Respiratory event recognition method and device, storage medium and computer program product

Abstract

The invention discloses a respiratory event recognition method and device, a storage medium and a computer program product, and relates to the technical field of medical monitoring, the method comprises the following steps: collecting throat vibration data through a three-axis acceleration sensor of throat snore-ceasing equipment, and generating a vibration data sequence according to the throat vibration data; calculating an environmental noise baseline according to the vibration data sequence, and identifying a breathing emergency based on the environmental noise baseline; and calculating the current breathing amplitude according to the vibration data sequence, and calculating a breathing amplitude ratio based on the current breathing amplitude and a pre-constructed breathing cycle amplitude model so as to identify an apnea event. The throat vibration data is collected through the three-axis acceleration sensor, and the synergistic effect of the environmental noise baseline and the respiratory cycle amplitude model is combined, so that the recognition precision and recognition efficiency of the apnea event and the respiratory emergency are improved.

Description

Technical Field

[0001] This application relates to the field of medical monitoring technology, and in particular to respiratory event identification methods, devices, storage media, and computer program products. Background Technology

[0002] Accurate identification and timely intervention of respiratory emergencies such as snoring and talking during sleep, as well as sleep apnea events, have significant clinical and practical implications. Currently, intelligent pulse anti-snoring devices designed for sleep apnea primarily detect and capture respiratory emergencies such as snoring and talking through sound and vibration. However, because airflow weakens or stops during sleep apnea events, there is usually no obvious sound or vibration similar to snoring, making it impossible for sensors to capture effective identification signals. Consequently, these intelligent pulse anti-snoring devices cannot effectively identify and intervene in sleep apnea events.

[0003] Therefore, how to improve the accuracy and efficiency of the laryngeal anti-snoring device in recognizing sleep apnea events and respiratory emergencies has become a technical problem that this application urgently needs to solve.

[0004] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this application is to provide a respiratory event recognition method, device, storage medium, and computer program product, aiming to solve the technical problem of how to improve the recognition accuracy and efficiency of laryngeal anti-snoring devices for sleep apnea events and respiratory emergencies.

[0006] To achieve the above objectives, this application proposes a respiratory event recognition method, the method comprising: The laryngeal anti-snoring device collects laryngeal vibration data using a three-axis accelerometer and generates a vibration data sequence based on the laryngeal vibration data. Calculate the environmental noise baseline based on the vibration data sequence, and identify respiratory emergencies based on the environmental noise baseline; The current respiratory amplitude is calculated based on the vibration data sequence, and the respiratory amplitude ratio is calculated based on the current respiratory amplitude and the pre-built respiratory cycle amplitude model to identify apnea events.

[0007] In one embodiment, the step of acquiring laryngeal vibration data through the triaxial accelerometer of the laryngeal anti-snoring device and generating a vibration data sequence based on the laryngeal vibration data includes: Within a preset time period, the three-axis accelerometer of the laryngeal anti-snoring device collects laryngeal vibration data at a preset sampling frequency; The throat vibration data are arranged in chronological order to obtain a vibration data sequence.

[0008] In one embodiment, the step of calculating an environmental noise baseline based on the vibration data sequence and identifying respiratory emergencies based on the environmental noise baseline includes: Select a first data sequence from the vibration data sequence whose sequence length is greater than a first threshold, and calculate the first difference sequence of the first data sequence; The variance of the difference sequence is calculated based on the first difference sequence, and the variance of the difference sequence is used as the environmental noise baseline. At any point in time in the vibration data sequence, an instantaneous difference sequence is acquired forward, and the instantaneous variance of the instantaneous difference sequence is calculated. The presence of a respiratory emergency is determined based on the instantaneous variance and the environmental noise baseline. If the respiratory emergency occurs, the type of the respiratory emergency is identified by the spectral distribution obtained after performing a Fourier transform on the vibration data sequence.

[0009] In one embodiment, the step of identifying the type of respiratory emergency based on the spectral distribution obtained after performing a Fourier transform on the vibration data sequence includes: A second data sequence is generated based on the vibration data sequence, and a second difference sequence of the second data sequence is calculated; The second difference sequence is subjected to Fourier transform to obtain the spectral distribution, and the type of respiratory emergency is identified based on the spectral distribution.

[0010] In one embodiment, the step of calculating the current respiratory amplitude based on the vibration data sequence and calculating the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify apnea events further includes: Select a third data sequence from the vibration data sequence, the sequence length of which is multiple respiratory cycles; The median smoothing curve is obtained by performing a weighted average calculation on the third data sequence using a pre-set smoothing coefficient. Calculate the first overlap area between the median smoothing curve and the third data sequence, and calculate the model breathing amplitude based on the first overlap area; The respiratory amplitude model is updated by rolling forward the time series to obtain the respiratory cycle amplitude model.

[0011] In one embodiment, the step of calculating the current respiratory amplitude based on the vibration data sequence, and calculating the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify apnea events includes: Select a fourth data sequence from the vibration data sequence, the sequence length of which is a single respiratory cycle; Calculate the second overlap area between the fourth data sequence and the pre-generated median smoothed curve, and calculate the current respiratory amplitude based on the second overlap area; The respiratory amplitude ratio is calculated based on the current respiratory amplitude and the model respiratory amplitude in the pre-built respiratory cycle amplitude model; The breathing amplitude ratio is compared with a predefined apnea threshold, and apnea events are identified based on the comparison result.

[0012] In one embodiment, after the steps of calculating the current respiratory amplitude based on the vibration data sequence and calculating the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify apnea events, the method further includes: Calculate the number of overlaps between the pre-generated median smoothed curve and the third data sequence, and correct the smoothing coefficient in the respiratory cycle amplitude model based on the number of overlaps; If the respiratory emergency occurs, the respiratory cycle amplitude model is discarded, and the respiratory cycle amplitude model is reconstructed based on the vibration data sequence.

[0013] Furthermore, to achieve the above objectives, this application also proposes a respiratory event recognition device, which includes: The data acquisition module is used to acquire laryngeal vibration data through the triaxial accelerometer of the laryngeal antisnoring device, and generate a vibration data sequence based on the laryngeal vibration data; A respiratory emergency identification module is used to calculate an environmental noise baseline based on the vibration data sequence and to identify respiratory emergencies based on the environmental noise baseline. The sleep apnea event identification module is used to calculate the current breathing amplitude based on the vibration data sequence, and to calculate the breathing amplitude ratio based on the current breathing amplitude and a pre-built breathing cycle amplitude model to identify sleep apnea events.

