Sleep breathing monitoring method and non-contact sleep breathing monitoring device

Abstract

The present invention provides a sleep breathing monitoring method and a non-contact sleep breathing monitoring device, including: collecting air pressure vibrations caused by the vibrations of the sleeper's body based on an air pressure sensor to obtain air pressure vibration data; extracting breathing-related features from the air pressure vibration data; performing cluster analysis on the breathing-related features to obtain corresponding cluster clusters; performing pattern recognition on the cluster clusters to determine the breathing pattern category; generating a breathing pattern model according to the breathing pattern category, and monitoring the breathing state of the sleeper based on the breathing pattern model to monitor whether the sleeper's breathing is normal. In the present invention, the air pressure vibrations caused by the vibrations of the sleeper's body are collected based on the air pressure sensor, without contacting the sleeper and without causing interference to the sleeper; the air pressure vibration data is analyzed, thereby realizing contactless sleep breathing monitoring of the user.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a sleep respiration monitoring method and a non-contact sleep respiration monitoring device. Background Art

[0002] As people pay more attention to sleep health, sleep monitoring technology is gaining increasing attention.

[0003] Traditional sleep monitoring equipment has many limitations. For example, contact monitoring equipment will affect sleep comfort due to the contact between the sensor and the human body, and even cause the user's sleep quality to decline. In addition, some existing monitoring equipment has a single function and cannot simultaneously meet the accurate monitoring of multiple important sleep parameters. Summary of the invention

[0004] The main purpose of the present invention is to provide a sleep breathing monitoring method and a non-contact sleep breathing monitoring device, aiming to overcome the defects of current contact monitoring devices that affect sleep comfort and have a single function.

[0005] To achieve the above object, the present invention provides a sleep breathing monitoring method, comprising the following steps:

[0006] The air pressure vibration caused by the sleeper's body vibration is collected based on the air pressure sensor to obtain air pressure vibration data;

[0007] Extracting breathing-related features from the air pressure vibration data; wherein the breathing-related features include breathing frequency and breathing amplitude;

[0008] Performing cluster analysis on the breathing-related features to obtain corresponding clusters; performing pattern recognition on the clusters to determine the breathing pattern category;

[0009] A breathing pattern model is generated according to the breathing pattern category, and the breathing state of the sleeper is monitored based on the breathing pattern model to monitor whether the breathing of the sleeper is normal.

[0010] Furthermore, the air pressure sensor is installed inside or at the bottom of the mattress.

[0011] Further, extracting breathing-related features from the air pressure vibration data includes:

[0012] Performing time domain analysis on the air pressure vibration data, calculating the time intervals between adjacent peaks or troughs, taking the average value of the time intervals as a preliminary breathing cycle, and calculating the breathing frequency;

[0013] The amplitude of the air pressure vibration data is statistically analyzed to determine the maximum amplitude and the minimum amplitude in a breathing cycle, and the difference between the two is used as the breathing amplitude.

[0014] Furthermore, pattern recognition is performed on the clusters to determine the breathing pattern categories, including:

[0015] Construct a breathing pattern category library, which includes feature templates of normal breathing patterns and multiple abnormal breathing patterns, including apnea, tachypnea, and shallow and slow breathing;

[0016] Calculate the statistical parameters of each cluster; wherein the statistical parameters include mean and variance;

[0017] The statistical parameters of the clusters are calculated similarly to the feature templates in the breathing pattern category library, and the breathing pattern category to which the clusters belong is determined according to the similarity.

[0018] Furthermore, after monitoring whether the sleeping person's breathing is normal, the method further includes:

[0019] If it is detected that the sleeper has abnormal breathing, the air pressure vibration data and breathing-related characteristics are packaged to obtain packaged data;

[0020] generating an authorization code based on the air pressure vibration data and the breathing-related characteristics;

[0021] Obtaining device information of the management terminal, and generating a request code based on the device information;

[0022] Sending the request code to the management terminal for verification, and sending the permission code to the management terminal for storage;

[0023] After the management terminal verifies that the request code is passed, the packaged data is encrypted based on the permission code and transmitted to the management terminal.

[0024] Further, based on the air pressure vibration data and breathing-related characteristics, an authorization code is generated, including:

[0025] Encoding the air pressure vibration data to obtain encoded data;

[0026] Add the characters in the coded data in sequence to the matrix to obtain a data matrix;

[0027] generating a corresponding frequency curve based on the respiratory frequency, and generating a corresponding amplitude curve based on the respiratory amplitude;

[0028] Extracting features from the frequency curve and the amplitude curve to obtain frequency curve features and amplitude curve features;

[0029] Transforming the data matrix based on the frequency curve characteristics and the amplitude curve characteristics to obtain a transformed data matrix;

[0030] The permission code is generated based on the transformed data matrix and the frequency curve and amplitude curve.

[0031] Further, based on the air pressure vibration data and breathing-related characteristics, an authorization code is generated, including:

[0032] Performing time-frequency analysis on the air pressure vibration data, obtaining a time-frequency graph, and calculating the image entropy of the time-frequency graph as an air pressure vibration characteristic value;

[0033] Performing joint probability distribution statistics on the respiratory frequency and respiratory amplitude to obtain a joint probability distribution matrix, and calculating the determinant value of the joint probability distribution matrix as a respiratory joint eigenvalue;

[0034] Taking the respiratory frequency as the abscissa and the respiratory amplitude as the ordinate, a respiratory phase plane is constructed, and the curvature change rate of the trajectory on the respiratory phase plane is calculated to obtain a curvature feature sequence;

[0035] The air pressure vibration eigenvalue, the breathing joint eigenvalue, and the curvature feature sequence are combined to form high-dimensional feature data;

[0036] The permission code is generated based on the high-dimensional feature data.

[0037] Further, based on the high-dimensional feature data, generating the permission code includes:

[0038] Performing principal component analysis on the high-dimensional feature data to extract principal component vectors;

[0039] Constructing a surface in three-dimensional space with the principal component vector as a parameter, calculating the Gaussian curvature and the mean curvature of the surface, and obtaining a curvature value pair;

[0040] Using the curvature value pair as a control point, a space curve is generated using a cubic spline interpolation method;

[0041] Parameterizing the space curve, and converting the parameterized parameter values ​​into binary codes;

[0042] A cyclic redundancy check calculation is performed on the binary code to obtain a check code; the check code is concatenated with the binary code to obtain a final permission code.

[0043] The present invention also provides a non-contact sleep breathing monitoring device, which is used to perform the steps of any of the above methods in the claims.

