Respiration depth measurement method, device and storage medium

Abstract

The present application discloses a method, device and storage medium for measuring respiratory depth. The method includes: obtaining a pressure signal generated by a pressure matrix, and determining the area where the monitored object is located and the sleeping position of the monitored object based on the pressure signal, wherein the area where the monitored object is located includes the chest and abdomen area of ​​the monitored object; determining the contact matrix between the chest and abdomen area of ​​the monitored object and each airbag, and determining the contact airbag based on the contact matrix; determining the pressure baseline value and respiratory signal amplitude of each contact airbag; determining the pressure mean of the pressure submatrix corresponding to each contact airbag; obtaining the respiratory depth corresponding to each contact airbag based on the sleeping position of the monitored object, the contact matrix, the pressure mean, the pressure baseline value and the respiratory signal amplitude, and performing weighted summation on the respiratory depth corresponding to each contact airbag to obtain the final respiratory depth. Accurately perceive the pressure changes in different positions and sleeping positions, reduce the influence of different factors on the accuracy of respiratory measurement, and provide stable and accurate respiratory monitoring services.

Description

Technical Field

[0001] The present application relates to the field of smart home technology, and in particular to a breathing depth measurement method, device, and storage medium. Background Art

[0002] Breathing is vital to human health, and different respiratory behaviors are closely related to health status. Abnormalities such as shortness of breath, hypopnea, or brief respiratory pauses when not exercising may be caused by respiratory diseases. For users with a history of related diseases, monitoring respiratory status can effectively reduce the damage caused by emergencies; for healthy users, it can also provide timely risk alerts and prevent risks in advance. Respiration depth, as a key characteristic for measuring respiratory intensity, is of great significance.

[0003] To accurately measure respiratory status, various respiratory monitoring methods have been developed. Traditional methods, including spirometers, breathing masks, and breathing belts, primarily obtain respiratory metrics by directly measuring respiratory airflow or the intensity of the floating movement of the respiratory area. Taking the spirometer as an example, the measurement requires the user to blow air through a specific device with their mouth. The total airflow is measured under preset time and other rules, and then converted into vital capacity and related respiratory metrics through an internal algorithm. While these methods are highly accurate, prolonged monitoring can seriously affect user comfort.

[0004] In recent years, monitoring methods that use optical, acoustic, and electrical sensors to measure human physiological signals have gradually gained popularity, such as in wearable devices like wristbands and watches. A common approach is to wear a watch on the wrist, collect wrist photoplethysmography (PPG) signals, and apply signal processing methods to derive respiratory-related parameter indicators. These methods often rely on a relatively simple sensor type, such as PPG, making it difficult to fully capture the complex physiological signal changes during breathing. Signal interference is also prone to fluctuations in ambient light, uneven wear of the device, or slight user movement, resulting in relatively low accuracy. Summary of the Invention

[0005] Based on this, it is necessary to provide a breathing depth measurement method, device and storage medium to address the above technical problems, so as to solve at least one problem existing in the above-mentioned prior art.

[0006] In a first aspect, a method for measuring breathing depth is provided, which is applied to an intelligent monitoring device, wherein the intelligent monitoring device is provided with a pressure matrix composed of multiple pressure sensing units and multiple airbags, and each airbag deployment area corresponds to a pressure sub-matrix composed of corresponding pressure sensing units. The method includes:

[0007] Obtaining a pressure signal generated by the pressure matrix, and determining a region where the monitored subject is located and a sleeping position of the monitored subject based on the pressure signal, wherein the region where the monitored subject is located includes a chest and abdomen region of the monitored subject;

[0008] determining a contact matrix between the chest and abdomen area of ​​the monitored subject and each airbag, and determining a contact airbag based on the contact matrix;

[0009] Determining the pressure baseline value and the respiratory signal amplitude of each contact airbag;

[0010] Determine the pressure mean of the pressure submatrix corresponding to each of the contact airbags;

[0011] Based on the sleeping posture of the monitored subject, the contact matrix, the pressure mean, the pressure baseline value and the respiratory signal amplitude, the respiratory depth corresponding to each contact airbag is obtained, and the respiratory depth corresponding to each contact airbag is weighted and summed to obtain the final respiratory depth.

[0012] In one possible implementation, determining a contact matrix between the chest and abdomen region of the monitored subject and each airbag includes:

[0013] Determining whether the number of chest and abdominal regions of the monitored subject is greater than a preset number;

[0014] If yes, set all contact matrices to default values;

[0015] If not, obtaining the overlapping area between the airbag deployment area of ​​each contact airbag and the chest and abdomen area of ​​the monitored object;

[0016] The pressure sensing units corresponding to the overlapping area and the pressure sensing units corresponding to the non-overlapping area are respectively encoded into corresponding values ​​to obtain the contact matrix.

[0017] In one possible implementation, determining the pressure baseline value and the respiratory signal amplitude of each contact airbag includes:

[0018] Acquiring an airbag pressure signal generated by each of the contact airbags;

[0019] Performing low-pass filtering on the airbag pressure signal to obtain the pressure baseline signal;

[0020] performing band-pass filtering on the airbag pressure signal to obtain a breathing signal;

[0021] Based on the air pressure baseline signal, obtaining the air pressure baseline signal value;

[0022] The respiratory signal amplitude is obtained based on the respiratory signal.

[0023] In one possible implementation, obtaining the respiratory signal amplitude based on the respiratory signal includes:

[0024] Performing minimum and maximum point detection on the respiratory signal to obtain effective maximum and effective minimum points;

[0025] Determining effective peak points and effective trough points based on the effective maximum point and the effective minimum point;

[0026] The respiratory signal amplitude is obtained based on the effective peak point and the effective trough point.

[0027] In one possible implementation, obtaining the respiratory depth corresponding to each contact airbag based on the monitored subject's sleeping posture, contact matrix, pressure mean, pressure baseline value, and respiratory signal amplitude includes:

[0028] Combining the sleeping posture, contact matrix, pressure mean, air pressure baseline value, and respiratory signal amplitude of the monitored subject into a feature array;

[0029] performing a flattening operation on the feature array;

[0030] The flattened feature array is input into a preset breathing depth prediction model for prediction processing to obtain the breathing depth corresponding to the contact airbag.

[0031] In one possible implementation, performing weighted summation on the breathing depths corresponding to the contact airbags to obtain a final breathing depth includes:

[0032] Determining the distance between the center point of the chest and abdomen area of ​​the monitored subject and the center point of each of the contact airbag deployment areas;

[0033] Obtaining the contact state and contact area of ​​each of the contact airbags;

[0034] Determining a weighting coefficient corresponding to each of the contact airbags based on the distance, contact state, and contact area;

[0035] Based on the weighting coefficient, the breathing depth corresponding to each of the contact airbags is weighted accordingly to obtain the final breathing depth.

[0036] In one possible implementation, determining the area where the monitored subject is located and the sleeping position of the monitored subject based on the pressure signal includes:

[0037] determining whether the pressure signal is valid;

[0038] If the pressure signal is valid, performing a convolution operation on the pressure signal;

[0039] Perform region extraction on the pressure signal after the convolution operation to obtain the target area;

[0040] Performing a flattening operation on the target area, and performing a full connection operation on the target area after the flattening operation;

[0041] The target area after the full connection operation is activated by a preset activation function to obtain the sleeping posture of the monitored object, and the area where the monitored object is located is obtained by performing linear transformation weighting on the target area after the full connection operation.