[0014] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the breathing event recognition method described above.

[0015] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the breathing event recognition method described above.

[0016] One or more technical solutions proposed in this application have at least the following technical effects: The device collects laryngeal vibration data using a three-axis accelerometer and generates a vibration data sequence based on this data. An environmental noise baseline is calculated from the vibration data sequence, and respiratory emergencies are identified based on this baseline. The current respiratory amplitude is calculated from the vibration data sequence, and a ratio of the current respiratory amplitude to a pre-built respiratory cycle amplitude model is calculated to identify apnea events. First, by collecting laryngeal vibration data using a three-axis accelerometer, no additional sensors or computing power are required; data acquisition is achieved directly using the device's existing hardware, ensuring real-time data transmission while reducing hardware costs and data processing complexity, thus improving recognition efficiency. Second, calculating the environmental noise baseline and identifying respiratory emergencies based on it effectively distinguishes between normal noise and signals related to sudden breathing, reducing misjudgments caused by non-respiratory interference factors and improving the accuracy of identifying respiratory emergencies. Distinguishing between respiratory emergencies and apnea events also provides a foundation for accurate identification of apnea events. Furthermore, the ratio of the current respiratory amplitude to a pre-built respiratory cycle amplitude model is calculated to identify apnea events. This quantitative comparison enables accurate judgment of the breathing state, ensuring the accuracy of apnea event identification. In this application, laryngeal vibration data is collected by a triaxial accelerometer, and the combined effect of environmental noise baseline and respiratory cycle amplitude model is used to improve the accuracy and efficiency of identifying apnea events and respiratory emergencies. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the first embodiment of the respiratory event recognition method of this application; Figure 2 A schematic diagram of the respiratory curve provided in this application; Figure 3 This is a flowchart illustrating the second embodiment of the respiratory event recognition method of this application; Figure 4 This is a flowchart illustrating the third embodiment of the respiratory event recognition method of this application; Figure 5 This is a flowchart illustrating the fifth embodiment of the respiratory event recognition method of this application; Figure 6 This is a schematic diagram of the module structure of the respiratory event recognition device according to an embodiment of this application; Figure 7 This is a schematic diagram of the device structure of the hardware operating environment involved in the respiratory event recognition method in the embodiments of this application.

[0020] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] The main solution of this application embodiment is: to collect larynx vibration data through a triaxial accelerometer of a laryngeal anti-snoring device, and generate a vibration data sequence based on the larynx vibration data; to calculate an environmental noise baseline based on the vibration data sequence, and to identify respiratory emergencies based on the environmental noise baseline; to calculate the current respiratory amplitude based on the vibration data sequence, and to calculate the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify sleep apnea events.

[0024] This application's embodiments take into account the significant clinical and practical implications of accurately identifying and promptly intervening in sudden respiratory events such as snoring and talking during sleep, as well as sleep apnea events. Currently, intelligent pulse anti-snoring devices designed for sleep apnea primarily capture sudden respiratory events like snoring and talking through sound and vibration recognition. However, because airflow weakens or stops during sleep apnea events, there is usually no obvious sound or vibration similar to snoring, making it impossible to capture effective recognition signals through sensors. Therefore, intelligent pulse anti-snoring devices cannot effectively identify and intervene in sleep apnea events. While ventilators can identify sleep apnea events by monitoring changes in airflow and pressure, they rely on direct monitoring of airflow and pressure, requiring dedicated airflow sensors, pressure regulation modules, and tubing connections. Since intelligent anti-snoring devices are small, wearable devices that need to fit snugly against the skin of the throat, their structural design, size, and hardware configuration (such as sensor type and computing power) differ fundamentally from ventilators, making it impossible to integrate the airflow / pressure monitoring modules and related hardware required by ventilators. Therefore, the sleep apnea recognition technology of ventilators cannot be directly applied to intelligent anti-snoring devices for the throat, making it difficult for these devices to use the technology to solve the problem of sleep apnea event recognition.

[0025] Therefore, this application provides a solution that collects laryngeal vibration data using a triaxial accelerometer of the laryngeal anti-snoring device and generates a vibration data sequence based on the laryngeal vibration data; calculates an environmental noise baseline based on the vibration data sequence, and identifies respiratory emergencies based on the environmental noise baseline; calculates the current respiratory amplitude based on the vibration data sequence, and calculates the ratio of the current respiratory amplitude to a pre-built respiratory cycle amplitude model to identify apnea events. First, by collecting laryngeal vibration data using a triaxial accelerometer, no additional sensors or computing power are needed; data acquisition is achieved directly using the existing hardware of the device, ensuring real-time data transmission while reducing hardware costs and data processing complexity, and improving recognition efficiency. Second, calculating the environmental noise baseline and identifying respiratory emergencies based on it effectively distinguishes between normal noise and signals related to sudden breathing, reducing misjudgments caused by non-respiratory interference factors, improving the accuracy of identifying respiratory emergencies, and differentiating between respiratory emergencies and apnea events, thus providing a foundation for accurate identification of apnea events. Furthermore, by calculating the ratio of the current respiratory amplitude to a pre-built respiratory cycle amplitude model to identify apnea events, accurate judgment of the breathing state is achieved through quantitative comparison, ensuring the accuracy of apnea event identification. In this application, laryngeal vibration data is collected by a triaxial accelerometer, and the combined effect of environmental noise baseline and respiratory cycle amplitude model is used to improve the accuracy and efficiency of identifying apnea events and respiratory emergencies.

[0026] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or respiratory event recognition device capable of performing the above functions. The following description uses a respiratory event recognition device as an example to illustrate this embodiment and the subsequent embodiments.

[0027] Based on this, this application provides a respiratory event recognition method applied to a laryngeal anti-snoring device, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the respiratory event recognition method of this application.