[0044] The sleep breathing monitoring method and non-contact sleep breathing monitoring device provided by the present invention include: collecting air pressure vibrations caused by the vibrations of the sleeper's body based on an air pressure sensor to obtain air pressure vibration data; extracting breathing-related features from the air pressure vibration data; wherein the breathing-related features include breathing frequency and breathing amplitude; clustering analysis of the breathing-related features to obtain corresponding clusters; pattern recognition of the clusters to determine the breathing pattern category; generating a breathing pattern model according to the breathing pattern category, and monitoring the breathing state of the sleeper based on the breathing pattern model to monitor whether the sleeper's breathing is normal. In the present invention, the air pressure vibrations caused by the vibrations of the sleeper's body are collected based on the air pressure sensor, without contacting the sleeper and without causing interference to the sleeper; the air pressure vibration data is analyzed to obtain multiple parameters such as breathing frequency and breathing amplitude, thereby realizing non-contact sleep breathing monitoring of the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a schematic diagram of the steps of a sleep breathing monitoring method in one embodiment of the present invention;

[0046] Figure 2 is a structural block diagram of a sleep breathing monitoring device in one embodiment of the present invention;

[0047] Figure 3 It is a schematic block diagram of the structure of a non-contact sleep breathing monitoring device according to an embodiment of the present invention.

[0048] The implementation, functional features and advantages of the present invention will be further described in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0050] Reference Figure 1 In one embodiment of the present invention, a sleep breathing monitoring method is provided, comprising the following steps:

[0051] Step S1, collecting air pressure vibration caused by the sleeper's body vibration based on the air pressure sensor to obtain air pressure vibration data;

[0052] Step S2, extracting breathing-related features from the air pressure vibration data; wherein the breathing-related features include breathing frequency and breathing amplitude;

[0053] Step S3, performing cluster analysis on the breathing-related features to obtain corresponding clusters; performing pattern recognition on the clusters to determine the breathing pattern category;

[0054] Step S4, generating a breathing pattern model according to the breathing pattern category, and monitoring the breathing state of the sleeper based on the breathing pattern model to monitor whether the sleeper's breathing is normal.

[0055] In this embodiment, as described in step S1 above, the sensitive characteristics of the air pressure sensor are used to transmit the tiny vibrations of the sleeper's body during sleep due to breathing, heartbeat, body turning, etc. to the air medium in the sensor, so that the air produces pressure changes. The sensor converts this pressure change into a measurable electrical signal, thereby obtaining air pressure vibration data. This method can non-invasively collect data related to the sleeper's physiological state, avoid the discomfort caused to the sleeper by traditional contact monitoring equipment (such as chest strap respiratory monitors), and improve the naturalness and comfort of sleep monitoring.

[0056] Compared with other types of sensors (such as infrared sensors, accelerometers, etc.), air pressure sensors are more sensitive to weak vibration signals and can more accurately capture the pressure changes caused by the sleeper's body vibration, thereby obtaining more accurate air pressure vibration data and providing more reliable basic data for subsequent respiratory monitoring.

[0057] In the home sleep monitoring scenario, the monitoring device equipped with an air pressure sensor is placed under the mattress. Data can be collected when the sleeper falls asleep normally without affecting the sleep quality. It can continuously monitor the sleeper's sleep breathing conditions for a long time, which helps to detect potential sleep breathing problems, such as sleep apnea hypopnea syndrome.

[0058] As described in step S2 above, by performing signal processing and analysis algorithms on the collected air pressure vibration data, the fluctuation characteristics related to the respiratory cycle are identified. The key features reflecting the respiratory state can be accurately extracted from the complex air pressure vibration data, providing a quantitative basis for subsequent respiratory pattern analysis. Compared with directly using the original data for analysis, feature extraction can reduce the data dimension and the amount of calculation, while highlighting the information related to breathing and improving the analysis efficiency and accuracy.

[0059] In sleep research, researchers can use the breathing frequency and amplitude data extracted in this step to analyze the differences in sleep breathing characteristics among people of different age groups and health conditions, and provide data support for sleep medicine research; in smart mattress products, by extracting breathing characteristics in real time, real-time monitoring and feedback of the sleeper's breathing status can be achieved, providing users with sleep quality assessment and health advice.

[0060] As described in the above-mentioned step S3, cluster analysis divides the breathing-related characteristic data into different groups (cluster clusters) based on the similarity between the data points. For example, the distance between different data points (combination of breathing frequency and amplitude) is calculated using a distance metric method (such as Euclidean distance), and the data points with similar distances are classified as the same cluster cluster. Then, for each cluster cluster, by comparing and matching with a predefined breathing pattern template or model, a pattern recognition algorithm (such as a neural network algorithm, a support vector machine algorithm, etc.) is used to identify the breathing pattern category represented by the cluster cluster. These pattern categories may include normal breathing, rapid breathing, slow breathing, apnea, etc. Cluster analysis helps to discover the inherent structure and regularity in the data, automatically identifies different types of breathing patterns, does not need to manually pre-set category boundaries, and is suitable for complex and diverse breathing data situations. The pattern recognition process can accurately determine the specific type of breathing pattern, providing a basis for subsequent personalized monitoring and intervention.

[0061] As described in the above step S4, according to the different breathing pattern categories determined in step S3, a breathing pattern model is constructed using machine learning or statistical modeling methods. For example, for each breathing pattern category, its characteristic distribution and change rules are learned through a large amount of training data, and a corresponding probability model or classification model is established. During the monitoring process, the breathing-related features obtained in real time are input into the breathing pattern model. The model calculates the matching degree or deviation degree between the current breathing state and the normal breathing pattern according to the pattern features and rules learned in advance, thereby judging whether the breathing is normal. If the deviation from the normal range exceeds a certain threshold, the breathing is judged to be abnormal. The breathing pattern model generated based on actual data is targeted and adaptable, and can more accurately reflect the individual's breathing characteristics and change trends. Through real-time monitoring and model judgment, abnormal breathing can be discovered in time, providing real-time warning and protection for the health of the sleeper.

[0062] In personal health management devices, such as smart bracelets or sleep monitors, users can wear the devices for a long time to generate personalized breathing pattern models based on their personal breathing data. In daily use, the device monitors the user's sleep breathing status in real time. Once an abnormality is found, it will promptly remind the user to pay attention to their health status, such as suggesting that the user adjust their sleeping posture, seek medical examination, etc. In elderly care institutions, caregivers can use this technology to continuously monitor the sleep breathing of the elderly, promptly detect potential health risks, and improve the quality of care and the safety of life of the elderly.

[0063] In one embodiment, the air pressure sensor is installed inside or under the mattress.

[0064] In one embodiment, extracting breathing-related features from the air pressure vibration data includes:

[0065] Performing time domain analysis on the air pressure vibration data, calculating the time intervals between adjacent peaks or troughs, taking the average value of the time intervals as a preliminary breathing cycle, and calculating the breathing frequency;

[0066] The amplitude of the air pressure vibration data is statistically analyzed to determine the maximum amplitude and the minimum amplitude in a breathing cycle, and the difference between the two is used as the breathing amplitude.

