059] In a specific example, the abnormal wearing determination also incorporates the attitude data of the inertial measurement unit. When both the dual-mode signals are extremely low and the inertial measurement unit detects a sudden change in the device angle, it is determined as a case of the device falling off, further improving the accuracy of the determination.

[0060] In this embodiment, the technical problem that a single sensor solution cannot distinguish between obstructive and central sleep apnea is solved, providing a precise pathological classification basis for subsequent targeted and differentiated interventions.

[0061] Furthermore, in this embodiment, physiological feature signals can be introduced as auxiliary verification indicators to participate in the classification determination.

[0062] In one specific example, the physiological characteristic signal is a blood oxygen saturation signal, which is acquired by a PPG (Photoplethysmography) sensor.

[0063] Specifically, please refer to Table 1 below. When the classification result is suspected obstructive sleep apnea or central sleep apnea, a secondary confirmation is performed by combining the decreasing trend of blood oxygen saturation. The time correlation between the continuous decrease of blood oxygen saturation and the apnea event can significantly improve the confidence of the judgment and reduce misjudgment.

[0064] Table 1 In another specific example, the physiological characteristic signal may also include a heart rate variability (HRV) signal, which is used to assist in assessing the state of the autonomic nervous system and further improve the accuracy of classification.

[0065] Furthermore, in this embodiment, the weights of the vibration signal, the pure respiratory airflow signal, and the physiological characteristic signal in the classification judgment can be dynamically adjusted by combining the collected posture data.

[0066] Specifically: In a quiet environment, the weight of the pure breathing airflow signal is increased, and its rich spectral information is used to assist in sleep staging; in a high-noise environment, the system automatically switches to vibration signal as the dominant signal, with the pure breathing airflow signal only used for auxiliary verification; in side-lying or compression scenarios, when the inertial measurement unit detects posture changes that cause poor contact between the vibration sensor and the bones, the weight of the vibration signal is automatically reduced, and the weight of the pure breathing airflow signal and physiological characteristic signal is increased to ensure monitoring stability under various sleeping positions.

[0067] In this embodiment, the aforementioned dynamic weight adjustment mechanism enables the system to adapt to different sleep scenarios and maintain stable monitoring accuracy under interference factors such as environmental noise and changes in sleeping posture.

[0068] Furthermore, the monitoring method provided by the present invention also includes step S3: performing differentiated intervention based on the classification judgment result.

[0069] Specifically, when obstructive sleep apnea is diagnosed, a specific frequency sound wave or vibration signal is output to stimulate the tension of the throat muscles, prevent airway collapse, and expand the airway.

[0070] In a specific example, the specific frequency is 70-100Hz. Based on the physical characteristics of soft tissue collapse in the pharynx, micro-vibrations in this frequency range can effectively stimulate the contraction of the pharyngeal muscles, and the vibration frequency is positively correlated with the pharyngeal muscle tension.

[0071] When central sleep apnea is diagnosed, progressively stronger tactile pulses or sounds of a specific rhythm are output to stimulate the brain to resume breathing and awaken the respiratory center.

[0072] In one specific example, the tactile pulse frequency is 0.5-2Hz. Based on the neural stimulation effect of low-frequency vibration on the brainstem respiratory center, the progressively enhanced tactile pulse can effectively activate the respiratory center without causing excessive awakening.

[0073] For a specific example, please refer to Figure 6 and Figure 8 This includes the following situations: Scenario A: At a certain moment, S_vpu shows significant vibration (the user is breathing forcefully), but S_clean shows no airflow sound; this is diagnosed as obstructive sleep apnea (OSA). Initiate airway dilation micro-vibration intervention, focusing on airway dilation by playing low-frequency sound waves or micro-vibrations of a specific frequency to stimulate pharyngeal muscle tone and prevent airway collapse.

[0074] Scenario B: At a certain moment, S_vpu has no vibration, S_clean has no sound, and PPG blood oxygen levels decrease; this is diagnosed as central apnea (CSA). A progressive intervention to awaken the respiratory center is initiated, focusing on stimulating the respiratory center using progressively stronger tactile impulses or rhythmic sounds to encourage the brain to resume breathing commands.