[0042] In one possible implementation, determining whether the pressure signal is valid includes:

[0043] Determining whether the pressure signal corresponds to a pressure signal generated when the monitored subject is on the intelligent monitoring device; and / or

[0044] determining whether the pressure signal corresponds to a pressure signal generated under steady state conditions; and / or

[0045] Determine whether the signal quality of the pressure signal meets a preset signal quality condition.

[0046] In a second aspect, a breathing depth measurement device is provided for use in an intelligent monitoring device. The intelligent monitoring device is provided with a pressure matrix composed of multiple pressure sensing units and multiple airbags, and each airbag deployment area corresponds to a pressure sub-matrix composed of corresponding pressure sensing units. The device includes:

[0047] a prediction unit, configured to obtain a pressure signal generated by the pressure matrix, and determine a region where the monitored subject is located and a sleeping position of the monitored subject based on the pressure signal, wherein the region where the monitored subject is located includes a chest and abdomen region of the monitored subject;

[0048] a contact airbag determining unit, configured to determine a contact matrix between the chest and abdomen region of the monitored subject and each airbag, and determine the contact airbag based on the contact matrix;

[0049] An air pressure baseline value and a respiratory signal amplitude determination unit, configured to determine an air pressure baseline value and a respiratory signal amplitude of each of the contact airbags;

[0050] a pressure average value determining unit, configured to determine a pressure average value corresponding to each of the contact airbags;

[0051] The breathing depth determination unit is used to obtain the breathing depth corresponding to each of the contact airbags based on the sleeping posture of the monitored subject, the contact matrix, the pressure mean, the pressure baseline value and the breathing signal amplitude, and perform weighted summation on the breathing depth corresponding to each of the contact airbags to obtain the final breathing depth.

[0052] In a third aspect, a computer device is provided, comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the processor implements the above-mentioned breathing depth measurement method when executing the computer-readable instructions.

[0053] In a fourth aspect, a readable storage medium of computer-readable instructions is provided. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the above-mentioned breathing depth measurement method.

[0054] The above-mentioned breathing depth measurement method, device and medium, the method of which is implemented, includes: obtaining the pressure signal generated by the pressure matrix, and based on the pressure signal, determining the area where the monitored object is located and the sleeping position of the monitored object, the area where the monitored object is located includes the chest and abdomen area of ​​the monitored object; determining the contact matrix between the chest and abdomen area of ​​the monitored object and each airbag, and determining the contact airbag based on the contact matrix; determining the pressure baseline value and breathing signal amplitude of each contact airbag; determining the pressure mean of the pressure submatrix corresponding to each contact airbag; based on the sleeping position of the monitored object, the contact matrix, the pressure mean, the pressure baseline value and the breathing signal amplitude, obtaining the breathing depth corresponding to each contact airbag, and performing weighted summation on the breathing depth corresponding to each contact airbag to obtain the final breathing depth. In the embodiment of the present application, the combination of a matrix pressure sensing unit and multiple airbags can effectively reduce the influence of factors such as user position, sleeping position, weight and bedding on the accuracy of respiratory state measurement. The pressure sensing unit can accurately sense the pressure changes under different positions and sleeping positions, and provide comprehensive data support. With the help of parameters such as the average pressure of the pressure sensing unit above the airbag, the airbag pressure baseline value and the respiratory signal amplitude, the system can automatically calibrate the data to ensure that the measurement is not affected by the user's weight and differences in bedding. It provides stable and accurate respiratory monitoring services with high practical value for people with a history of disease and those pursuing health prevention. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0056] Figure 1 This is a schematic diagram of an implementation environment of a pressure sensing unit and an airbag deployment method of an intelligent monitoring device in an embodiment of the present application;

[0057] Figure 2 This is a flow chart of a method for measuring breathing depth in one embodiment of the present application;

[0058] Figure 3 This is a schematic diagram of a model structure of a multi-target detection model in one embodiment of the present application;

[0059] Figure 4 This is a schematic diagram of an implementation environment for a scene in which the chest and abdomen areas come into contact with an airbag in one embodiment of the present application;

[0060] Figure 5 This is a signal diagram of an airbag pressure signal in one embodiment of the present application;

[0061] Figure 6 This is a signal diagram of an air pressure baseline signal in one embodiment of the present application;

[0062] Figure 7 This is a signal diagram of a respiratory signal in one embodiment of the present application;

[0063] Figure 8 This is a signal diagram of a method for selecting peaks and troughs of a respiratory signal in one embodiment of the present application;

[0064] Figure 9 This is a schematic structural diagram of a breathing depth measurement device in one embodiment of the present application;

[0065] Figure 10 Schematic diagram of a computer device in one embodiment of the present application. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0067] In one embodiment, if Figure 2 As shown, a method for measuring breathing depth is provided, which is applied to an intelligent monitoring device. The intelligent monitoring device is provided with a pressure matrix composed of multiple pressure sensing units and multiple airbags, and each airbag deployment area corresponds to a pressure sub-matrix composed of corresponding pressure sensing units, including the following steps:

[0068] In step S110, a pressure signal generated by the pressure matrix is ​​obtained, and based on the pressure signal, the area where the monitored object is located and the sleeping position of the monitored object are determined, wherein the area where the monitored object is located includes the chest and abdomen area of ​​the monitored object;

[0069] It should be noted that the intelligent monitoring device can be a piece of intelligent furniture, such as a smart bed, smart mattress, or sofa, for users to sit, lie down, or rest. It can be equipped with multiple pressure-sensing units arranged in an array, such as a rectangular array, a circular array, or a hexagonal array. This arrangement can evenly sense pressure changes at different locations on the surface of the intelligent monitoring device, such as a mattress. Regardless of whether the monitored subject is lying on their back, side, or stomach, the pressure distribution of various body parts on the mattress can be accurately captured.

[0070] The pressure sensing unit is a component that can sense pressure changes and convert them into electrical signals or other detectable signals, such as a pressure sensor.

[0071] It should be noted that the intelligent monitoring device is also equipped with multiple airbags distributed throughout the device. These airbags can be connected to pressure sensors to monitor the air pressure within them. When the user is in any position within the intelligent monitoring device, the corresponding pressure sensor can monitor changes in air pressure within the corresponding airbag. It is understood that the pressure sensing unit can be positioned above the airbag, and multiple pressure sensing units can be deployed in a single airbag deployment area. That is, each airbag can correspond to a pressure submatrix composed of multiple pressure sensing units. This design enables precise monitoring of the pressure within each airbag area. When a particular airbag experiences a change in pressure, the corresponding pressure submatrix can quickly and accurately sense this change and convert it into data such as electrical signals for transmission and analysis. For example, in a smart mattress, when a user's body part presses against an area corresponding to an airbag, the corresponding pressure submatrix can accurately measure information such as the pressure level and change trend. By analyzing this data, the user's stress level in that area can be inferred, providing data support for analyzing the user's sleeping posture and micro-movements during breathing.

[0072] like Figure 1As shown, a schematic diagram of the deployment of pressure sensing units and airbags in an intelligent monitoring device is provided. S11 is a pressure matrix, which can be 64*32, with 32 pressure sensors per row, for a total of 64 rows, and a sampling accuracy of 0.01 kPa. S12-S17 are airbags, which can be connected to air pressure sensors with a range of 0-40 kPa and a sampling accuracy of 0.5 Pa. Taking the intelligent monitoring device as a double mattress and the monitored object as an example, the airbags can be deployed based on the supine position of the person. S12 can be the location of the airbag on the left shoulder of the user, S13 the location of the airbag on the left waist of the user, S14 the location of the airbag on the left hip of the user, S15 the location of the airbag on the right shoulder of the user, S16 the location of the airbag on the right waist of the user, and S17 the location of the airbag on the right hip of the user. Deploying S12-S17 in this manner effectively senses changes in airbag pressure in the chest and abdomen of the person, thereby better measuring breathing depth. It should be noted that when the intelligent monitoring equipment is different and the monitoring objects are different, the deployment methods of the airbag and the air pressure matrix are different, and they can be deployed accordingly according to actual needs.