[0067] In this embodiment, time domain analysis is to process and analyze the signal directly in the time domain. In sleep breathing monitoring, the human body's breathing process will cause regular vibrations of the body, and this vibration is converted into an air pressure vibration data signal by an air pressure sensor. The peaks and troughs represent the extreme points of the signal on the time axis, and the time interval between adjacent peaks or troughs corresponds to the time span of a breath, that is, the breathing cycle. By calculating the average value of a large number of adjacent peaks or troughs, the influence of single measurement errors can be reduced, and a relatively stable and accurate preliminary breathing cycle estimate can be obtained. The breathing frequency is the number of breaths per unit time. According to the inverse relationship between the breathing cycle and the frequency (breathing frequency = 1 / breathing cycle), the breathing frequency can be calculated.

[0068] The amplitude is the change in the signal in the vertical direction, which reflects the intensity of body vibration caused by breathing. In a complete breathing cycle, the contraction and relaxation of the respiratory muscles will cause the amplitude of body vibration to change. By statistically analyzing the amplitude changes of air pressure vibration data in a breathing cycle, its maximum and minimum values ​​can be determined. The difference between these two values ​​represents the range of changes in the amplitude of body vibration during breathing, that is, the breathing amplitude. The size of the breathing amplitude is related to the depth of breathing. Deeper breathing will produce a larger body vibration amplitude, and vice versa.

[0069] As an important indicator of breathing depth, the respiratory amplitude can be directly obtained from the air pressure vibration data through this simple statistical analysis method, providing additional information for a comprehensive assessment of the sleeping person's breathing state. Combined with the respiratory rate, a more comprehensive understanding of the sleeping person's breathing pattern can be obtained. For example, when judging apnea hypopnea syndrome (SAHS), changes in respiratory rate and amplitude are both of great significance.

[0070] In one embodiment, performing pattern recognition on the clusters to determine the breathing pattern category includes:

[0071] Construct a breathing pattern category library, which includes feature templates of normal breathing patterns and multiple abnormal breathing patterns, including apnea, tachypnea, and shallow and slow breathing;

[0072] Calculate the statistical parameters of each cluster; wherein the statistical parameters include mean and variance;

[0073] The statistical parameters of the clusters are calculated similarly to the feature templates in the breathing pattern category library, and the breathing pattern category to which the clusters belong is determined according to the similarity.

[0074] In this embodiment, the construction of the breathing pattern category library is to provide a reference standard for subsequent pattern recognition. By collecting a large number of breathing-related data (such as characteristic data such as breathing frequency and breathing amplitude) of different individuals in normal breathing state and abnormal breathing patterns such as apnea, tachypnea, shallow breathing, etc., using statistical analysis and data mining technology, the typical features of each breathing pattern are extracted to form a feature template. These feature templates can be described by mathematical models, vector representations or other data structures so as to be compared in the pattern recognition process. For example, for a normal breathing pattern, its feature template is a relatively stable breathing frequency and a moderate breathing amplitude within a certain range; the feature template of the apnea pattern is manifested as the absence or significant weakening of the breathing signal over a period of time; the tachypnea pattern has a higher breathing frequency and a larger breathing amplitude change; the shallow breathing pattern presents a lower breathing frequency and a smaller breathing amplitude.

[0075] The mean is the average value of all data points in the cluster, which reflects the central position of the cluster in the feature space. For respiratory-related feature data (such as respiratory frequency, respiratory amplitude, etc.), calculating the mean can summarize the characteristic performance of the respiratory pattern represented by the cluster at the average level. The variance measures the degree of dispersion of the data points relative to the mean, which reflects the stability and degree of change of the respiratory pattern within the cluster. For example, in a cluster representing normal breathing, the mean of the respiratory frequency should be within the normal respiratory frequency range and the variance is small, indicating that the respiratory frequency is relatively stable; while in a cluster that may represent abnormal breathing, the mean may deviate from the normal range and the variance may be large, indicating that the respiratory pattern has large fluctuations and uncertainties.

[0076] The mean and variance can concisely summarize the central tendency and dispersion of the data within the cluster, and use a small number of statistical parameters to represent the characteristics of the entire cluster, reducing the dimension of the data, facilitating subsequent comparison and calculation with the templates in the breathing pattern category library, and improving the efficiency of pattern recognition.

[0077] The above similarity calculation is a method to measure the similarity between two data objects. In this technical solution, the calculated statistical parameters of the cluster cluster (mean and variance, etc.) are regarded as a vector, and the similarity is calculated with the vector corresponding to each feature template in the breathing pattern category library. Commonly used similarity calculation methods include Euclidean distance, cosine similarity, etc. For example, when using Euclidean distance calculation, the smaller the distance, the more similar the statistical parameters of the cluster cluster are to the feature template; cosine similarity measures the similarity by calculating the cosine value of the angle between the two vectors. The closer the cosine value is to 1, the more similar it is. According to the pre-set similarity threshold, when the similarity between the cluster cluster and a feature template exceeds the threshold, it can be determined that the cluster cluster belongs to the corresponding breathing pattern category.

[0078] The similarity measurement method based on mathematical calculation provides an objective and quantitative way to determine the breathing pattern category to which the cluster belongs, avoiding the error of subjective judgment and improving the accuracy and reliability of recognition.

[0079] In one embodiment, after monitoring whether the sleeping person's breathing is normal, the method further includes:

[0080] If it is detected that the sleeper has abnormal breathing, the air pressure vibration data and breathing-related characteristics are packaged to obtain packaged data;

[0081] generating an authorization code based on the air pressure vibration data and the breathing-related characteristics;

[0082] Obtaining device information of the management terminal, and generating a request code based on the device information;

[0083] Sending the request code to the management terminal for verification, and sending the permission code to the management terminal for storage;

[0084] After the management terminal verifies that the request code is passed, the packaged data is encrypted based on the permission code and transmitted to the management terminal.

[0085] In this embodiment, when the monitoring system detects that the sleeper is breathing abnormally, in order to facilitate the subsequent comprehensive analysis and processing of the abnormal situation, it is necessary to integrate the relevant raw data (air pressure vibration data) and the extracted valuable information (breathing-related features). The process of packaging data is to organize these data into a data packet according to a certain format or rule to ensure the integrity and relevance of the data. For example, the air pressure vibration data can be stored in a specific file format (such as a binary file), and the identification and value of breathing-related features (such as breathing frequency, breathing amplitude and other parameters) can be added to the file header or other locations to form a data packet containing complete abnormal breathing event information.

[0086] The generation of the above permission codes is intended to provide security for packaged data, ensure the confidentiality, integrity and availability of data during transmission and storage, and realize permission management for data access. Through specific algorithms and functions, air pressure vibration data and breathing-related features are used as input, and permission codes are generated after a series of complex calculations and transformations. For example, a hash function can be used to process the data. The hash function has unidirectional and anti-collision properties, and the data is converted into a hash value of a fixed length, which is part of the permission code; or combined with an encryption algorithm (such as symmetric encryption or asymmetric encryption algorithm), the key information in the data is used to generate an encryption key, and then the key is encoded or transformed to obtain the permission code. The generation process of the permission code allows the data to be identified and protected without exposing the original content. Even if the data is intercepted during transmission, the data content cannot be obtained or tampered without the correct permission code, which effectively prevents the leakage of patient privacy and malicious attacks on the data.