[0028] In this embodiment, the respiratory event recognition method includes steps S10 to S30: Step S10: Collect throat vibration data using the triaxial accelerometer of the throat anti-snoring device, and generate a vibration data sequence based on the throat vibration data; A laryngeal anti-snoring device refers to a low-power wearable electronic device worn on the user's throat, equipped with a microprocessor and a three-axis accelerometer. Its core task is to perform real-time judgment and intervention on respiratory events in the absence of airflow / pressure channels. These respiratory events include sudden respiratory events such as talking and snoring, as well as sleep apnea events.

[0029] A triaxial accelerometer is a microelectromechanical system (MEMS) accelerometer that integrates three mutually orthogonal sensing axes (X, Y, and Z) within the same package and can synchronously output three-axis instantaneous acceleration values.

[0030] Laryngeal vibration data refers to the mechanical vibrations generated by the coupling of respiratory airflow, soft palate vibration, vocal cord opening and closing, and chest wall movement to the soft tissues of the neck, which are represented as acceleration signals at the triaxial accelerometer.

[0031] Vibration data sequence refers to a discrete time sequence formed by performing analog-to-digital conversion on the original triaxial signals at a fixed sampling frequency and then writing them sequentially into a circular buffer with timestamps. Each element in the sequence usually contains the instantaneous amplitude of triaxial acceleration or the amplitude after vector synthesis into a scalar.

[0032] Additionally, it should be noted that the purpose of data acquisition is to provide high signal-to-noise ratio and high temporal resolution raw input for subsequent respiratory event recognition. By relying solely on laryngeal vibrations for respiratory monitoring, the airflow channels and masks required by traditional ventilators can be avoided, significantly lowering the compliance threshold. Furthermore, because the laryngeal tissue is closest to the airway, the vibration signals generated by breathing and snoring experience minimal attenuation; therefore, selecting laryngeal data acquisition significantly increases the proportion of respiratory-related signal energy, reducing the false positive rate of subsequent algorithms. Simultaneously, three-axis data fusion can offset the changes in gravitational projection caused by slight head and neck rotations, ensuring sequence stability.

[0033] It should also be noted that the acquisition process adopts a continuous rolling caching strategy to ensure that the latest 5–10s data is available in real time. At the same time, the built-in anti-aliasing low-pass filter suppresses high-frequency noise above 20Hz, laying the signal quality foundation for subsequent differential, Fourier transform and other operations.

[0034] Step S20: Calculate the environmental noise baseline based on the vibration data sequence, and identify respiratory emergencies based on the environmental noise baseline; The environmental noise baseline refers to the acceleration variance statistic contributed by respiratory coupling vibration and minor external mechanical disturbances in a resting, awake or stable sleep state without significant body movement, talking, or loud snoring. Its value reflects the inherent fluctuation level of the mechanical environment in which the system is currently located.

[0035] A respiratory emergency refers to a short-term event that occurs suddenly during sleep, such as snoring, talking, coughing, or turning over, causing the instantaneous energy to be significantly higher than the ambient noise baseline.

[0036] The overall approach to calculating the environmental noise baseline and identifying respiratory emergencies is as follows: First, long-window difference variance calculation is used on the vibration data sequence to lock the background noise average. Then, short-window difference variance is used to capture instantaneous energy mutations. When the mutation factor exceeds an empirical threshold, it is marked as a suspected emergency. In one possible implementation, the difference variance calculation can be replaced with second-order difference to further suppress drift.

[0037] refer to Figure 2 , Figure 2 A schematic diagram of the respiratory curve provided in this application. Figure 2 As shown, section A (yellow) represents data on snoring during sleep; section B (brown) represents data on normal breathing; section C (red) represents data on hypopnea and apnea; and section D (green) represents data on snoring and normal breathing after breathing has resumed. Figure 2 It can be clearly shown that the vibration data generated at different breathing states will also be different, so different breathing events can be identified through vibration data sequences.

[0038] Step S30: Calculate the current breathing amplitude based on the vibration data sequence, and calculate the breathing amplitude ratio based on the current breathing amplitude and the pre-built respiratory cycle amplitude model to identify apnea events.

[0039] The current respiratory amplitude refers to the integral of the absolute difference between the measured vibration data sequence and the ideal smooth template curve within a single respiratory cycle (approximately 4–6 seconds), and its magnitude is proportional to the intensity of thoracic expansion / contraction.

[0040] The respiratory cycle amplitude model refers to a template respiratory amplitude obtained from statistics of multiple past respiratory cycles. This template is continuously updated through a sliding weighted average to characterize the normal respiratory intensity of an individual in the current time period.

[0041] The breathing amplitude ratio refers to the dimensionless ratio of the current breathing amplitude to the model breathing amplitude. When this ratio is lower than a preset threshold, it indicates that the breathing airflow has decreased to varying degrees.

[0042] An apnea event is defined in medicine as a state in which the airflow during breathing through the mouth and nose decreases by ≥90% and lasts for ≥10 seconds. This application uses the amplitude ratio for equivalent discrimination.

[0043] Additionally, it should be noted that identifying apnea events by using the ratio of current breathing amplitude to model amplitude, rather than the absolute amplitude, can eliminate individual differences in absolute values ​​caused by different weights, neck circumferences, and sensor fit.

[0044] This embodiment provides a respiratory event recognition method. Based on a triaxial accelerometer to collect laryngeal vibration data, it eliminates the need for additional sensors or computing power, directly utilizing existing hardware to acquire data. This ensures real-time data transmission while reducing hardware costs and data processing complexity, thus improving recognition efficiency. Secondly, by calculating an environmental noise baseline and using it to identify respiratory emergencies, it effectively distinguishes between normal noise and signals related to sudden breathing, reducing misjudgments caused by non-respiratory interference factors and improving the accuracy of respiratory emergencies recognition. It also differentiates between respiratory emergencies and sleep apnea events, providing a foundation for accurate sleep apnea event recognition. Furthermore, by calculating the ratio between the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify sleep apnea events, it achieves accurate judgment of respiratory status through quantitative comparison, ensuring the accuracy of sleep apnea event recognition. In this application, the triaxial accelerometer collects laryngeal vibration data, and the combined effect of the environmental noise baseline and the respiratory cycle amplitude model improves the accuracy and efficiency of sleep apnea and respiratory emergencies recognition.