[0073] Specifically, after the pressure signal is generated through the pressure matrix, the pressure signal can be tested for stability and quality to see if it is the pressure signal corresponding to when the user is on the smart monitoring device, in order to determine whether the pressure signal is valid. The stability and quality testing of the pressure signal can eliminate invalid or interfering signals, ensuring that the data input to the model is true and reliable, thereby improving the accuracy of the prediction of sleeping posture and area. If the pressure signal is valid, the pressure signal can be input into the pre-trained prediction model to predict the area where the monitoring object is located and the sleeping posture, wherein the sleeping posture may include supine, prone, side, and other sleeping postures, and the area where the monitoring object is located may include the horizontal coordinate of the center of the chest and abdomen, the vertical coordinate of the center of the chest and abdomen, the width of the chest and abdomen window, the height of the chest and abdomen window, and the body inclination angle. Therefore, the multi-target detection model that has been trained can finally output the sleeping posture probabilities corresponding to the above four sleeping postures and five types of position information.

[0074] In step S120, a contact matrix between the chest and abdomen area of ​​the monitored object and each airbag is determined, so as to determine the contact airbag based on the contact matrix;

[0075] Specifically, after determining the sleeping posture of the monitored object, the sleeping posture with the highest probability can be selected as the target sleeping posture from the various sleeping posture probabilities, and the target sleeping posture area can be determined. If the number of target sleeping posture areas is less than 3, it means that the number of people detected to be on the smart monitoring device at the same time is no more than 2. At this time, the overlapping area between the chest and abdomen area of ​​the monitored object and the airbag deployment area can be determined, and the pressure sensing units located in the overlapping area can be determined, and the pressure sensing units in the overlapping area and the pressure sensing units in the non-overlapping area can be encoded as different values. Each airbag corresponds to a pressure submatrix composed of multiple pressure sensing units. After assigning values ​​to these pressure submatrices according to the above rules, the contact matrix corresponding to each airbag can be obtained. If the number of people detected to be on the smart monitoring device at the same time exceeds 2, the values ​​of all pressure sensing units in the contact matrix can be set to 0 to reduce the impact of multi-person scenarios on measurement accuracy.

[0076] It should be noted that a contact airbag refers to an airbag that is in contact with the monitored object, such as a human body. If the contact matrix contains corresponding values ​​for overlapping areas, the airbag is considered a contact airbag; if the contact matrix contains only corresponding values ​​for non-overlapping areas, the airbag is considered a non-contact airbag.

[0077] In step S130, the pressure baseline value and the respiratory signal amplitude of each contact airbag are determined;

[0078] The air pressure baseline is a reference value for low-frequency variations, such as the average air pressure within the airbag when the subject is exposed to it for a long time on a smart monitoring device. The respiratory signal amplitude refers to the strength or range of variation of the respiratory signal. A larger amplitude may indicate deeper breathing or greater body movement during breathing, while a smaller amplitude may indicate shallower or more steady breathing. By analyzing changes in the respiratory signal amplitude, we can determine whether the user's breathing pattern is normal and whether there are any abnormalities (such as rapid breathing or shallow breathing).

[0079] Optionally, the airbag pressure signals generated by each contact airbag can be filtered using a preset filtering method to obtain a pressure baseline signal and a respiration signal. For example, the airbag pressure signal can be low-pass filtered to obtain the pressure baseline signal, and the airbag pressure signal can be band-pass filtered to obtain the respiration signal. The mean of the pressure baseline signal within a preset window time can then be calculated as the pressure baseline value. Simultaneously, the difference between the effective peak value and the effective trough value of the respiration signal within the preset window time can be calculated as the respiration signal amplitude.

[0080] It should be noted that after acquiring the signal, in order to ensure the real-time and accuracy of the data, key parameters need to be continuously updated. Therefore, the pressure baseline value and the respiratory signal amplitude can be updated using a second-by-second sliding window. For example, within each sliding window, the pressure baseline value and the respiratory signal amplitude are calculated or adjusted. For example, based on the pressure baseline signal and the respiratory signal within the current 1-second time window, the pressure baseline value and the respiratory signal amplitude are re-determined. As the sliding window continues to move, it is updated based on the new 1-second data each time. This can reflect the changes in the pressure baseline and respiratory signal amplitude over time in real time and track the dynamic characteristics of the signal in a timely manner.

[0081] In addition, low-pass filtering can be implemented by using filters such as IIR type Butterworth low-pass filter, Kalman low-pass filter, Gaussian low-pass filter, etc., and band-pass filtering can be implemented by using filters such as IIR type Butterworth band-pass filter, IIR type Chebyshev band-pass filter, finite impulse response (FIR) filter, etc.

[0082] In step S140, the pressure mean value of the pressure sub-matrix corresponding to each of the contact airbags is determined;

[0083] It should be noted that each contact airbag corresponds to a pressure submatrix. The pressure sensing cells in the pressure submatrix are assigned different values ​​when in contact with the monitored object and when in non-contact areas. For example, a pressure sensing cell in the contact area can be coded as 1, while a pressure sensing cell in the non-contact area can be coded as 0. Then, the pressure values ​​collected by the pressure sensing cells with non-zero values ​​are selected and averaged within a preset time window TH11, such as 10 seconds, to obtain the average pressure value of 0 corresponding to the contact airbag.

[0084] In step S150, based on the sleeping posture of the monitored subject, the contact matrix, the pressure mean, the pressure baseline value and the respiratory signal amplitude, the respiratory depth corresponding to each contact airbag is obtained, and the respiratory depth corresponding to each contact airbag is weighted and summed to obtain the final respiratory depth.

[0085] Specifically, different sleeping positions produce different respiratory amplitudes. For example, prone breathing produces stronger amplitude than supine breathing, which in turn produces stronger amplitude than side sleeping. The contact position and area of ​​the airbag also affect respiratory amplitude. For example, if the entire body is pressing against the airbag, the same respiratory amplitude will result in a stronger compression of the airbag than if only a portion of the body is compressing the airbag. The higher the air pressure within the airbag, the stiffer it is, and the smaller the amplitude of the pressure fluctuations caused by breathing. The mean pressure value can be used to preliminarily assess the weight load above the airbag. The greater the mean pressure value, the greater the load above the airbag, and the greater the breathing depth. The respiratory signal amplitude is a direct indicator of respiratory depth: the greater the respiratory signal amplitude, the greater the breathing depth. Therefore, by monitoring the subject's sleeping position, the contact matrix, the mean pressure value, the pressure baseline value, and the respiratory signal amplitude, and jointly determining respiratory depth, we can reduce interference factors and improve measurement accuracy.

[0086] Optionally, features such as sleeping posture, contact matrix, pressure mean, air pressure baseline value and respiratory signal amplitude of a specified group of people in various sleeping postures are collected to construct a training data set. At the same time, a breathing mask is worn to collect the changes in gas volume during the breathing process, and the average value of the changes in the amount of gas exhaled and inhaled per minute is used as the true breathing depth. The above features are then used as input to the breathing depth prediction model obtained by modeling with breathing depth as the target value for iterative training until the preset convergence conditions are met, and a trained breathing depth prediction model can be obtained. Then, the acquired sleeping posture, contact matrix, pressure mean, air pressure baseline value and respiratory signal amplitude of the monitored object can be input into the trained breathing depth prediction model, and the breathing depth corresponding to each contact airbag can be obtained. Then, the breathing depth corresponding to each contact airbag is weighted and summed to obtain the final breathing depth. For example, there are 3 contact airbags in total, and the breathing depths corresponding to each contact airbag are A, B, and C, respectively, and the weight coefficient is 、 、 , then the final breathing depth = A* +B* +C* .