[0087] The permission code generated based on air pressure vibration data and breathing-related characteristics is unique and personalized, closely related to specific abnormal breathing events, and can accurately distinguish different monitoring data, facilitating data management and permission control in a multi-user or multi-device environment.

[0088] The permission code can be used as an identifier and index of the data. When necessary, the corresponding original data and respiratory-related characteristics can be traced back through the permission code, which facilitates the auditing, analysis and management of the data, and helps to improve the quality of medical care and the standardization of data management.

[0089] In order to ensure the security and legality of data transmission, the management terminal needs to be authenticated before sending data to the management terminal. Obtain the device information of the management terminal (such as device model, serial number, IP address, MAC address, etc.), which can uniquely identify the management terminal device. Then, the device information is processed through a specific algorithm or function to generate a request code. For example, a digital signature algorithm can be used to use the device information as the message content of the signature, and the private key is used to sign it. The generated signature result is the request code; or a hash function can be used to hash the device information to obtain a fixed-length hash value as the request code. The request code contains the characteristic information of the management terminal device, but does not directly expose the device information itself, ensuring a certain degree of security.

[0090] The generation and verification process of the request code realizes the identity authentication of the management terminal device, ensuring that only legally authorized devices can receive and process monitoring data, preventing illegal devices from accessing the system to obtain sensitive information, and improving the data security of the entire monitoring system. Since the request code is generated based on the device information and is closely associated with the device information during the transmission process, if the device information is tampered with during the transmission process, the request code will not be able to pass the verification, thereby promptly discovering and preventing malicious tampering during the data transmission process and ensuring the integrity of the data. The request code is associated with the management terminal device information. During the data transmission and processing process, the request code can be traced back to the receiving device of the data, which facilitates the monitoring and management of the data flow. When problems occur, the responsible party can be quickly located, improving the manageability and security of the system.

[0091] After receiving the request code, the management terminal uses the corresponding verification algorithm or key to verify the request code. If the request code is verified, it indicates that the device sending the data is a legally authorized device, and the management terminal allows the subsequent data to be received. At the same time, the management terminal receives and stores the permission code, which will be used for subsequent encryption and decryption operations of the packaged data and control of data access rights. When verifying the request code, the management terminal determines the validity of the request code based on pre-set verification rules (such as comparing with the stored device information database, using the public key to verify the digital signature, etc.). When storing the permission code, store it in a secure storage area (such as a specific field in a database or an encrypted file), and establish an association with the relevant packaged data or device so that the permission code can be accurately obtained and used later.

[0092] The request code verification mechanism ensures that only legitimate devices can establish a data transmission channel with the management terminal, effectively preventing illegal devices from attacking the management terminal and stealing data, laying the foundation for subsequent secure data transmission. The storage of permission codes provides a basis for subsequent access control of packaged data. Only users or system modules with the correct permission code (or with the corresponding permission to obtain the permission code) can decrypt and access the packaged data, realizing fine-grained data access permission management and protecting the confidentiality and privacy of the data. The permission code is stored after the request code verification is passed, ensuring the integrity and consistency of the entire data transmission and storage process. If the request code verification fails, the data transmission will be interrupted, preventing incomplete or illegal data from entering the management terminal system and ensuring the quality and security of the system data.

[0093] When the management terminal successfully verifies the request code, it uses the previously stored permission code as the encryption key or encryption parameter to encrypt the packaged data. The encryption algorithm can be selected according to the security requirements and performance requirements of the system, such as the Advanced Encryption Standard (AES) encryption algorithm, the RSA encryption algorithm, etc. The encryption process converts the packaged data into ciphertext form, making it impossible for unauthorized parties to read and understand it during transmission. Even if the data is intercepted, it cannot be decrypted to obtain the original data content without the correct permission code. The encrypted packaged data is transmitted to the management terminal via a secure transmission protocol (such as the HTTPS protocol). After receiving the ciphertext data, the management terminal performs a decryption operation according to the permission code to restore the original packaged data for subsequent processing.

[0094] Encryption based on permission codes ensures the confidentiality of packaged data during transmission, prevents the leakage of patients' sensitive health information (such as data related to abnormal respiratory events), protects patients' privacy, and complies with relevant regulatory requirements on medical data security and privacy protection.

[0095] The encryption process not only protects the data content from being leaked, but also prevents the data from being tampered with during transmission. Any illegal modification of the ciphertext data will result in decryption failure, thus ensuring the integrity and authenticity of the data during transmission.

[0096] In one embodiment, based on the air pressure vibration data and the breathing-related characteristics, an authorization code is generated, including:

[0097] Encoding the air pressure vibration data to obtain encoded data;

[0098] Add the characters in the coded data in sequence to the matrix to obtain a data matrix;

[0099] generating a corresponding frequency curve based on the respiratory frequency, and generating a corresponding amplitude curve based on the respiratory amplitude;

[0100] Extracting features from the frequency curve and the amplitude curve to obtain frequency curve features and amplitude curve features;

[0101] Transforming the data matrix based on the frequency curve characteristics and the amplitude curve characteristics to obtain a transformed data matrix;

[0102] The permission code is generated based on the transformed data matrix and the frequency curve and amplitude curve.

[0103] In this embodiment, encoding is to convert the original air pressure vibration data into a representation form that is more suitable for subsequent processing and storage. Through a specific encoding algorithm, the continuous air pressure vibration data is mapped to discrete encoding values. For example, differential encoding can be used to calculate the difference between adjacent sampling points and convert these differences into binary codes; or an adaptive encoding method based on data statistical characteristics can be used to dynamically adjust the encoding rules according to the distribution of the data to improve the encoding efficiency. The purpose of encoding is to reduce data redundancy while retaining data information as much as possible, so as to facilitate data transmission, storage and further processing.

[0104] Organizing the encoded data into a matrix form helps to take advantage of the advantages of matrix operations for subsequent data processing. The characters in the encoded data are filled into the element positions of the matrix one by one in order. It can be a one-dimensional matrix (vector) or a two-dimensional matrix, depending on the characteristics of the data and subsequent processing requirements. For example, if the encoded data has certain time series characteristics, it can be arranged into a two-dimensional matrix, where the rows represent different fragments in the time series and the columns represent the data encoding values ​​in each fragment. Such a data matrix structure provides a convenient data organization form for subsequent matrix transformation, feature extraction and other operations, so that the data has a clearer structural relationship in space, which is convenient for overall analysis and processing.