[0045] In one feasible implementation, step S10 may include steps S11-S12: Step S11: Within a preset time period, the laryngeal vibration data is collected by the triaxial accelerometer of the laryngeal antisnoring device at a preset sampling frequency. The preset time period refers to the continuous collection interval set according to the shortest effective duration of clinical sleep monitoring. The length is usually no less than 5 hours to ensure that it covers the user's complete physiological cycle from falling asleep to deep sleep and then waking up in the morning.

[0046] The preset sampling frequency refers to the fixed sampling rate written to the sensor register when the throat anti-snoring device is powered on and initialized. It is generally set to 400Hz, which corresponds to outputting 400 acceleration sample values ​​per second. This frequency not only meets the Nyquist requirements of the 0.1 to 0.5Hz component of the respiratory signal, but also takes into account the balance between power consumption and storage.

[0047] Additionally, it should be noted that the purpose of continuously collecting laryngeal vibration data within a preset time period is to capture the complete dynamics of the breathing curve throughout the night, providing a high temporal resolution data source for subsequent environmental noise baseline estimation, respiratory event identification, and apnea determination. By fixing the sensor to the larynx rather than the chest, abdomen, or wrist, it can be placed as close as possible to the airway vibration source, reducing the attenuation of the signal by fat and clothing, thereby increasing the energy proportion of breathing-related signals. At the same time, the 400Hz sampling rate can retain most of the physiological and snoring characteristics within 50Hz in the frequency domain, providing sufficient bandwidth for subsequent FFT spectrum analysis.

[0048] Step S12: Arrange the throat vibration data in chronological order to obtain a vibration data sequence.

[0049] Arranging laryngeal vibration data in chronological order means that the respiratory event recognition device writes the laryngeal vibration data collected by the triaxial accelerometer into a linear buffer or linked list in the order of its sampling time, forming a continuous, time-reversed, and packet-free discrete time series.

[0050] This discrete-time series is stored in memory at intervals a0, a... 1,...， a n This is represented as n = sampling frequency × time period. Each element a... i It contains the acceleration values ​​of the X, Y, and Z axes at the same moment, thus obtaining a vibration data sequence with strictly monotonically increasing timestamps and time intervals that are always equal to 1 / sampling frequency.

[0051] Based on the first embodiment of this application, a second embodiment of this application is proposed. In the second embodiment of this application, content that is the same as or similar to that in the first embodiment described above can be referred to the above description and will not be repeated hereafter.

[0052] Based on this, please refer to Figure 3 , Figure 3 This is a schematic diagram of the second embodiment of this application. Figure 3 As shown, step S20, which calculates the environmental noise baseline based on the vibration data sequence and identifies respiratory emergencies based on the environmental noise baseline, includes steps S21 to S25: Step S21: Select a first data sequence with a sequence length greater than a first threshold from the vibration data sequence, and calculate the first difference sequence of the first data sequence; The first threshold refers to the minimum sample length set to ensure the statistical reliability of environmental noise. Its value is equal to the number of sampling points corresponding to 5 minutes, which is 120,000 points at a sampling rate of 400Hz. For example, a data sequence A with a duration of 5 minutes is taken from the vibration data sequence. [0，120000] This data sequence keeps rolling forward over time.

[0053] The first data sequence refers to a continuous sub-segment, denoted as A, that starts from the beginning of the vibration data sequence and has a length greater than or equal to the first threshold. [0，m-1] , where m≥120,000.

[0054] The first difference sequence refers to the data sequence obtained after performing difference calculations on the first data sequence. The specific calculation formula is as follows: B [i] =A [i] -A [i-1] (i=1…m-1) It should be noted that B [0] Set to 0 to eliminate sensor zero drift and slow body position drift, while retaining high-frequency micro-vibration information such as breathing and heartbeat.

[0055] Additionally, it should be noted that the overall purpose of selecting the first data sequence greater than the first threshold and calculating the first difference sequence is to provide a statistically sufficiently long "stable segment" for the subsequent environmental noise baseline, avoiding baseline fluctuations due to excessively short samples. The difference operation can convert the low-frequency drift of the original acceleration into a near-zero mean process of the difference sequence, so that the variance can truly reflect the energy of random noise. The sliding window method is used to scroll forward once every 10 seconds to ensure that the baseline is updated slowly over time, adapting to changes in sensor fit or temperature throughout the night.

[0056] In one possible implementation, the first threshold can be dynamically adjusted according to the sampling rate, and when the sampling rate drops to 200Hz, the threshold is changed to 60,000 points accordingly.

[0057] Step S22: Calculate the variance of the difference sequence based on the first difference sequence, and use the variance of the difference sequence as the environmental noise baseline; The variance of a difference sequence refers to the variance of the first difference sequence B. The specific calculation formula is as follows: Vol = variation(B0, B1, ..., B m-1 ) The variance of the difference sequence of the first variance sequence is calculated by calculating the variance of the first variance sequence. The variance of the difference sequence is directly used as the environmental noise baseline, and the golden noise memory is periodically updated forward to track the minute environmental changes throughout the night.

[0058] Step S23: Acquire an instantaneous difference sequence forward from any time point in the vibration data sequence, and calculate the instantaneous variance of the instantaneous difference sequence; Any point in time refers to any timestamp t in the vibration data sequence, covering the entire vibration data sequence. Acquiring an instantaneous difference sequence forward from any point in time means extracting a short segment of 1 second in length from any timestamp t in the vibration data sequence and using this short segment as the instantaneous difference sequence.

[0059] The instantaneous variance of the instantaneous difference sequence is calculated using the following formula: C [k] =A [t-399+k] -A [t-400+k] (k=0…399) Act = variation(C0, ..., C k ) Among them, C [k]Let represent the instantaneous difference sequence, Act represent the instantaneous variance, A represent the original vibration data sequence, and t represent any timestamp.

[0060] In one possible implementation, the window length can be extended to 2 seconds based on the duration of snoring. If the sampling frequency is 400 Hz, then the number of data points sampled within this window length is 800. Extending the window length improves the spectral resolution. It is understood that the sequence length of the short segment truncated forward from any timestamp t in the vibration data sequence can vary depending on the specific circumstances. In this embodiment, 1 second is chosen as the sequence length only for better illustration.