[0087] In the embodiment of the present application, by combining a matrix pressure sensing unit with multiple airbag pressure sensors, the impact of different user positions on the accuracy of the measurement results can be effectively reduced. Regardless of the user's position, the pressure sensing unit can accurately sense pressure changes, ensure the reliability of respiratory status monitoring, and avoid monitoring errors caused by position differences. At the same time, different sleeping positions may cause different body pressure distributions during breathing, and the matrix pressure sensing unit can capture these changes in real time, effectively reducing the interference of the monitored subject's sleeping position on the measurement results, providing more comprehensive data support for accurately analyzing respiratory status. In addition, with the help of parameters such as the average pressure above the airbag, as well as the airbag pressure baseline value and the respiratory signal amplitude, the impact of factors such as user weight and other objects such as paving on the accuracy of the measurement results is successfully reduced. Even when facing users of different weights or when the paving varies, by using the baseline value, the system can automatically calibrate the measurement data to ensure that the respiratory monitoring results are not interfered with by these factors, thereby providing users with a more stable and accurate respiratory status monitoring service. Whether it is for users with a history of illness who need to pay close attention to respiratory health, or for the general public pursuing health prevention, it has important practical value.

[0088] In one embodiment of the present application, determining the area where the monitored object is located and the sleeping posture of the monitored object based on the pressure signal includes:

[0089] determining whether the pressure signal is valid;

[0090] If the pressure signal is valid, performing a convolution operation on the pressure signal;

[0091] Perform region extraction on the pressure signal after the convolution operation to obtain the target area;

[0092] Performing a flattening operation on the target area, and performing a full connection operation on the target area after the flattening operation;

[0093] The target area after the full connection operation is activated by a preset activation function to obtain the sleeping posture of the monitored object, and the area where the monitored object is located is obtained by performing linear transformation weighting on the target area after the full connection operation.

[0094] Optionally, after generating a pressure signal using the pressure matrix, the pressure signal can be tested for stability and quality to determine if it corresponds to the pressure signal corresponding to the user being on the smart monitoring device, thereby confirming its validity. If so, it can be input into a pre-set multi-target detection model to predict the sleeping position and the area where the monitored object is located.

[0095] The multi-target detection model can be Figure 3As shown, it may include an input layer S31, a first convolutional layer S32, a region extraction S33, a second convolutional layer S24, a first fully connected layer S35, a second fully connected layer S36 and an output layer S37.

[0096] The collected pressure signal is input into the input layer. The input tensor can be 64*32*1. It is then fed into the first convolutional layer for convolution processing, resulting in a tensor of 64*32*16. In this first convolutional layer, K3 represents a convolution kernel size of 3*3, s1 indicates a stride of 1, p1 indicates padded maxpooling, and c16 indicates the number of convolution kernels is 16. Region extraction is performed on the tensor obtained from the first convolutional layer. For example, this can be obtained using a fixed-size window and a single-step sliding window. The fixed window sizes include 8*4, 8*8, 8*16, 8*32, 16*4, 16*8, 16*16, 16*32, 32*4, 32*8, 32*16, 32*32, 64*4, 64*8, 64*16, and 64*32. The single-step sliding window moves only one data point horizontally or vertically each time the window is updated. The region-extracted tensor undergoes a 1x1 convolution, or flattening, operation to produce a one-dimensional array of size 16RC, where R is the fixed window row size and C is the fixed window column size. For an 8x4 fixed window, for example, R is 8 and C is 4, resulting in a one-dimensional array of size 512. A fully connected feature array of this flattened array is then fully connected to produce a one-dimensional array of size 1000. This 1000-dimensional array is then fully connected again to produce a one-dimensional array of size 500. Finally, the output layer outputs a one-dimensional vector of length 9, which includes four sleeping posture probabilities: supine, prone, sideways, and other sleeping postures, which can be obtained using an activation function such as softmax. Five additional probabilities are the horizontal and vertical coordinates of the center of the chest and abdomen, the width and height of the chest and abdomen window, and the body tilt angle, which are obtained by applying a weighted linear transformation to the previous fully connected layer.

[0097] It should be noted that the coordinate system of the above-mentioned horizontal and vertical coordinates can be established with the vertex in the lower left corner of the pressure matrix as the coordinate origin, the horizontal rightward direction as the positive direction of the horizontal axis, and the vertical upward direction as the positive direction of the vertical axis. The body inclination angle refers to the angle between the straight line connecting the head and feet of the person and the vertical upward direction. It should be pointed out that for different fixed window sizes, the S35-S37 model parameters are not shared, that is, the corresponding parameters are not the same. For each extracted area, there is a one-dimensional output of length 9, that is, for the original input of 64*32*1, there will be a total of 10080 (140*72, 59+49+33+1, 29+25+17+1) one-dimensional outputs of length 9.

[0098] In one embodiment of the present application, determining whether the pressure signal is valid includes:

[0099] Determining whether the pressure signal corresponds to a pressure signal generated when the monitored subject is on the intelligent monitoring device; and / or

[0100] determining whether the pressure signal corresponds to a pressure signal generated under steady state conditions; and / or

[0101] Determine whether the signal quality of the pressure signal meets a preset signal quality condition.

[0102] Alternatively, because the monitored subject may leave the intelligent monitoring device due to external factors, the pressure signal collected at this time cannot be used to determine the sleeping position and the area where the monitored subject is located. Therefore, it is possible to first determine whether the pressure signal corresponds to the pressure signal generated when the monitored subject is on the intelligent monitoring device. Taking a mattress as an example, a judgment of leaving the bed can be made based on the pressure signal to eliminate the pressure signal when the user is out of bed. In addition, a signal stability judgment can be made to eliminate unstable data generated when the user is not sleeping or is in an active state such as turning over or moving. In addition, the signal quality can be tested and weak signal data can be eliminated to prevent the signal from interfering with the final measurement results and affecting the measurement accuracy.

[0103] Among them, when judging whether the monitored object is on the said intelligent monitoring device, the sum of all pressure signals in the pressure matrix can be calculated. If the sum is greater than the preset value TH1; and the number of pressure values ​​in the pressure signal greater than the preset value TH2 is greater than TH3; and the preset window time length, such as the sum of pressure signals within 5s; it is band-pass filtered, such as [0.1, 2] Hz. If the sum of the absolute values ​​of the pressure signals after band-pass filtering is greater than the preset value TH4, and the above three conditions are met at the same time, it is considered that the monitored object is on the said intelligent monitoring device, otherwise it is considered that the monitored object is not on the said intelligent monitoring device.