[0105] Respiratory frequency and respiratory amplitude are important features that reflect the respiratory state of a sleeper. With time as the horizontal axis and respiratory frequency and respiratory amplitude as the vertical axis, corresponding curves can be drawn. The respiratory frequency curve shows the changing trend of respiratory frequency over a period of time, while the respiratory amplitude curve reflects the changes in breathing depth. The generation of these curves helps to intuitively observe the dynamic changes of breathing patterns and provides a visual basis for subsequent feature extraction and pattern recognition. For example, under normal breathing conditions, the respiratory frequency curve may show relatively stable periodic fluctuations, while the respiratory amplitude curve may fluctuate within a certain range; when abnormal breathing occurs (such as apnea, shortness of breath, etc.), the curves will show obvious changes, such as interruption of the frequency curve, abnormal increase or decrease of the amplitude curve, etc.

[0106] The frequency curve and amplitude curve present the breathing status in a graphical way, which is easier to understand and observe than a simple data table. Doctors, researchers or users can quickly judge whether the sleeping person's breathing is normal, whether there is an abnormal breathing pattern, and the time and duration of its occurrence by observing the shape, period, amplitude and other characteristics of the curve.

[0107] As a continuous function representation, the curve facilitates the extraction of more complex features. For example, the slope, curvature, peak and valley values ​​and other geometric features of the curve can be calculated. These features can further characterize the detailed changes in the breathing pattern and provide more basis for accurately identifying breathing abnormalities. At the same time, the spectrum analysis of the curve (through Fourier transform and other methods) can reveal the distribution characteristics of the breathing frequency and amplitude in the frequency domain, which helps to discover the deep information hidden in the signal.

[0108] Feature extraction is to extract representative and discriminative information from the original frequency curve and amplitude curve. For the frequency curve, the features that can be extracted include average frequency, frequency standard deviation, frequency change rate, number of frequency peaks, frequency peak-to-valley difference, etc.; for the amplitude curve, the features that can be extracted include average amplitude, amplitude standard deviation, amplitude change rate, area under the amplitude curve, maximum amplitude duration, etc. These features can characterize the characteristics of the breathing pattern from different angles. For example, the average frequency and average amplitude reflect the basic rhythm and depth of breathing; the frequency standard deviation and amplitude standard deviation reflect the stability of breathing, and a larger standard deviation may indicate irregular breathing or abnormal fluctuations; the frequency change rate and amplitude change rate can capture the dynamic changes of the breathing pattern, which helps to detect the occurrence of abnormal conditions such as shortness of breath or slow breathing; the number of frequency peaks and the area under the amplitude curve will show obvious changes in certain abnormal breathing patterns (such as compensatory shortness of breath after apnea), which can serve as an important basis for diagnosis.

[0109] Transforming the data matrix using features extracted from the frequency curve and amplitude curve is a way to integrate respiratory frequency and amplitude information into the data matrix to enhance the data matrix's ability to characterize the respiratory state. The specific transformation method can be selected according to the nature of the features and the data processing requirements. For example, the weights of the corresponding elements in the data matrix can be adjusted according to the frequency curve characteristics so that the information related to the respiratory frequency is prominently reflected in the matrix; or the matrix can be transformed nonlinearly according to the amplitude curve characteristics, such as stretching or compressing the matrix elements according to the amplitude, so that the data matrix can better reflect the impact of changes in respiratory amplitude on air pressure vibrations. This transformation operation is intended to enable the data matrix to be more closely associated with the respiratory state after processing, providing a more representative and secure data basis for the subsequent generation of permission codes.

[0110] By integrating the frequency curve characteristics and amplitude curve characteristics into the data matrix, each element in the data matrix is ​​associated with the respiratory frequency and amplitude information, which improves the correlation between the data matrix as a whole and the respiratory state. Since the transformed data matrix contains more complex information related to breathing, it is difficult for attackers to obtain the original respiratory data or crack the permission code through simple analysis methods. Even if part of the data is leaked, since the data has been transformed based on the respiratory characteristics, it is difficult for attackers to restore the real breathing pattern, thereby protecting the privacy and data security of patients. There are differences in the breathing patterns of different individuals. The transformation based on the frequency curve characteristics and amplitude curve characteristics can personalize the data matrix according to the respiratory characteristics of each individual. The permission code generated in this way not only has a universal security function, but can also be customized according to the respiratory characteristics of each individual, further improving the security and applicability of the permission code.

[0111] The process of generating the permission code comprehensively considers the information in the transformation data matrix and the frequency curve and amplitude curve. One possible method is to perform mathematical operations on the transformation data matrix (such as hash operations, eigenvalue combination after matrix decomposition, etc.) to obtain an intermediate result, and at the same time perform feature encoding on the frequency curve and amplitude curve (such as converting the key eigenvalues ​​of the curve into a specific encoding format), and then splice the intermediate result and the feature encoding or further encrypt them to finally obtain the permission code. For example, the transformation data matrix can be first subjected to singular value decomposition, and some singular values ​​can be taken as the intermediate result. Then, the eigenvalues ​​of the frequency curve and the amplitude curve (such as average frequency, average amplitude, etc.) are converted into binary codes and spliced ​​with the intermediate results. Finally, the spliced ​​results are encrypted (such as using a symmetric encryption algorithm or an asymmetric encryption algorithm), and the ciphertext obtained is the permission code. The generation process of the permission code aims to use the multiple features of the data to generate an identifier that is unique, secure, and closely related to the respiratory state, which is used for permission management and security protection of the data.

[0112] In one embodiment, based on the air pressure vibration data and the breathing-related characteristics, an authorization code is generated, including:

[0113] Performing time-frequency analysis on the air pressure vibration data, obtaining a time-frequency graph, and calculating the image entropy of the time-frequency graph as an air pressure vibration characteristic value;

[0114] Performing joint probability distribution statistics on the respiratory frequency and respiratory amplitude to obtain a joint probability distribution matrix, and calculating the determinant value of the joint probability distribution matrix as a respiratory joint eigenvalue;

[0115] Taking the respiratory frequency as the abscissa and the respiratory amplitude as the ordinate, a respiratory phase plane is constructed, and the curvature change rate of the trajectory on the respiratory phase plane is calculated to obtain a curvature feature sequence;

[0116] The air pressure vibration eigenvalue, the breathing joint eigenvalue, and the curvature feature sequence are combined to form high-dimensional feature data;

[0117] The permission code is generated based on the high-dimensional feature data.

[0118] In this embodiment, time-frequency analysis is a method of simultaneously analyzing a signal in two dimensions, time and frequency. For air pressure vibration data, it changes over time and contains different frequency components (such as low-frequency components related to breathing frequency, etc.). The data is converted into a time-frequency diagram through time-frequency analysis techniques (such as short-time Fourier transform, wavelet transform, etc.). In the time-frequency diagram, the horizontal axis represents time, the vertical axis represents frequency, and the grayscale or color intensity of the image represents the energy or amplitude of the signal at this time and frequency. Image entropy is an indicator to measure the richness or uncertainty of image information. Calculating the image entropy of the time-frequency diagram can quantify the complexity and distribution characteristics of air pressure vibration data in the time-frequency domain. Higher image entropy may indicate that there are more frequency changes or irregularities in the respiratory signal, while lower image entropy may correspond to a more stable and regular breathing pattern.