[0061] Step S24: Determine whether a respiratory emergency exists based on the instantaneous variance and the environmental noise baseline; Determining the presence of a respiratory emergency based on instantaneous variance and environmental noise baseline involves calculating the instantaneous variance and environmental noise baseline using the following formula: Rate_act=Act / Vol Here, Rate_act represents the ratio curve of instantaneous variance to the environmental noise baseline as a function of time, Act represents instantaneous variance, and Vol represents the environmental noise baseline.

[0062] The Rate_act is compared with an empirical threshold. For example, the empirical threshold is 3. If Rate_act≥3, a respiratory emergency is determined to have occurred. The empirical threshold can be reset according to individual differences caused by different wearing tightness, weight, and mattress firmness.

[0063] Step S25: If the respiratory emergency exists, the type of the respiratory emergency is identified based on the spectral distribution obtained after performing a Fourier transform on the vibration data sequence.

[0064] Fourier transform refers to the process of resampling the vibration data sequence within the segment of a respiratory emergency in the presence of such an event, obtaining a new data sequence, performing a 512-point Fast Fourier Transform on this new data sequence to obtain the spectral distribution, and identifying the type of respiratory emergency based on the spectral distribution. Types of respiratory emergencies include: talking, snoring, turning over, etc.

[0065] In this embodiment, by automatically establishing and updating the environmental noise baseline throughout the night's sleep, and by using spectral characteristics to accurately distinguish snoring from interference such as talking and body movement in the event of a respiratory emergency, the system can quickly identify and intervene in only true snoring or hypopnea, significantly reducing the false alarm rate and power consumption.

[0066] In one feasible implementation, step S25 may include steps S251-S252: Step S251: Generate a second data sequence based on the vibration data sequence, and calculate the second difference sequence of the second data sequence; The second data sequence refers to a short segment consisting of 512 data points, 256 points forward and 256 points backward from the original vibration data sequence, centered on the starting point of the respiratory emergency. This segment covers the entire process of laryngeal vibration and is used to preserve the complete transient characteristics of the emergency. The second difference sequence refers to the new sequence obtained by performing a first-order backward difference on the second data sequence.

[0067] For example, after determining that a respiratory emergency has occurred at time point t, data d is taken for two seconds before and after point t. [0,…,799] , for d [0,…,799] After performing the difference calculation, we get: e0=0; e1=d1-d0;….e 799 =d 799 -d 798 The initial second difference sequence is obtained. Since Fourier transform typically takes a power of 2 of the data points, 256 points are taken before and after the midpoint value of the initial second difference sequence, for a total of 512 points, to obtain the second difference sequence.

[0068] In one possible implementation, the length of the second data sequence can be extended to 1024 points to improve spectral resolution; in another possible implementation, the difference can be replaced with a second-order difference to further suppress low-frequency drift.

[0069] Step S252: Perform a Fourier transform on the second difference sequence to obtain a spectral distribution, and identify the type of respiratory emergency based on the spectral distribution.

[0070] The Fourier transform refers to performing a fast Fourier transform (RFFT) on the second difference sequence, outputting 256 complex spectrum bins with a frequency resolution of Δf = 400Hz / 512 ≈ 0.78Hz.

[0071] Spectral distribution refers to the energy density spectrum P[k] obtained by taking the modulus of the complex spectrum and squaring it, k=0,…,255, covering 0~200Hz, used to quantify the distribution characteristics of energy in each frequency band of sudden events.

[0072] Respiratory emergencies can be categorized into four types: snoring, talking, coughing, and large movements. Snoring has a main frequency concentrated in the range of 30-120Hz with a narrow peak, talking has an energy distribution in the range of 80-250Hz, coughing has a broadband burst in the range of 40-150Hz, and large movements have an energy peak below 20Hz with an extremely wide bandwidth.

[0073] Additionally, it should be noted that by identifying event types through spectral distribution, it is possible to accurately distinguish between speech and body movements that do not require intervention and snoring that requires immediate intervention, thus avoiding accidental triggering of laryngeal electrical stimulation.

[0074] Specifically, first extract the main frequency F_max=Max(P[k]), then calculate the proportion of energy in the 30~120Hz range to the total energy R_snore. If F_max∈[30,120] and R_snore>60%, it is judged as snoring; if F_max<20Hz and the proportion of energy in the 0~40Hz range is >70%, it is judged as a large movement; in other cases, speaking and coughing are further subdivided according to the threshold chain.

[0075] In this embodiment, by identifying the types of respiratory emergencies, various feasible measures can be taken to intervene and improve the user's sleep quality.

[0076] Based on the first and / or second embodiments of this application, a third embodiment of this application is proposed. In the third embodiment of this application, content that is the same as or similar to the first and / or second embodiments described above can be referred to the above description and will not be repeated hereafter.

[0077] Based on this, please refer to Figure 4 , Figure 4 This is a schematic diagram of the process of the third embodiment of this application. Figure 4 As shown, in this embodiment, before step S30, which calculates the current respiratory amplitude based on the vibration data sequence and calculates the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify apnea events, steps S01 to S04 are also included: Step S01: Select a third data sequence from the vibration data sequence, with a sequence length of multiple respiratory cycles; A respiratory cycle refers to the time required for a user to complete one inhalation and exhalation while asleep. The normal range for adults is about 2 to 5 seconds, corresponding to 800 to 2000 sampling points at 400 Hz. Multiple respiratory cycles generally refer to 3 to 5 complete cycles, with a sequence length of about 2400 to 10000 points. The default value is 15 seconds as the rolling window length.

[0078] The third data sequence refers to the vibration data segment of fixed duration 15 seconds extracted from the current sampling time t, denoted as aa. [0,5999] This sub-segment is used to subsequently establish or update personalized respiratory intensity benchmarks. Its selection principle is to avoid known large movement markers and ensure data stability.

[0079] Additionally, it should be noted that the purpose of selecting a third data sequence with a length of multiple respiratory cycles is to provide a sufficiently long "template segment" for subsequent smoothing and area calculation, so that the model's respiratory amplitude can reflect the user's true average respiratory intensity. By fixing the window length to 15 seconds, the system can calculate the average number of cycles k within 3 to 5 cycles, providing a basis for the adaptive correction of the smoothing coefficient α (α=1 / k). If the truncated length is less than 3 cycles, the system will extend it forward to at least one respiratory cycle to ensure statistical validity.