[0075] Scenario C: S_vpu and S_mic simultaneously and suddenly drop to zero, and the inertial measurement unit (IMU) detects a sudden change in the headphone angle; this is determined to be a headphone detachment. Monitoring is paused, and no alarm is triggered.

[0076] In this embodiment, the above-mentioned differentiated intervention strategy designs corresponding intervention methods for different pathological mechanisms of OSA and CSA. Compared with the traditional fixed threshold loud wake-up scheme, it significantly improves the success rate of intervention, while effectively protecting sleep structure and improving user comfort.

[0077] In a specific example, the differentiated intervention is implemented according to the following tiered intensities to achieve an optimal balance between intervention effectiveness and sleep protection: Level 1 (Invisible Induction): The intervention duration is less than the preset first duration, outputting micro-vibration or white noise with extremely low intensity, without waking the user, prioritizing the protection of the sleep structure. In a specific example, the first duration is 15 seconds.

[0078] Level 2 (tactile alert): When the intervention duration reaches the first duration, or when the current sleep stage is deep sleep, it is upgraded to non-periodic pulse vibration to enhance the stimulation effect. The determination of sleep stage can be achieved based on the fusion data of the inertial measurement unit and the PPG sensor, which is a common method used by those skilled in the art and will not be described in detail here.

[0079] Level 3 (Auditory Arousal): When the intervention duration exceeds the preset second duration or the blood oxygen saturation is lower than the preset threshold, the intervention is upgraded to a gradually increasing auditory arousal sound to forcibly wake the user to resume breathing. In a specific example, the second duration is 30 seconds.

[0080] If intervention fails after a preset number of consecutive attempts or if the heart rate exceeds a preset abnormal threshold, the intervention will stop and an external terminal alarm will be triggered.

[0081] In one specific example, the preset number of times is 3. The safety circuit breaker mechanism, as a last resort, immediately stops all local stimulation when intervention is ineffective or dangerous signals such as abnormal heart rate occur. It then sends a strong alarm via Bluetooth to the mobile phone, notifying the user or their family to take further measures, effectively preventing safety risks caused by delayed intervention.

[0082] In summary, this invention uses vibration signals as a reference signal to denoise audio signals. Leveraging the physical complementarity between vibration and audio signals at their source, it removes environmental noise components from the audio signal at the algorithmic level, obtaining a pure respiratory airflow signal with a high signal-to-noise ratio. This solves the technical problem of high false alarm rates caused by environmental noise interference in traditional air conduction microphone solutions during home sleep scenarios. Furthermore, a two-dimensional judgment matrix is ​​constructed based on the vibration signal amplitude and the intensity of the pure respiratory airflow signal. The combined characteristics of the dual-mode signals accurately distinguish between obstructive apnea and central apnea. Simultaneously, the feature of simultaneous absence of both dual-mode signals is used to identify abnormal device wearing, solving the problem of being unable to distinguish between central apnea and device detachment.

[0083] Example 2 This embodiment provides a wearable device for implementing the sleep breathing monitoring method described in the above embodiment. The wearable device can be in the form of a TWS (True Wireless Stereo) earphone, but this embodiment is not limited to this; other ear-worn wearable devices are also applicable.

[0084] Please refer to Figure 9 The wearable device includes a vibration sensor 1, a microphone 2, a processor 6, and a memory. The memory stores a computer program, and the processor 6 executes the computer program to implement the method described in Embodiment 1.

[0085] The modules are connected via an internal device bus, and the processor 6 uniformly schedules the working status of each module.

[0086] Vibration sensor 1 is placed close to the ear canal or auricular bone to collect bone conduction vibration signals.

[0087] The bone conduction vibration signal is not affected by environmental noise and can truly reflect the user's own breathing effort vibration, heartbeat and bone conduction snoring, serving as a reference signal for subsequent noise reduction processing.

[0088] Vibration sensor 1 operates continuously in low-power mode.