[0119] Time-frequency analysis overcomes the limitations of traditional time-domain or frequency-domain analysis when used alone, and can simultaneously capture the changing characteristics of air pressure vibration data in time and frequency, providing more comprehensive information. Image entropy as a eigenvalue further condenses this information, and a single value can characterize the overall characteristics of the time-frequency graph, facilitating subsequent data processing and analysis.

[0120] Joint probability distribution statistics are used to describe the probability of two random variables (i.e., respiratory rate and respiratory amplitude) occurring at the same time. By collecting respiratory rate and respiratory amplitude data over a certain period of time, a joint probability distribution matrix is ​​constructed. Each element in the matrix represents the probability of occurrence under a specific respiratory rate and respiratory amplitude combination. The determinant value is an important mathematical feature of the matrix. It integrates the information of all elements in the matrix and reflects the characteristics of the data structure represented by the matrix to a certain extent. For the joint probability distribution matrix, its determinant value can be used as an overall indicator to characterize the relationship between respiratory rate and respiratory amplitude and the complexity of the joint distribution. For example, if there is a strong linear relationship between respiratory rate and respiratory amplitude, the determinant value may be small; if the relationship between the two is more complex and nonlinear, the determinant value may be large.

[0121] Constructing the respiratory frequency and respiratory amplitude into a respiratory phase plane is a method to intuitively represent the dynamic changes of the respiratory state. On this two-dimensional plane, the respiratory state at each moment can be represented by a point, and these points form a trajectory over time. The curvature of the trajectory reflects the rate and direction of change in the respiratory state. Calculating the curvature change rate of the trajectory further describes the change of curvature over time, that is, the acceleration of the change of the respiratory state. For example, during normal breathing, the trajectory on the respiratory phase plane may be relatively smooth and the curvature change rate is small; while in abnormal breathing (such as shortness of breath or apnea recovery stage), the trajectory may be sharply bent or twisted, resulting in an increase in the curvature change rate. By calculating the sequence of curvature change rates over the entire respiratory cycle or a period of time, the dynamic change characteristics of the respiratory state can be captured, which are important for accurately identifying breathing patterns and judging respiratory abnormalities.

[0122] The air pressure vibration eigenvalues, respiratory joint eigenvalues ​​and curvature feature sequences extracted from different angles are combined to construct a high-dimensional feature data vector. Each eigenvalue or feature sequence represents an aspect of the respiratory state, and a more comprehensive and representative feature description is formed by combination. Mathematically, these features can be arranged into a vector in a certain order, where each element corresponds to a eigenvalue or a data point in the feature sequence. For example, if the air pressure vibration eigenvalue is a scalar, the respiratory joint eigenvalue is a scalar, and the curvature feature sequence is a vector containing multiple data points, then the combined high-dimensional feature data vector can be expressed as [air pressure vibration eigenvalue, respiratory joint eigenvalue, the first data point in the curvature feature sequence, the second data point in the curvature feature sequence, ...]. Such a high-dimensional feature data vector can comprehensively reflect the information of the respiratory state in multiple aspects such as the air pressure vibration characteristics, the joint distribution of respiratory frequency and amplitude, and the dynamic changes of the respiratory state.

[0123] The generation of the permission code is based on high-dimensional feature data, aiming to use these data containing rich respiratory state information to create a unique and secure identification. A common method is to use a hash function to process the high-dimensional feature data. The hash function has the characteristics of one-way and anti-collision. After the high-dimensional feature data is input into the hash function, a hash value of fixed length will be obtained, and the above hash value can be used as the permission code. Since the high-dimensional feature data contains many aspects of the respiratory state, different respiratory patterns and individual feature data will produce different hash values, thereby ensuring the uniqueness of the permission code. At the same time, the characteristics of the hash function make it difficult to reversely derive the original high-dimensional feature data from the permission code, ensuring the security of the data. In addition, the hash value can be further encrypted in combination with other encryption technologies (such as asymmetric encryption algorithms) to increase the security of the permission code and prevent the permission code from being cracked and forged.

[0124] In one embodiment, generating the permission code based on the high-dimensional feature data includes:

[0125] Performing principal component analysis on the high-dimensional feature data to extract principal component vectors;

[0126] Constructing a surface in three-dimensional space with the principal component vector as a parameter, calculating the Gaussian curvature and the mean curvature of the surface, and obtaining a curvature value pair;

[0127] Using the curvature value pair as a control point, a space curve is generated using a cubic spline interpolation method;

[0128] Parameterizing the space curve, and converting the parameterized parameter values ​​into binary codes;

[0129] A cyclic redundancy check calculation is performed on the binary code to obtain a check code; the check code is concatenated with the binary code to obtain a final permission code.

[0130] In the present embodiment, principal component analysis (PCA) is a commonly used data dimension reduction technology. For high-dimensional feature data (composed of air pressure vibration eigenvalues, breathing joint eigenvalues, curvature feature sequences, etc.), its dimension may be higher, containing a large amount of information, but some of the information may be redundant or relevant. PCA projects the original high-dimensional data to a low-dimensional space (usually selecting several principal components that retain the main information) by linear transformation, while keeping the variance (i.e. information content) of the data as much as possible. In this process, the covariance matrix of the data is calculated, and its eigenvalues ​​and eigenvectors are sought, and the main eigenvectors are selected as principal component vectors according to the size of the eigenvalues. These principal component vectors are the main representation directions of the original high-dimensional feature data in the low-dimensional space, and they can summarize most of the variation information of the original data, thereby retaining key information while realizing data dimension reduction.

[0131] The above process reduces the dimensionality of high-dimensional feature data, reduces the subsequent computational effort and storage space requirements, and highlights the main information in the data. For example, if the original high-dimensional feature data has dozens of dimensions, only a few (such as 2-3) principal component vectors may be retained after PCA analysis. These principal component vectors can explain most of the variance of the original data, making data processing more efficient without losing too much key information.

[0132] Using the principal component vector as a parameter to construct a surface in three-dimensional space is a method of mapping low-dimensional vector information to three-dimensional space so as to further extract features using the geometric properties of the surface. In three-dimensional space, the coordinates of each point are determined by the elements of the principal component vector, and a continuous surface is constructed through a mathematical model (such as based on spline functions or other surface fitting methods). Gaussian curvature and mean curvature are two important geometric features of the surface. Gaussian curvature describes the curvature of the surface at a certain point, which reflects the local shape characteristics of the surface near the point. For example, positive Gaussian curvature indicates that the surface is convex outward like a sphere at this point, and negative Gaussian curvature indicates that it is concave inward like a saddle. Mean curvature is an average measure of the curvature of the surface, which comprehensively considers the curvature of the surface in all directions. By calculating the Gaussian curvature and mean curvature of the surface at different points, a series of curvature value pairs are obtained, which can characterize the distribution and changes of the respiratory state characteristics represented by the principal component vector in three-dimensional space from a geometric perspective.