[0080] In one possible implementation, the window length of multiple respiratory cycles can be dynamically adjusted based on the real-time detected instantaneous respiratory rate to ensure that five complete cycles are always captured; in another possible implementation, the third data sequence can be captured starting from the "first stable breathing zone after falling asleep" to improve the accuracy of the initial model.

[0081] Step S02: Use a pre-set smoothing coefficient to perform a weighted average calculation on the third data sequence to obtain a median smoothing curve; The pre-set smoothing coefficient refers to the α value assigned during system initialization, which controls the memory length of the exponentially weighted average. The larger the α value, the closer the curve is to the original data, and the smaller the value, the smoother the curve. The weighted average calculation refers to performing first-order exponential smoothing on the third data sequence.

[0082] For example, a third data sequence of multiple respiratory cycles aa with a sampling frequency of 400Hz and a default 15s. [0,5999] For example, the median smoothing curve bb [0,5999] The specific calculation process is as follows: bb0 = aa0 × α; bb1 = aa0 × (1 - α) + aa1 × α; bb2 = aa1 × (1 - α) + aa2 × α; … bb 5999 =aa 5998 ×(1-α)+aa 5999 ×α Additionally, it should be noted that the purpose of obtaining the median smoothing curve through exponential weighted averaging is to provide an idealized respiratory baseline for subsequent overlap area calculations, ensuring that the area value reflects only the true respiratory amplitude rather than random fluctuations. The smoothing process is completed in a single pass in the time domain, with a computational complexity of O(n) and memory usage only one more floating-point buffer than the input. The smoothing coefficient α is then adjusted in real time based on the number of zero crossovers k between the third data sequence and the median smoothing curve, ensuring that the smoothing window length always equals the average respiratory cycle, thus improving the model's adaptability.

[0083] Step S03: Calculate the first overlap area between the median smoothing curve and the third data sequence, and calculate the model breathing amplitude based on the first overlap area; The first overlap area refers to the integral of the absolute difference between the third data sequence and its corresponding median smoothed curve at the same sampling point, using multiple respiratory cycles of the third data sequence aa with a sampling frequency of 400Hz and a default 15s. [0,5999] Median smoothing curve bb [0,5999] For example, its first overlapping area M is:

[0084] in, This represents the raw vibration data of the nth sampling point (6000 data points in 15 seconds, n=0 to 5999). This represents the median smoothed curve data for the nth sampling point (the trend curve calculated using the smoothing coefficient α).

[0085] Dividing the first overlapping area by the length of the third data sequence yields the model's respiratory amplitude. For example, if the length of the third data sequence is 15 seconds, the formula for calculating the model's respiratory amplitude is: M / 15.

[0086] Step S04: Update the respiratory amplitude of the model by rolling forward with the time series to obtain the respiratory cycle amplitude model.

[0087] The forward rolling update of the time series refers to sliding the window's start and end points forward by one point for each new sampling point, forming a new third data sequence. This process of smoothing, area recognition, and normalization is repeated to obtain the updated model respiratory amplitude M. new , replacing the old value M.

[0088] Additionally, it should be noted that the rolling update mechanism enables the model to adaptively fit the physiological fluctuations throughout the night, avoiding misjudgments caused by a fixed template; when large movements such as turning over or getting out of bed are detected (Rate_act>threshold), the model is immediately reset, and 15 seconds of data are extracted again to establish a new baseline to prevent model collapse; the update process adopts a double-buffered ping-pong structure, with computation and data acquisition in parallel to ensure real-time performance.

[0089] In this embodiment, laryngeal vibration data of multiple respiratory cycles are continuously captured throughout the night's sleep. An adaptive smoothing coefficient is used to generate a median smoothing curve in real time, and a personalized respiratory cycle amplitude model is continuously updated using an overlap area normalization method. This accurately tracks the slow changes in the user's breathing intensity and resets immediately upon encountering large movements, ensuring that the model always fits the current physiological state. This provides a highly reliable and timely benchmark for subsequent medical threshold judgment of hypoventilation and apnea events, significantly improving recognition accuracy and reducing false alarms.

[0090] Based on the above embodiments of this application, a fourth embodiment of this application is proposed. In this fourth embodiment, content that is the same as or similar to that in the above embodiments can be referred to the above description, and will not be repeated hereafter.

[0091] In this embodiment, step S30, which calculates the current respiratory amplitude based on the vibration data sequence and calculates the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify apnea events, may include steps S31 to S34: Step S31: Select a fourth data sequence from the vibration data sequence with a sequence length of a single respiratory cycle; A single respiratory cycle refers to the average time required for a user to complete one inhalation and exhalation while asleep. The normal range for adults is approximately 2-5 seconds, corresponding to 800-2000 data points at a 400Hz sampling rate. The fourth data sequence refers to a short segment of 5 seconds preceding the current sampling point t, denoted as aa. [4000,5999] The sequence length is 2000 data points, which can completely cover a respiratory cycle and is used to reflect the instantaneous intensity of the most recent breath in real time. The truncation operation slides once every 25 data points in the scroll window to ensure the real-time recognition of subsequent apnea.

[0092] Additionally, it should be noted that the purpose of selecting the fourth data sequence, with a sequence length of one respiratory cycle, is to provide a benchmark for comparison of the current respiratory amplitude with the same dimensions and physical meaning as the model's respiratory amplitude. Step S32: Calculate the second overlap area between the fourth data sequence and the pre-generated median smoothing curve, and calculate the current respiratory amplitude based on the second overlap area; The second overlap area refers to the integral of the absolute difference between the fourth data sequence and its corresponding median smoothing curve at the same sampling point, with a sampling frequency of 400Hz and a default 5-second single respiratory cycle for the fourth data sequence aa. [4000,5999] Median smoothing curve bb [0,5999] For example, its first overlapping area S is:

[0093] in, This represents the raw vibration data of the nth sampling point (2000 data points over 5 seconds, n=4000 to 5999). This represents the median smoothed curve data for the nth sampling point (the trend curve calculated using the smoothing coefficient α).