[0104] It should be noted that TH1 is strongly correlated with sensor characteristics. Data from the matrix pressure sensor can be collected from a monitored subject on the smart monitoring device in a quiet environment. The current matrix pressure sensor signal is summed, and the lower quartile of the summed signal across all scenarios is calculated. TH1 is calculated by multiplying the lower quartile by a certain coefficient, such as 0.5. TH2 can be an empirical value. For example, a fixed size (e.g., 5cm*5cm) and weight (e.g., 250g) can be placed on different mattresses at different locations. The maximum pressure matrix values ​​in the placement areas are calculated, and the average of these maximum values ​​is calculated as TH2. TH3 can also be an empirical value. First, determine the coverage area of ​​a single sensor. For example, if the entire pressure sensor sheet covers an area of ​​150cm*160cm, the number of sensors is 32*32, and the coverage area of ​​a single sensor is 4.68cm*5cm. Then, the lower quartile of the contact area of ​​the subject lying on the smart monitoring device is multiplied by a certain coefficient, such as 0.5. Finally, the resulting area is divided by the coverage area of ​​a single sensor, and this value is rounded down to obtain the corresponding TH3. TH4 is strongly related to the sensor characteristics. It can collect matrix pressure sensor data in scenarios where different people are distributed in bed and there is no obvious body movement. The matrix pressure sensor signals at the current moment are summed, and then the signal sum is band-pass filtered at [0.1, 2] Hz. The sum of the absolute values ​​of the filtered signals within the window time is calculated. The window time can be 5 seconds, and the lower quartile of the sum of the absolute values ​​of all window signals is counted. Then TH4 is obtained by multiplying the lower quartile by a certain coefficient, such as 0.3.

[0105] To determine whether the pressure signal corresponds to a pressure signal generated in a stable state, the pressure signal may be summed to obtain a pressure signal sum value, which is then subjected to first-order difference processing and its absolute value taken. If this absolute value is not greater than a preset value TH5; and the absolute value of the difference between the number of pressure values ​​in the pressure signal greater than the preset value TH2 and the number of pressure values ​​greater than TH2 at the previous moment is not greater than TH6, the state is considered stable; otherwise, it is considered unstable.

[0106] It should be noted that TH5 is strongly correlated with sensor characteristics. Matrix pressure sensor data can be collected from different people in bed without noticeable movement. The matrix pressure sensor signals at the current moment are summed, and the upper quartile of the first-order difference of the sum of the absolute values ​​of the signals is calculated. TH5 is then calculated as the upper quartile multiplied by a certain coefficient, such as 1.5. TH6 can be an empirical value. This threshold is used to assess whether significant movement has occurred, that is, whether there has been a significant change in the contact area of ​​the pressure matrix between moments. It is calculated by dividing the allowable area change threshold by the coverage area of ​​a single sensor, rounded up. For example, if the expected effective contact area change (greater than TH2) is less than 25cm*25cm and the coverage area of ​​a single sensor is 4.68cm*5cm, the threshold is 27.

[0107] Among them, when determining whether the signal quality of the pressure signal meets the preset signal quality conditions, the [0.1, 2] Hz passband signal f1 within the window time length of 5s in the pressure matrix and the high-frequency component f2 with a cutoff frequency of 2 Hz in the pressure matrix can be extracted, and the ratio of the sum of the absolute values ​​of f1 to the absolute value of f2 is calculated. If the ratio is less than the preset value TH7, the signal quality is considered to be too low; otherwise, the signal quality is considered to meet the requirements.

[0108] It should be noted that TH7 may be an empirical value used to evaluate the signal-to-noise ratio and is related to the sensor, for example, 1.5.

[0109] In the embodiment of the present application, determining the contact matrix between the chest and abdomen area of ​​the monitored subject and each airbag includes:

[0110] Determining whether the number of chest and abdominal regions of the monitored subject is greater than a preset number;

[0111] If yes, set all contact matrices to default values;

[0112] If not, obtaining the overlapping area between the airbag deployment area of ​​each contact airbag and the chest and abdomen area of ​​the monitored object;

[0113] The pressure sensing units corresponding to the overlapping area and the pressure sensing units corresponding to the non-overlapping area are respectively encoded into corresponding values ​​to obtain the contact matrix.

[0114] Optionally, the sleeping posture probabilities corresponding to the extracted areas of the chest and abdomen of all the monitored subjects can be determined, and the maximum value of all sleeping posture probabilities in each extracted area can be taken. If the maximum sleeping posture probability of a certain extracted area is greater than the preset threshold TH8, the extracted area is used as a candidate area. Since there are multiple people on the intelligent monitoring device, if the number of candidate areas is greater than the preset number, such as 3, all contact matrices can be set to preset values, such as 0, to avoid the impact of multiple people on the measurement accuracy. If there are less than 3, the overlapping part of the airbag deployment position and the area where the user's chest and abdomen are located can be determined, and the pressure sensing unit of the overlapping part is set to 1, and the pressure sensing unit of the non-overlapping part is set to 0, so as to obtain the contact matrix of each airbag.

[0115] Among them, the TH8 experience value generally ranges from [0.75, 0.95], for example 0.8.

[0116] A coordinate system can be established based on the pressure matrix. For example, the horizontal and vertical coordinate systems can be based on the vertex in the lower left corner of the pressure matrix as the coordinate origin, with the horizontal axis pointing rightward and the vertical axis pointing upward. This coordinate system can then be used to determine the coordinates of the chest and abdomen regions and the airbag deployment zone. By comparing the positional relationships between the two at each coordinate point, it can be determined which areas fall within both the chest and abdomen and the airbag deployment zone. These areas are considered to be overlapping.

[0117] For example, take a single person on a smart monitoring device as an example. Figure 4 As shown, S11 is an array pressure sensor, with each small grid representing a sensor. S12-S17 are the six airbag distribution areas. S221 is the user's chest and abdomen area, S222 is the contact area between the user's chest and abdomen and the left shoulder airbag, S223 is the contact area between the user's chest and abdomen and the left waist airbag, S224 is the contact area between the user's chest and abdomen and the right shoulder airbag, and S225 is the contact area between the user's chest and abdomen and the right waist airbag. Each airbag position corresponds to an array pressure sensor in a fixed area. Assuming that the area where the left shoulder airbag is located is 3*3 in size, and the area where the user's chest and abdomen are located is only one area in the lower left corner, then its contact matrix is: The same applies to other airbags.

[0118] In one embodiment of the present application, determining the pressure baseline value and the respiratory signal amplitude of each contact airbag includes:

[0119] Acquiring an airbag pressure signal generated by each of the contact airbags;

[0120] Performing low-pass filtering on the airbag pressure signal to obtain the pressure baseline signal;

[0121] performing band-pass filtering on the airbag pressure signal to obtain a breathing signal;

[0122] Based on the air pressure baseline signal, obtaining the air pressure baseline signal value;

[0123] The respiratory signal amplitude is obtained based on the respiratory signal.

[0124] like Figure 5 As shown, a schematic diagram of the airbag pressure signal is provided. Figure 5 After low-pass filtering of the airbag pressure signal shown in Figure 6 The pressure baseline signal is shown in . Figure 5 After bandpass filtering of the airbag pressure signal in the Figure 7The respiratory signal shown in FIG. Here, low-pass filtering can be implemented using filters such as an IIR type Butterworth low-pass filter, a Kalman low-pass filter, and a Gaussian low-pass filter, and band-pass filtering can be implemented using filters such as an IIR type Butterworth band-pass filter, an IIR type Chebyshev band-pass filter, and a finite impulse response (FIR) filter.