[0089] Microphone 2 is located at the headphone vent or the ear canal and is used to collect the breathing airflow sound and environmental noise in the ear canal as the main input signal for noise reduction processing.

[0090] Microphone 2 adopts an on-demand start-stop working mode, which is only activated when vibration sensor 1 detects a suspected abnormal breathing, and immediately turns off after classification and judgment are completed, effectively reducing overall power consumption.

[0091] Processor 6 is a low-power microcontroller unit (MCU) or digital signal processor 6 (DSP) for the headphone end, responsible for running embedded algorithms, including sensor driving, data preprocessing, adaptive noise reduction, feature extraction, classification and judgment, and intervention strategy execution.

[0092] The memory stores a computer program, and the processor 6 executes the computer program to implement the sleep breathing monitoring method of the present invention.

[0093] The inertial measurement unit 4 (IMU) is used to monitor the user's head posture, body movements, and wearing status.

[0094] The output data of the inertial measurement unit 4 is used to determine the user's sleeping state to trigger the monitoring process, detect changes in posture such as lying on one's side to dynamically adjust the signal weights of each sensor, and combine the dual-mode signal missing characteristics to assist in determining whether the device has been worn off.

[0095] PPG sensor 3 is used to collect at least one of the user's heart rate, blood oxygen saturation (SpO2) and heart rate variability (HRV) as auxiliary verification indicators to participate in classification and judgment, and to provide heart rate abnormality monitoring in the safety circuit breaker mechanism.

[0096] The PPG sensor 3 is located at the point where the earphone contacts the auricle or ear canal. It collects physiological characteristic signals through photoplethysmography. Its specific implementation is a common practice for those skilled in the art and will not be described in detail here.

[0097] The actuator includes at least one of a speaker 5 and a linear motor, and is used to output a corresponding intervention signal based on the classification result.

[0098] For obstructive sleep apnea, a specific frequency of sound waves or vibration signals are output to stimulate the tension of the throat muscles and dilate the airway; for central sleep apnea, progressively enhanced tactile pulses or auditory awakening sounds are output to stimulate the respiratory center.

[0099] The speaker 5 and the linear motor correspond to the auditory and tactile intervention channels, respectively, and can be used individually or in combination according to the intervention grading strategy.

[0100] In one specific example, the wearable device also includes a Bluetooth communication module 7, which is used to transmit the processed sleep data to a mobile APP or cloud server to realize the visualization of sleep reports, long-term trend analysis, and training and distribution of deep learning models.

[0101] In summary, the wearable device provided in this invention integrates both vibration sensors and a microphone, leveraging their complementary physical characteristics along the signal acquisition path to achieve dual-mode collaborative operation. The vibration sensor operates continuously in low-power mode as a trigger, while the microphone is activated on demand for secondary confirmation and then deactivated upon confirmation. This achieves a balance between monitoring accuracy and overall power consumption at the hardware level, resolving the contradiction in traditional solutions where a constantly on microphone leads to excessive power consumption or a degraded accuracy when the microphone is off. Furthermore, by integrating an inertial measurement unit, a PPG sensor, and an actuator, the device can output differentiated intervention signals for obstructive and central sleep apnea based on adaptive multimodal fusion judgment, achieving a complete closed loop from monitoring to intervention, significantly improving intervention success rate and user comfort.

[0102] Accordingly, other embodiments of this application may also provide a storage medium storing computer-executable instructions, which, when executed by a processor, implement the various method embodiments of this application. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information may be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0103] In summary, this invention discloses a sleep breathing monitoring method, a wearable device, and a storage medium. It utilizes vibration signals as reference signals for adaptive noise reduction of audio signals, eliminating environmental noise without requiring continuous recording of ambient sounds, thus effectively protecting user privacy. A two-dimensional judgment matrix based on dual-mode signals is used to differentiate between obstructive and central sleep apnea. Abnormal device wear is accurately identified by the simultaneous absence of dual-mode signals. A dual-mode intelligent start-stop strategy uses a low-power vibration sensor as a continuous trigger, waking the microphone only when an abnormality is suspected, achieving stable monitoring with low power consumption. Targeted intervention is highly effective, executing airway dilation stimulation and respiratory center arousal based on precise classification, combined with a graded intensity control strategy to improve intervention success rate while effectively protecting sleep structure.