[0094] The current respiratory amplitude can be obtained by dividing the second overlapping area by the sequence length of the fourth data sequence. For example, if the sequence length of the fourth data sequence is 5 seconds, then the current respiratory amplitude is calculated as: S / 5.

[0095] Step S33: Calculate the respiratory amplitude ratio based on the current respiratory amplitude and the model respiratory amplitude in the pre-built respiratory cycle amplitude model; The respiratory amplitude ratio refers to the ratio of the current respiratory amplitude to the model's respiratory amplitude, and its specific calculation formula is as follows: Rate_fudu (Ratio of Respiratory Amplitude) = Current Respiratory Amplitude / Model Respiratory Amplitude Step S34: Compare the breathing amplitude ratio with a predefined apnea threshold, and identify apnea events based on the comparison result.

[0096] The predefined apnea thresholds are determined in the firmware based on medical standards and clinical big data: hypoventilation threshold 0.5, apnea threshold 0.33, and recovery threshold 0.8. Apnea events refer to a pathological state where Rate_fudu is continuously below 0.33 for ≥10 seconds, requiring intervention. The comparison logic is: if Rate_fudu < 0.33, enter the apnea state; if > 0.8, exit the apnea state; 0.33~0.5 is hypoventilation, and 0.5~0.8 is normal.

[0097] Additionally, it should be noted that using a fixed but medically validated threshold ensures consistency and comparability across different individuals; the respiratory event recognition device has a built-in 10-second countdown timer to prevent misjudgment due to short-term fluctuations; once apnea is confirmed, it immediately drives an electrical pulse in the larynx and records the start time, duration, and minimum ratio to generate a sleep apnea event report for the next day.

[0098] In this embodiment, laryngeal vibration data is captured in real time using a single respiratory cycle as a window, and the overlap area is calculated with the continuously updated median smoothing curve to obtain the current respiratory amplitude that is highly synchronized with the individual's physiological state. Then, it is compared with the model respiratory amplitude to obtain the respiratory amplitude ratio, and then judged by a medical threshold to accurately identify the apnea event and immediately trigger intervention. At the same time, it adaptively eliminates individual differences caused by obesity, body position or sensor fit, so as to achieve low latency, low false alarm, high consistency apnea monitoring and immediate improvement.

[0099] Based on the above embodiments of this application, a fifth embodiment of this application is proposed. In this fifth embodiment, content that is the same as or similar to that in the above embodiments can be referred to the above description, and will not be repeated hereafter.

[0100] Based on this, please refer to Figure 5 , Figure 5This is the schematic diagram of the process of the fifth embodiment of the present application. As Figure 5 shown, in this embodiment, after step S30 of calculating the current breathing amplitude according to the vibration data sequence and calculating the breathing amplitude ratio based on the current breathing amplitude and the pre-constructed breathing cycle amplitude model to identify apnea events, steps S40 to S50 are further included: Step S40, calculating the number of overlaps between the pre-generated median smoothing curve and the third data sequence, and correcting the smoothing coefficient in the breathing cycle amplitude model according to the number of overlaps; The number of overlaps refers to the zero-crossing count of the mutual crossing between the median smoothing curve and the third data sequence within the window. The number of overlaps directly reflects the true number of breathing cycles k within the window. The smoothing coefficient α is corrected through the following formula: α = 1 / k where, the normal range of the k value is 2 < k < 8, and the α value is continuously corrected over time, which can more accurately calculate a single breathing cycle. At the same time, in actual applications, the 15-second sampling model time and the 5-second single breathing time can be corrected to the real 5 breathing cycles and 1 breathing cycle. Thus, the algorithm is more accurate in actual applications.

[0101] Step S50, if there is the breathing emergency event, discarding the breathing cycle amplitude model and reconstructing the breathing cycle amplitude model according to the vibration data sequence.

[0102] In the case of a breathing emergency event with large movements, the original breathing cycle amplitude model can no longer represent the current breathing level. If it continues to be used, it will lead to misjudgment of apnea. By actively discarding the breathing cycle amplitude model and re-establishing it, it is ensured that the subsequent pause recognition continues to be accurate.

[0103] In this embodiment, by adaptively modifying the smoothing coefficient and reconstructing the breathing cycle amplitude model, it is ensured that the breathing cycle amplitude model is always synchronized with the user's real-time breathing frequency and wearing state, eliminating both the over-smoothing or under-smoothing errors caused by fixed parameters, and being able to re-establish an effective benchmark after large movements such as turning over and getting up, thereby significantly reducing the misjudgment of apnea caused by model mismatch and realizing all-night high-robust and high-precision non-invasive breathing monitoring and instant intervention.

[0104] The present application also provides a breathing event recognition device. Please refer to Figure 6 , the breathing event recognition device includes: A data acquisition module 10, configured to collect throat vibration data through the triaxial acceleration sensor of the throat snore stopper and generate a vibration data sequence according to the throat vibration data; The respiratory emergency identification module 20 is used to calculate the environmental noise baseline based on the vibration data sequence and identify respiratory emergencies based on the environmental noise baseline. The apnea event identification module 30 is used to calculate the current breathing amplitude based on the vibration data sequence, and to calculate the breathing amplitude ratio based on the current breathing amplitude and the pre-built breathing cycle amplitude model to identify apnea events.

[0105] The respiratory event recognition device provided in this application, employing the respiratory event recognition method in the above embodiments, can solve the technical problem of respiratory event recognition. Compared with the prior art, the beneficial effects of the respiratory event recognition device provided in this application are the same as those of the respiratory event recognition method provided in the above embodiments, and other technical features in the respiratory event recognition device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0106] This application provides a respiratory event recognition device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the respiratory event recognition method in the above embodiment 1.

[0107] The following is for reference. Figure 7 The diagram illustrates a structural schematic suitable for implementing a respiratory event recognition device according to embodiments of this application. The respiratory event recognition device in embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 7 The respiratory event recognition device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0108] like Figure 7As shown, the respiratory event recognition device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the respiratory event recognition device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the respiratory event recognition device to communicate wirelessly or wiredly with other devices to exchange data. While the figures show respiratory event recognition devices with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0109] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0110] The respiratory event recognition device provided in this application, employing the respiratory event recognition method in the above embodiments, can solve the technical problem of respiratory event recognition. Compared with the prior art, the beneficial effects of the respiratory event recognition device provided in this application are the same as those of the respiratory event recognition method provided in the above embodiments, and other technical features in this respiratory event recognition device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0111] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0112] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0113] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the respiratory event recognition method in the above embodiments.