[0125] Specifically, after obtaining the pressure baseline signal, the preset window time TH11 can be calculated, such as the mean value of the pressure baseline signal within 10 seconds, and the mean value of the pressure baseline signal can be used as the pressure baseline value. After obtaining the respiratory signal, the difference between the effective peak value and the effective trough value within the preset window time TH11 can be calculated, and the difference can be used as the respiratory signal amplitude, wherein the effective peak value refers to the value corresponding to the peak that is above a certain amplitude threshold and whose duration meets specific conditions, such as the effective peak value is greater than 0.005KPa and lasts for 0.2 seconds, and the same applies to the effective trough value. The preset window time TH11 is set to 1 second. Based on this, the pressure baseline value and the respiratory signal amplitude are updated using a second-by-second sliding window. For example, in each sliding window, the pressure baseline value and the respiratory signal amplitude are calculated or adjusted. Based on the pressure baseline signal and the respiratory signal within the current 1-second time window, the pressure baseline value and the respiratory signal amplitude are re-determined. As the sliding window continues to move, it is updated based on new 1-second data each time. This can reflect the changes in the pressure baseline and respiratory signal amplitude over time in real time and track the dynamic characteristics of the signal in a timely manner.

[0126] In an embodiment of the present application, obtaining the respiratory signal amplitude based on the respiratory signal includes:

[0127] Performing minimum and maximum point detection on the respiratory signal to obtain effective maximum and effective minimum points;

[0128] Determining effective peak points and effective trough points based on the effective maximum point and the effective minimum point;

[0129] The respiratory signal amplitude is obtained based on the effective peak point and the effective trough point.

[0130] Alternatively, as Figure 8The figure shows a schematic diagram of detecting effective peaks and effective troughs of a respiratory signal. Minimum points are detected for the respiratory signal. If a minimum point is detected and the corresponding minimum value is less than a preset threshold value TH9, the minimum point flag can be set to a preset value, such as 1, and the cached effective minimum points and preset threshold value TH9 are updated. Similarly, maximum point detection can be performed on the respiratory signal. If a maximum point is detected and the corresponding maximum value is greater than a preset threshold value TH10, and a valid minimum point (minimum point flag is 1) occurs before the maximum point, the point is considered a valid peak point, the cached effective maximum value and TH10 are updated, and the valid minimum point flag is set to 0.

[0131] In addition, when performing effective trough point detection, the respiratory signal can first be tested for maximum points. If a maximum point is detected, and the corresponding maximum value is greater than the preset threshold TH10, the maximum point flag can be set to a preset value, such as 0, and the cached effective maximum value and TH10 are updated. Similarly, the respiratory signal can be tested for minimum points. If a minimum point is detected, and the corresponding minimum value is less than the preset threshold TH9, and a valid maximum point appears before the minimum point (the valid maximum point flag is 0), then it is considered that a valid trough point has occurred, and the cached effective minimum point and the preset threshold TH9 are updated. At the same time, the valid maximum point flag is set to 1.

[0132] TH9 is strongly correlated with sensor characteristics. Initial TH9 is statistically derived. Matrix pressure sensor data from different people in bed with no noticeable movement is collected. The currently acquired respiratory signal is extracted, the respiratory signal minimum is determined, and the upper quartile of the minimum value is calculated. TH9 is calculated by multiplying the upper quartile by a certain coefficient, such as 0.5. Subsequent updates to the TH9 are performed using the minimum value cache array. TH10 is also strongly correlated with sensor characteristics. Matrix pressure sensor data from different people in bed with no noticeable movement is collected. The currently acquired respiratory signal maximum is extracted, and the lower quartile of the maximum value is calculated. TH10 can be calculated by multiplying the lower quartile by a certain coefficient, such as 0.5. Subsequent updates to the TH10 are performed using the maximum value cache array.

[0133] In one embodiment of the present application, the breathing depth corresponding to each contact airbag is obtained based on the sleeping posture of the monitored subject, the contact matrix, the pressure mean, the pressure baseline value, and the respiratory signal amplitude, including:

[0134] Combining the sleeping posture, contact matrix, pressure mean, air pressure baseline value, and respiratory signal amplitude of the monitored subject into a feature array;

[0135] performing a flattening operation on the feature array;

[0136] The flattened feature array is input into a preset breathing depth prediction model for prediction processing to obtain the breathing depth corresponding to the contact airbag.

[0137] Optionally, since there are several different sleeping positions (such as supine, side, and prone), each position has a corresponding probability value, indicating the likelihood of the user being in that position. These probability values ​​form a probability distribution vector. For example, if there are three sleeping positions (supine, side, and prone), with probabilities of 0.3, 0.6, and 0.1, respectively, then the sleeping position probability vector is [0.3, 0.6, 0.1]. The contact matrix is ​​a two-dimensional array that can be expanded row-wise or column-wise to obtain a corresponding one-dimensional array. The pressure matrix mean, air pressure baseline value, and respiratory signal amplitude are all specific numbers. In this case, a feature array can be directly generated by placing these numbers into an array of length k, where k represents the number of contact air cells. The processed sleeping position probability vector (i.e., the array consisting of the probabilities of each sleeping position), the expanded contact matrix array, the pressure matrix mean array, the air pressure baseline value array, and the respiratory signal amplitude array are then concatenated in sequence. The one-dimensional array is input into a pre-trained preset breathing depth prediction model. The preset breathing depth prediction model may include two fully connected layers. The first fully connected layer receives the flattened one-dimensional array as input, extracts and transforms the input data, and introduces nonlinearity through activation functions (such as ReLU, etc.), so that the model can learn more complex relationships. The second fully connected layer further processes and integrates the features and finally outputs a value, which is the breathing depth of the current airbag. , where i represents the i-th airbag.

[0138] In one embodiment of the present application, performing weighted summation on the breathing depths corresponding to the contact airbags to obtain the final breathing depth includes:

[0139] Determining the distance between the center point of the chest and abdomen area of ​​the monitored subject and the center point of each of the contact airbag deployment areas;

[0140] Obtaining the contact state and contact area of ​​each of the contact airbags;

[0141] Determining a weighting coefficient corresponding to each of the contact airbags based on the distance, contact state, and contact area;

[0142] Based on the weighting coefficient, the breathing depth corresponding to each of the contact airbags is weighted accordingly to obtain the final breathing depth.

[0143] Optionally, a coordinate system can be established based on the pressure matrix. For example, the coordinate system of the horizontal and vertical coordinates can be based on the vertex in the lower left corner of the pressure matrix as the coordinate origin, with the horizontal right direction being the positive direction of the horizontal axis and the vertical upward direction being the positive direction of the vertical axis. After obtaining the user's chest and abdomen area, the coordinates of the center point of the user's chest and abdomen area can be determined, such as , and the coordinates of the center points of each contact airbag deployment area, such as , where i represents the i-th contact airbag, and then the distance between the two is calculated based on their coordinates. .

[0144] Then, the contact status of each contact airbag can be obtained , if the contact matrix has a non-zero value and only one user's chest and abdomen cover the airbag, then is 1, otherwise it is 0. The contact area of ​​each airbag can also be obtained , the contact area is the sum of the number of non-zero values ​​in the contact matrix, and the breathing depth can be expressed as:

[0145] ;

[0146] in, The weighting coefficient for each airbag can be calculated as follows:

[0147] ;

[0148] in, For the above contact state, is the contact area of ​​each airbag mentioned above, For the above distance, It is a fixed preset value.