[0133] Cubic spline interpolation is a mathematical method for constructing smooth curves. It generates a smooth curve passing through a given series of control points (here, curvature value pairs), and the curve segments between adjacent control points are cubic polynomial functions. For points in three-dimensional space determined by curvature value pairs, the cubic spline interpolation method calculates a continuous and smooth curve in three-dimensional space, making the curve as close to these control points as possible, while ensuring that the curve has continuous first-order and second-order derivatives at the connection, that is, the smoothness of the curve. This spatial curve can connect the spatial points represented by each curvature value pair in a continuous and smooth manner, thereby integrating the discrete curvature value pair information into a continuous geometric object, further mining the potential patterns and trends in the data.

[0134] Converting discrete curvature value pairs into continuous spatial curves makes the data representation in space smoother and more natural. This smoothness helps reduce the impact of noise and mutations in the data on subsequent analysis and more accurately reflects the overall change trend of respiratory state characteristics. For example, during the breathing process, even if there are some minor fluctuations or measurement errors, the smooth curve generated by cubic spline interpolation can filter out these short-term irregular changes and highlight the long-term stable characteristics of the breathing pattern.

[0135] The purpose of parameterizing a space curve is to use a set of parameters to uniquely describe the points on the curve so that the curve can be represented in the form of a mathematical function. Common parameterization methods include arc length parameterization. By selecting appropriate parameters (such as arc length), each point on the curve can be associated with a parameter value, so that the entire curve can be represented as a function of this parameter. These parameter values ​​are then converted into binary encoding, which is the process of converting continuous curve parameter information into digital information for processing and storage in a computer system. Binary encoding can use fixed-length or variable-length encoding methods, and the number of bits of the encoding is determined according to the range and precision requirements of the parameter values. For example, if the parameter value range is between 0 and 1 and a certain precision is required, it can be converted into 8-bit or 16-bit binary encoding, each encoding corresponding to a specific range of parameter values, thereby converting the geometric information of the space curve into a binary digital sequence that the computer can understand and process.

[0136] The above-mentioned cyclic redundancy check (CRC) is a technology used to detect whether errors occur during data transmission or storage. It is based on the principle of polynomial division, and regards the binary code as a coefficient sequence of a polynomial. It divides it through a predefined generating polynomial, and the remainder obtained is the check code. The length of the check code is usually determined according to the importance of the data and the requirements for error detection capabilities, and is generally a fixed length (such as 8 bits, 16 bits, etc.). After the check code is generated, it is spliced ​​with the original binary code to form the final permission code. The splicing method can be to add a check code at the end of the binary code, or to insert the check code into a specific position of the binary code according to specific rules. In this way, when the permission code is used subsequently, the check code can be recalculated and compared with the check code part in the permission code to verify whether the data is complete and has not been tampered with. If the calculated check code is consistent with the check code in the permission code, it means that there is a high probability that there is no error in the data during transmission or storage; on the contrary, if the two are inconsistent, it means that the data may have been damaged or tampered with, and corresponding measures need to be taken (such as retransmitting the data, checking the source of the data, etc.).

[0137] In one embodiment, generating the permission code based on the transformation data matrix and the frequency curve and the amplitude curve includes:

[0138] According to a preset rule, the frequency curve and the amplitude curve are superimposed on the transformation data matrix;

[0139] Respectively calculating the positional relationship between each matrix position in the transformation data matrix and the frequency curve and the amplitude curve;

[0140] When the position relationship meets the preset requirements, the matrix element at the corresponding matrix position is used as the target character;

[0141] The target characters are combined to obtain the permission code.

[0142] In the present embodiment, the above-mentioned preset rules define how the frequency curve and the amplitude curve are fused with the transformation data matrix. A possible rule is to regard the frequency curve and the amplitude curve as curves on a two-dimensional plane, and after discretizing them, according to the coordinate system of the matrix, the point values ​​on the curve are mapped to the corresponding positions in the transformation data matrix. For example, the value of each data point on the frequency curve can be placed in a row (or column) of the matrix according to the time sequence or the index order of the data point, and the value of the data point corresponding to the time point or index on the amplitude curve is placed in another row (or column) of the matrix, so as to realize the superposition of the curve and the matrix in space. The purpose of doing so is to combine the respiratory dynamic feature information contained in the frequency curve and the amplitude curve with the information obtained by processing based on the air pressure vibration data in the transformation data matrix, so that the data based on which the authority code is generated can comprehensively reflect multiple aspects of the respiratory state.

[0143] Calculating the positional relationship requires defining a metric to describe the relative positional relationship between the matrix position and the point on the curve. A common method is to calculate the difference between the matrix position coordinates (e.g., row index and column index) and the time index or other related index of the corresponding data point on the curve. The distance from each matrix position to the nearest point on the frequency curve and amplitude curve can be calculated separately (under a certain defined distance metric, such as Euclidean distance), or it can be determined whether the matrix position is located in a specific area of ​​the curve (such as the rising section, falling section, near the peak of the curve, etc.), which can all be used as a measure of the positional relationship. Through this calculation, the degree of spatial correlation between each element in the matrix and the dynamic change characteristics of respiratory frequency and amplitude can be determined, providing a basis for subsequent screening of target characters.

[0144] The above preset requirements define which positional relationships are meaningful or related to specific breathing patterns. For example, the preset requirement can be that the distance from the matrix position to the peak of the frequency curve is within a certain range and is simultaneously located in the rising section of the amplitude curve; or that a certain combination of the matrix position and the frequency curve and the amplitude curve satisfies specific mathematical conditions (such as the sum of the distances is less than a certain threshold, or is located in the area near a specific intersection of the two curves, etc.). When a matrix position meets these preset requirements, it means that the element at that position has a specific association with the characteristics of the breathing frequency and amplitude, and may represent some key information of the breathing state, so it is selected as the target character. These target characters will serve as the basic elements for generating permission codes, and their combination can reflect the specific feature combination of the breathing state.

[0145] Finally, the selected target characters are combined in a certain order to form a string or digital sequence, which is the permission code. The order of combination can be determined according to preset rules, such as row priority or column priority of matrix position, or the order of occurrence of target characters and respiratory events. As a whole, the combined permission code represents a unique identifier generated according to the characteristics of a specific respiratory state (the target characters selected by the position relationship between the frequency curve, amplitude curve and the transformed data matrix). This permission code can be used for data access control, data integrity verification, respiratory state classification and identification, etc. For example, in data access control, only users or systems that provide the correct permission code can access the corresponding respiratory data; in data integrity verification, by recalculating the permission code and comparing it with the stored original permission code, it can be checked whether the data has been tampered with during transmission or storage; in respiratory state classification and identification, different permission code modes can correspond to different respiratory mode categories, and the general type of respiratory state can be quickly determined by analyzing the permission code.