[0114] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0115] The aforementioned computer-readable storage medium may be included in the respiratory event recognition device; or it may exist independently and not assembled into the respiratory event recognition device.

[0116] The aforementioned computer-readable storage medium carries one or more programs that, when executed by the respiratory event recognition device, cause the respiratory event recognition device to: acquire laryngeal vibration data through the triaxial accelerometer of the laryngeal anti-snoring device and generate a vibration data sequence based on the laryngeal vibration data; calculate an environmental noise baseline based on the vibration data sequence and identify a respiratory emergency based on the environmental noise baseline; calculate the current respiratory amplitude based on the vibration data sequence and calculate the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify a sleep apnea event.

[0117] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0118] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0119] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0120] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described respiratory event recognition method, thereby solving the technical problem of respiratory event recognition. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the respiratory event recognition method provided in the above embodiments, and will not be repeated here.

[0121] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the breathing event recognition method described above.

[0122] The computer program product provided in this application can solve the technical problem of respiratory event recognition. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the respiratory event recognition method provided in the above embodiments, and will not be repeated here.

[0123] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for identifying respiratory events, characterized in that, The respiratory event recognition method, applied to a laryngeal anti-snoring device, includes: The laryngeal anti-snoring device collects laryngeal vibration data using a three-axis accelerometer and generates a vibration data sequence based on the laryngeal vibration data. Calculate the environmental noise baseline based on the vibration data sequence, and identify respiratory emergencies based on the environmental noise baseline; The current respiratory amplitude is calculated based on the vibration data sequence, and the respiratory amplitude ratio is calculated based on the current respiratory amplitude and the pre-built respiratory cycle amplitude model to identify apnea events.

2. The respiratory event recognition method as described in claim 1, characterized in that, The step of acquiring laryngeal vibration data through the triaxial accelerometer of the laryngeal anti-snoring device and generating a vibration data sequence based on the laryngeal vibration data includes: Within a preset time period, the three-axis accelerometer of the laryngeal anti-snoring device collects laryngeal vibration data at a preset sampling frequency; The throat vibration data are arranged in chronological order to obtain a vibration data sequence.

3. The respiratory event recognition method as described in claim 1, characterized in that, The step of calculating an environmental noise baseline based on the vibration data sequence and identifying respiratory emergencies based on the environmental noise baseline includes: Select a first data sequence from the vibration data sequence whose sequence length is greater than a first threshold, and calculate the first difference sequence of the first data sequence; The variance of the difference sequence is calculated based on the first difference sequence, and the variance of the difference sequence is used as the environmental noise baseline. At any point in time in the vibration data sequence, an instantaneous difference sequence is acquired forward, and the instantaneous variance of the instantaneous difference sequence is calculated. The presence of a respiratory emergency is determined based on the instantaneous variance and the environmental noise baseline. If the respiratory emergency occurs, the type of the respiratory emergency is identified by the spectral distribution obtained after performing a Fourier transform on the vibration data sequence.

4. The respiratory event recognition method as described in claim 3, characterized in that, The step of identifying the type of respiratory emergency based on the spectral distribution obtained after performing a Fourier transform on the vibration data sequence includes: A second data sequence is generated based on the vibration data sequence, and a second difference sequence of the second data sequence is calculated; The second difference sequence is subjected to Fourier transform to obtain the spectral distribution, and the type of respiratory emergency is identified based on the spectral distribution.

5. The respiratory event recognition method as described in claim 1, characterized in that, The step of calculating the current respiratory amplitude based on the vibration data sequence and calculating the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify apnea events further includes: Select a third data sequence from the vibration data sequence, the sequence length of which is multiple respiratory cycles; The median smoothing curve is obtained by performing a weighted average calculation on the third data sequence using a pre-set smoothing coefficient. Calculate the first overlap area between the median smoothing curve and the third data sequence, and calculate the model breathing amplitude based on the first overlap area; The respiratory amplitude model is updated by rolling forward the time series to obtain the respiratory cycle amplitude model.

6. The respiratory event recognition method as described in claim 1, characterized in that, The steps of calculating the current respiratory amplitude based on the vibration data sequence, and calculating the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify apnea events include: Select a fourth data sequence from the vibration data sequence, the sequence length of which is a single respiratory cycle; Calculate the second overlap area between the fourth data sequence and the pre-generated median smoothed curve, and calculate the current respiratory amplitude based on the second overlap area; The respiratory amplitude ratio is calculated based on the current respiratory amplitude and the model respiratory amplitude in the pre-built respiratory cycle amplitude model; The breathing amplitude ratio is compared with a predefined apnea threshold, and apnea events are identified based on the comparison result.

7. The respiratory event recognition method as described in claim 1, characterized in that, The step of calculating the current respiratory amplitude based on the vibration data sequence and calculating the respiratory amplitude ratio based on the current respiratory amplitude and a pre-built respiratory cycle amplitude model to identify apnea events further includes: Calculate the number of overlaps between the pre-generated median smoothed curve and the third data sequence, and correct the smoothing coefficient in the respiratory cycle amplitude model based on the number of overlaps; If the respiratory emergency occurs, the respiratory cycle amplitude model is discarded, and the respiratory cycle amplitude model is reconstructed based on the vibration data sequence.

8. A respiratory event recognition device, characterized in that, The respiratory event recognition device includes: The data acquisition module is used to acquire laryngeal vibration data through the triaxial accelerometer of the laryngeal antisnoring device, and generate a vibration data sequence based on the laryngeal vibration data; A respiratory emergency identification module is used to calculate an environmental noise baseline based on the vibration data sequence and to identify respiratory emergencies based on the environmental noise baseline. The sleep apnea event identification module is used to calculate the current breathing amplitude based on the vibration data sequence, and to calculate the breathing amplitude ratio based on the current breathing amplitude and a pre-built breathing cycle amplitude model to identify sleep apnea events.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the respiratory event recognition method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the respiratory event recognition method as described in any one of claims 1 to 7.

Images