[0149] In the embodiment of the present application, by combining a matrix pressure sensing unit with multiple airbag pressure sensors, the impact of different user positions on the accuracy of the measurement results can be effectively reduced. Regardless of the user's position, the pressure sensing unit can accurately sense pressure changes, ensure the reliability of respiratory status monitoring, and avoid monitoring errors caused by position differences. At the same time, different sleeping positions may cause different body pressure distributions during breathing, and the matrix pressure sensing unit can capture these changes in real time, effectively reducing the interference of the monitored subject's sleeping position on the measurement results, providing more comprehensive data support for accurately analyzing respiratory status. In addition, with the help of parameters such as the average pressure above the airbag, as well as the airbag pressure baseline value and the respiratory signal amplitude, the impact of factors such as user weight and other objects such as paving on the accuracy of the measurement results is successfully reduced. Even when facing users of different weights or when the paving varies, by using the baseline value, the system can automatically calibrate the measurement data to ensure that the respiratory monitoring results are not interfered with by these factors, thereby providing users with a more stable and accurate respiratory status monitoring service. Whether it is for users with a history of illness who need to pay close attention to respiratory health, or for the general public pursuing health prevention, it has important practical value.

[0150] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0151] In one embodiment, a breathing depth measurement device is provided for use in an intelligent monitoring device. The intelligent monitoring device is provided with a pressure matrix composed of multiple pressure sensing units and multiple airbags, and each airbag deployment area corresponds to a pressure sub-matrix composed of corresponding pressure sensing units. The breathing depth measurement device corresponds one-to-one with the breathing depth measurement method in the above embodiment. Figure 9 As shown, the breathing depth measurement device includes a prediction unit 10, a contact airbag determination unit 20, an air pressure baseline value and breathing signal amplitude determination unit 30, a pressure mean value determination unit 40, and a breathing depth determination unit 50. The functional modules are described in detail as follows:

[0152] The prediction unit 10 is configured to obtain a pressure signal generated by the pressure matrix and determine a region where the monitored subject is located and a sleeping position of the monitored subject based on the pressure signal, wherein the region where the monitored subject is located includes a chest and abdomen region of the monitored subject;

[0153] a contact airbag determining unit 20, configured to determine a contact matrix between the chest and abdomen region of the monitored subject and each airbag, and to determine a contact airbag based on the contact matrix;

[0154] An air pressure baseline value and respiratory signal amplitude determination unit 30, configured to determine the air pressure baseline value and respiratory signal amplitude of each of the contact airbags;

[0155] a pressure average value determining unit 40, configured to determine a pressure average value corresponding to each of the contact airbags;

[0156] The breathing depth determination unit 50 is used to obtain the breathing depth corresponding to each of the contact airbags based on the sleeping posture of the monitored subject, the contact matrix, the pressure mean, the pressure baseline value and the breathing signal amplitude, and perform weighted summation on the breathing depths corresponding to each of the contact airbags to obtain the final breathing depth.

[0157] In one embodiment of the present application, the airbag contact determination unit 20 is further configured to:

[0158] Determining whether the number of chest and abdominal regions of the monitored subject is greater than a preset number;

[0159] If yes, set all contact matrices to default values;

[0160] If not, obtaining the overlapping area between the airbag deployment area of ​​each contact airbag and the chest and abdomen area of ​​the monitored object;

[0161] The pressure sensing units corresponding to the overlapping area and the pressure sensing units corresponding to the non-overlapping area are respectively encoded into corresponding values ​​to obtain the contact matrix.

[0162] In one embodiment of the present application, the air pressure baseline value and respiratory signal amplitude determination unit 30 is further configured to:

[0163] Acquiring an airbag pressure signal generated by each of the contact airbags;

[0164] Performing low-pass filtering on the airbag pressure signal to obtain the pressure baseline signal;

[0165] performing band-pass filtering on the airbag pressure signal to obtain a breathing signal;

[0166] Based on the air pressure baseline signal, obtaining the air pressure baseline signal value;

[0167] The respiratory signal amplitude is obtained based on the respiratory signal.

[0168] In one embodiment of the present application, the air pressure baseline value and respiratory signal amplitude determination unit 30 is further configured to:

[0169] Performing minimum and maximum point detection on the respiratory signal to obtain effective maximum and effective minimum points;

[0170] Determining effective peak points and effective trough points based on the effective maximum point and the effective minimum point;

[0171] The respiratory signal amplitude is obtained based on the effective peak point and the effective trough point.

[0172] In one embodiment of the present application, the breathing depth determination unit 50 is further configured to:

[0173] Combining the sleeping posture, contact matrix, pressure mean, air pressure baseline value, and respiratory signal amplitude of the monitored subject into a feature array;

[0174] performing a flattening operation on the feature array;

[0175] The flattened feature array is input into a preset breathing depth prediction model for prediction processing to obtain the breathing depth corresponding to the contact airbag.

[0176] In one embodiment of the present application, the breathing depth determination unit 50 is further configured to:

[0177] Determining the distance between the center point of the chest and abdomen area of ​​the monitored subject and the center point of each of the contact airbag deployment areas;

[0178] Obtaining the contact state and contact area of ​​each of the contact airbags;

[0179] Determining a weighting coefficient corresponding to each of the contact airbags based on the distance, contact state, and contact area;

[0180] Based on the weighting coefficient, the breathing depth corresponding to each of the contact airbags is weighted accordingly to obtain the final breathing depth.

[0181] In one embodiment of the present application, the prediction unit 10 is further configured to:

[0182] determining whether the pressure signal is valid;

[0183] If the pressure signal is valid, performing a convolution operation on the pressure signal;

[0184] Perform region extraction on the pressure signal after the convolution operation to obtain the target area;

[0185] Performing a flattening operation on the target area, and performing a full connection operation on the target area after the flattening operation;

[0186] The target area after the full connection operation is activated by a preset activation function to obtain the sleeping posture of the monitored object, and the area where the monitored object is located is obtained by performing linear transformation weighting on the target area after the full connection operation.

[0187] In one embodiment of the present application, the prediction unit 10 is further configured to:

[0188] Determining whether the pressure signal corresponds to a pressure signal generated when the monitored subject is on the intelligent monitoring device; and / or

[0189] determining whether the pressure signal corresponds to a pressure signal generated under steady state conditions; and / or

[0190] Determine whether the signal quality of the pressure signal meets a preset signal quality condition.

[0191] In the embodiment of the present application, by combining a matrix pressure sensing unit with multiple airbag pressure sensors, the impact of different user positions on the accuracy of the measurement results can be effectively reduced. Regardless of the user's position, the pressure sensing unit can accurately sense pressure changes, ensure the reliability of respiratory status monitoring, and avoid monitoring errors caused by position differences. At the same time, different sleeping positions may cause different body pressure distributions during breathing, and the matrix pressure sensing unit can capture these changes in real time, effectively reducing the interference of the monitored subject's sleeping position on the measurement results, providing more comprehensive data support for accurately analyzing respiratory status. In addition, with the help of parameters such as the average pressure above the airbag, as well as the airbag pressure baseline value and the respiratory signal amplitude, the impact of factors such as user weight and other objects such as paving on the accuracy of the measurement results is successfully reduced. Even when facing users of different weights or when the paving varies, by using the baseline value, the system can automatically calibrate the measurement data to ensure that the respiratory monitoring results are not interfered with by these factors, thereby providing users with a more stable and accurate respiratory status monitoring service. Whether it is for users with a history of illness who need to pay close attention to respiratory health, or for the general public pursuing health prevention, it has important practical value.

[0192] The specific definitions of the respiratory depth measurement device can be found in the definitions of the respiratory depth measurement method above and will not be repeated here. Each module in the aforementioned respiratory depth measurement device may be implemented in whole or in part via software, hardware, or a combination thereof. Each of these modules may be embedded in or independent of a processor in a computer device in hardware form, or may be stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0193] In one embodiment, a computer device is provided. The computer device may be a terminal device, and its internal structure diagram may be as follows: Figure 10As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a readable storage medium. The readable storage medium stores computer-readable instructions. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer-readable instructions are executed by the processor, a method for measuring respiratory depth is implemented. The readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0194] In an embodiment of the present application, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, the steps of the above-mentioned breathing depth measurement method are implemented.