[0146] Reference Figure 2 In one embodiment of the present invention, a sleep breathing monitoring device is provided, comprising:

[0147] A collection unit, used to collect air pressure vibration caused by the sleeper's body vibration based on the air pressure sensor to obtain air pressure vibration data;

[0148] An extraction unit, used to extract breathing-related features from the air pressure vibration data; wherein the breathing-related features include breathing frequency and breathing amplitude;

[0149] An analysis unit, configured to perform cluster analysis on the breathing-related features to obtain corresponding clusters; and perform pattern recognition on the clusters to determine the breathing pattern category;

[0150] The monitoring unit is used to generate a breathing pattern model according to the breathing pattern category, monitor the breathing state of the sleeper based on the breathing pattern model, and monitor whether the sleeping person's breathing is normal.

[0151] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0152] The present invention also provides a non-contact sleep breathing monitoring device, which is used to perform the steps of any of the above methods in the claims.

[0153] Its internal structure can be Figure 3As shown. It includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the above-mentioned processor is used to provide computing and control capabilities. The above-mentioned memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The above-mentioned internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The above-mentioned database is used to store the corresponding data in this embodiment. The above-mentioned network interface is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above-mentioned method is implemented.

[0154] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0155] In summary, the sleep breathing monitoring method and non-contact sleep breathing monitoring device provided in the embodiments of the present invention include: collecting air pressure vibrations caused by the vibrations of the sleeper's body based on an air pressure sensor to obtain air pressure vibration data; extracting breathing-related features from the air pressure vibration data; wherein the breathing-related features include breathing frequency and breathing amplitude; clustering analysis of the breathing-related features to obtain corresponding clusters; pattern recognition of the clusters to determine the breathing pattern category; generating a breathing pattern model according to the breathing pattern category, and monitoring the breathing state of the sleeper based on the breathing pattern model to monitor whether the sleeper's breathing is normal. In the present invention, the air pressure vibrations caused by the vibrations of the sleeper's body are collected based on the air pressure sensor, without contacting the sleeper and without causing interference to the sleeper; the air pressure vibration data is analyzed to obtain multiple parameters such as breathing frequency and breathing amplitude, thereby realizing non-contact sleep breathing monitoring of the user.

[0156] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0157] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0158] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A sleep breathing monitoring method, characterized in that: The following steps are involved: The air pressure vibration caused by the sleeper's body vibration is collected based on the air pressure sensor to obtain air pressure vibration data; Extracting breathing-related features from the air pressure vibration data; wherein the breathing-related features include breathing frequency and breathing amplitude; Performing cluster analysis on the breathing-related features to obtain corresponding clusters; performing pattern recognition on the clusters to determine the breathing pattern category; A breathing pattern model is generated according to the breathing pattern category, and the breathing state of the sleeper is monitored based on the breathing pattern model to monitor whether the breathing of the sleeper is normal.

2. The sleep breathing monitoring method according to claim 1, characterized in that: The air pressure sensor is installed inside or at the bottom of the mattress.

3. The sleep breathing monitoring method according to claim 1, characterized in that: Extracting breathing-related features from the air pressure vibration data includes: Performing time domain analysis on the air pressure vibration data, calculating the time intervals between adjacent peaks or troughs, taking the average value of the time intervals as a preliminary breathing cycle, and calculating the breathing frequency; The amplitude of the air pressure vibration data is statistically analyzed to determine the maximum amplitude and the minimum amplitude in a breathing cycle, and the difference between the two is used as the breathing amplitude.

4. The sleep breathing monitoring method according to claim 1, characterized in that: Perform pattern recognition on clusters to determine breathing pattern categories, including: Construct a breathing pattern category library, which includes feature templates of normal breathing patterns and multiple abnormal breathing patterns, including apnea, tachypnea, and shallow and slow breathing; Calculate the statistical parameters of each cluster; wherein the statistical parameters include mean and variance; The statistical parameters of the clusters are calculated similarly to the feature templates in the breathing pattern category library, and the breathing pattern category to which the clusters belong is determined according to the similarity.

5. The sleep breathing monitoring method according to claim 1, characterized in that: After monitoring whether the sleeping person's breathing is normal, the method further includes: If it is detected that the sleeper has abnormal breathing, the air pressure vibration data and breathing-related characteristics are packaged to obtain packaged data; generating an authorization code based on the air pressure vibration data and the breathing-related characteristics; Obtaining device information of the management terminal, and generating a request code based on the device information; Sending the request code to the management terminal for verification, and sending the permission code to the management terminal for storage; After the management terminal verifies that the request code is passed, the packaged data is encrypted based on the permission code and transmitted to the management terminal.

6. The sleep breathing monitoring method according to claim 5, characterized in that: Based on the air pressure vibration data and breathing-related characteristics, an authorization code is generated, including: Encoding the air pressure vibration data to obtain encoded data; Add the characters in the coded data in sequence to the matrix to obtain a data matrix; generating a corresponding frequency curve based on the respiratory frequency, and generating a corresponding amplitude curve based on the respiratory amplitude; Extracting features from the frequency curve and the amplitude curve to obtain frequency curve features and amplitude curve features; Transforming the data matrix based on the frequency curve characteristics and the amplitude curve characteristics to obtain a transformed data matrix; The permission code is generated based on the transformed data matrix and the frequency curve and amplitude curve.

7. The sleep breathing monitoring method according to claim 5, characterized in that: Based on the air pressure vibration data and breathing-related characteristics, an authorization code is generated, including: Performing time-frequency analysis on the air pressure vibration data, obtaining a time-frequency graph, and calculating the image entropy of the time-frequency graph as an air pressure vibration characteristic value; Performing joint probability distribution statistics on the respiratory frequency and respiratory amplitude to obtain a joint probability distribution matrix, and calculating the determinant value of the joint probability distribution matrix as a respiratory joint eigenvalue; Taking the respiratory frequency as the abscissa and the respiratory amplitude as the ordinate, a respiratory phase plane is constructed, and the curvature change rate of the trajectory on the respiratory phase plane is calculated to obtain a curvature feature sequence; The air pressure vibration eigenvalue, the breathing joint eigenvalue, and the curvature feature sequence are combined to form high-dimensional feature data; The permission code is generated based on the high-dimensional feature data.

8. The sleep breathing monitoring method according to claim 7, characterized in that: Generating the permission code based on the high-dimensional feature data includes: Performing principal component analysis on the high-dimensional feature data to extract principal component vectors; Constructing a surface in three-dimensional space with the principal component vector as a parameter, calculating the Gaussian curvature and the mean curvature of the surface, and obtaining a curvature value pair; Using the curvature value pair as a control point, a space curve is generated using a cubic spline interpolation method; Parameterizing the space curve, and converting the parameterized parameter values ​​into binary codes; A cyclic redundancy check calculation is performed on the binary code to obtain a check code; the check code is concatenated with the binary code to obtain a final permission code.

9. A non-contact sleep breathing monitoring device, used to perform the steps of the method according to any one of claims 1 to 8.

Images