[0104] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for monitoring sleep breathing, characterized in that, include: Vibration signals are collected as reference signals, and noise reduction processing is performed on the collected audio signals to obtain pure breathing airflow signals; Based on the amplitude of the vibration signal and the signal intensity of the pure respiratory airflow signal, the current respiratory state is classified and determined. When the amplitude of the vibration signal is higher than a first threshold and the signal intensity of the pure respiratory airflow signal is lower than a second threshold, it is determined to be obstructive sleep apnea. When the amplitude of the vibration signal is lower than the first threshold and the signal intensity of the pure respiratory airflow signal is lower than the second threshold, it is determined to be central sleep apnea.

2. The sleep breathing monitoring method according to claim 1, characterized in that, The noise reduction process employs an adaptive filtering algorithm or a deep learning mask network, using the vibration signal as a reference input and the audio signal as the main input, to extract the user's breathing components related to the vibration signal and remove environmental noise components unrelated to the vibration signal.

3. The sleep breathing monitoring method according to claim 1, characterized in that, The classification determination also includes: when the amplitude of the vibration signal and the signal strength of the pure breathing airflow signal are both lower than the third threshold, it is determined that the wearable device has fallen off.

4. The sleep breathing monitoring method according to claim 1, characterized in that, The vibration signal is continuously collected. When the amplitude of the vibration signal is lower than a preset percentage of the baseline value within a preset number of consecutive breathing cycles, the audio signal is collected, and the noise reduction processing and classification judgment are performed. After the classification determination is completed, the acquisition of the audio signal is stopped.

5. The sleep breathing monitoring method according to claim 4, characterized in that, Also includes: When the acquisition of the audio signal is triggered, and normal breathing airflow is detected after the noise reduction process, it is determined that the vibration signal is misjudged due to poor contact. The event is ignored and the weight of the vibration signal in the classification judgment is calibrated.

6. The sleep breathing monitoring method according to claim 1, characterized in that, Also includes: The system combines the collected motion data to determine the user's sleep state, and triggers the collection of the audio signal after confirming that the user has fallen asleep.

7. The sleep breathing monitoring method according to claim 1, characterized in that, This also includes implementing differentiated interventions based on the classification results: When obstructive sleep apnea is detected, a sound wave or vibration signal of a specific frequency is output; When central sleep apnea is diagnosed, a progressively stronger tactile pulse or a sound with a specific rhythm is output. Optionally, if the intervention fails after a preset number of consecutive attempts or if the heart rate exceeds a preset abnormal threshold, the intervention will be stopped and an external terminal alarm will be triggered.

8. The sleep breathing monitoring method according to claim 1, characterized in that, Also includes: Collect physiological characteristic signals and classify them based on the vibration signal, the pure respiratory airflow signal, and the physiological characteristic signals. Optionally, the weights of the vibration signal, the pure respiratory airflow signal, and the physiological characteristic signal in the classification determination can be dynamically adjusted by combining the collected posture data.

9. A wearable device, characterized in that, The device includes a vibration sensor, a microphone, a processor, and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement the method of any one of claims 1 to 8.

10. The wearable device according to claim 9, characterized in that, The vibration sensor is used to collect bone conduction vibration signals, and the microphone is used to collect breathing airflow sounds and environmental noise. Optionally, it also includes an inertial measurement unit for monitoring the user's head posture, body movements, and wearing status; Optionally, it also includes a PPG sensor for collecting at least one of the user's heart rate, blood oxygen saturation, and heart rate variability; Optionally, it may also include an actuator, which includes at least one of a speaker and a linear motor, for outputting an intervention signal.

11. A storage medium, characterized in that, The device contains a computer program that, when executed by a processor, implements the method of any one of claims 1 to 8.

Images