[0195] In an embodiment of the application, a readable storage medium is provided, which stores computer-readable instructions. When the computer-readable instructions are executed by a processor, the steps of the above-mentioned breathing depth measurement method are implemented.

[0196] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware through computer-readable instructions. The computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0197] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0198] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for measuring breathing depth, characterized in that: Applied to an intelligent monitoring device, the intelligent monitoring device is provided with a pressure matrix composed of multiple pressure sensing units and multiple airbags, and each airbag deployment area corresponds to a pressure sub-matrix composed of corresponding pressure sensing units. The method includes: Obtaining a pressure signal generated by the pressure matrix, and determining a region where the monitored subject is located and a sleeping position of the monitored subject based on the pressure signal, wherein the region where the monitored subject is located includes a chest and abdomen region of the monitored subject; determining a contact matrix between the chest and abdomen area of ​​the monitored subject and each airbag, and determining a contact airbag based on the contact matrix; Determining an air pressure baseline value and a respiratory signal amplitude of each of the contact airbags, wherein the air pressure baseline value is obtained by performing low-pass filtering on the airbag pressure signal of each contact airbag to obtain an air pressure baseline signal, and the respiratory signal amplitude is obtained by performing band-pass filtering on the airbag pressure signal of each contact airbag to obtain a respiratory signal; Determine the pressure mean of the pressure submatrix corresponding to each of the contact airbags; Inputting the sleeping posture, contact matrix, pressure mean, air pressure baseline value, and respiratory signal amplitude of the monitored subject into a preset respiratory depth prediction model to obtain the respiratory depth corresponding to each contact airbag, and performing weighted summation on the respiratory depth corresponding to each contact airbag to obtain the final respiratory depth; Among them, the preset breathing depth prediction model is obtained by training a training data set, and the training data set is composed of the sleeping posture, contact matrix, pressure mean, air pressure baseline value, breathing signal amplitude and the real breathing depth at the corresponding moment of a specified group of people in various sleeping postures. The real breathing depth refers to the average value of the change in the amount of exhaled and inhaled gas per minute obtained by collecting the change in gas volume during the breathing process when the specified group of people wears a breathing mask.

2. The breathing depth measurement method according to claim 1, wherein Determining the contact matrix between the chest and abdomen area of ​​the monitored subject and each airbag includes: Determining whether the number of chest and abdominal regions of the monitored subject is greater than a preset number; If yes, set all contact matrices to default values; If not, obtaining the overlapping area between the airbag deployment area of ​​each contact airbag and the chest and abdomen area of ​​the monitored object; The pressure sensing units corresponding to the overlapping area and the pressure sensing units corresponding to the non-overlapping area are respectively encoded into corresponding values ​​to obtain the contact matrix.

3. The breathing depth measurement method according to claim 1, wherein The respiratory signal amplitude is obtained as follows: Performing minimum and maximum point detection on the respiratory signal to obtain effective maximum and effective minimum points; Determining effective peak points and effective trough points based on the effective maximum point and the effective minimum point; The respiratory signal amplitude is obtained based on the effective peak point and the effective trough point.

4. The method for measuring breathing depth according to claim 1, wherein: The step of inputting the sleeping posture, contact matrix, pressure mean, air pressure baseline value, and respiratory signal amplitude of the monitored subject into a preset respiratory depth prediction model to obtain the respiratory depth corresponding to each contact airbag includes: Combining the sleeping posture, contact matrix, pressure mean, air pressure baseline value, and respiratory signal amplitude of the monitored subject into a feature array; performing a flattening operation on the feature array; The flattened feature array is input into a preset breathing depth prediction model for prediction processing to obtain the breathing depth corresponding to the contact airbag.

5. The method for measuring breathing depth according to claim 1, wherein: The weighted summation of the breathing depths corresponding to the contact airbags to obtain the final breathing depth includes: Determining the distance between the center point of the chest and abdomen area of ​​the monitored subject and the center point of each contact airbag deployment area; Obtaining the contact state and contact area of ​​each of the contact airbags; Determining a weighting coefficient corresponding to each of the contact airbags based on the distance, contact state, and contact area; Based on the weighting coefficient, the breathing depth corresponding to each of the contact airbags is weighted accordingly to obtain the final breathing depth.

6. The method for measuring breathing depth according to claim 1, wherein: The determining, based on the pressure signal, the area where the monitored object is located and the sleeping position of the monitored object includes: determining whether the pressure signal is valid; If the pressure signal is valid, performing a convolution operation on the pressure signal; Perform region extraction on the pressure signal after the convolution operation to obtain the target area; Performing a flattening operation on the target area, and performing a full connection operation on the target area after the flattening operation; The target area after the full connection operation is activated by a preset activation function to obtain the sleeping posture of the monitored object, and the area where the monitored object is located is obtained by performing linear transformation weighting on the target area after the full connection operation.

7. The method for measuring breathing depth according to claim 6, wherein: Determining whether the pressure signal is valid includes: Determining whether the pressure signal corresponds to a pressure signal generated when the monitored subject is on the intelligent monitoring device; and / or determining whether the pressure signal corresponds to a pressure signal generated under steady state conditions; and / or Determine whether the signal quality of the pressure signal meets a preset signal quality condition.

8. A breathing depth measuring device, characterized in that: Applied to intelligent monitoring equipment, the intelligent monitoring equipment is provided with a pressure matrix composed of multiple pressure sensing units and multiple airbags, and each airbag deployment area corresponds to a pressure sub-matrix composed of corresponding pressure sensing units. The device includes: a prediction unit, configured to obtain a pressure signal generated by the pressure matrix, and determine a region where the monitored subject is located and a sleeping position of the monitored subject based on the pressure signal, wherein the region where the monitored subject is located includes a chest and abdomen region of the monitored subject; a contact airbag determining unit, configured to determine a contact matrix between the chest and abdomen region of the monitored subject and each airbag, and determine the contact airbag based on the contact matrix; an air pressure baseline value and a breathing signal amplitude determination unit, configured to determine an air pressure baseline value and a breathing signal amplitude of each of the contact airbags, wherein the air pressure baseline value is obtained by performing low-pass filtering on the airbag air pressure signal of each contact airbag to obtain an air pressure baseline signal, and the breathing signal amplitude is obtained by performing band-pass filtering on the airbag air pressure signal of each contact airbag to obtain a breathing signal; a pressure average value determining unit, configured to determine a pressure average value corresponding to each of the contact airbags; a breathing depth determination unit, configured to input the monitored subject's sleeping posture, contact matrix, pressure mean, pressure baseline value, and breathing signal amplitude into a preset breathing depth prediction model, obtain the breathing depth corresponding to each contact airbag, and perform weighted summation of the breathing depths corresponding to each contact airbag to obtain a final breathing depth; Among them, the preset breathing depth prediction model is obtained through training with a training data set, and the training data set is composed of the sleeping posture, contact matrix, pressure mean, air pressure baseline value, breathing signal amplitude and the real breathing depth at the corresponding moment of a specified group of people in various sleeping postures. The real breathing depth refers to the average value of the change in the amount of exhaled and inhaled gas per minute obtained by collecting the change in gas volume during the breathing process when the specified group of people wears a breathing mask.

9. A readable storage medium storing computer-readable instructions, characterized in that: When the computer-readable instructions are executed by a processor, the steps of the breathing depth measurement method according to any one of claims 1 to 7 are implemented.

Images