Oronasal respiration data acquisition and analysis method and system

Abstract

The application discloses a kind of oral-nasal breathing data acquisition analysis method and system, it is related to medical monitoring technical field, for solving the problem that the adaptive threshold adjustment based on traditional image detection algorithm cannot be applicable to continuous repeating graphic mode, image feature is extracted using texture analysis method, based on image edge information is divided into multiple partial images, the oral-nasal breathing data anomaly rate of partial image, respiratory frequency variation intensity and texture complexity are collected, data analysis model is established, image evaluation coefficient is obtained and compared with image threshold, the number and proportion of important partial image are counted, the network state information obtained by network detection and the proportion of important part in image, the transmission duration is determined using fuzzy bayesian neural network, compared with preset transmission duration, according to the comparison result, the transmission image is adjusted, the limitation of traditional image detection algorithm is solved while reducing network resource consumption.

Description

Technical Field

[0001] This invention relates to the field of medical monitoring technology, and more specifically, to a method and system for collecting and analyzing oral and nasal breathing data. Background Technology

[0002] In recent years, with the development of medical technology and the increasing demand for health monitoring, oral and nasal breathing data acquisition and analysis systems have been widely used in clinical diagnosis, health monitoring and exercise physiology. Existing technologies can collect basic respiratory data such as respiratory rate, air flow rate and gas composition, and also have the function of storing and transmitting data, which can transmit data to PC or cloud server via wired or wireless means.

[0003] The existing technology has the following shortcomings:

[0004] Currently, wireless transmission mainly relies on network bandwidth to analyze image data collected by the nasal and oral respiration tubes and transmit it to a PC or cloud server. However, there are also factors such as network latency and signal interference. Existing image detection algorithms usually use the Otsu algorithm to obtain thresholds for threshold segmentation to solve the limitations of manually setting high and low thresholds in traditional algorithms. However, for nasal and oral respiration images that form repeating graphic patterns in the spectrogram, it is impossible to detect the degree of subtle changes, which leads to errors in data analysis and excessive consumption of network resources. Therefore, a method and system for nasal and oral respiration data acquisition and analysis is proposed.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a method and system for collecting and analyzing oral and nasal breathing data, which solves the problems mentioned in the background art by using different product inspection methods.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for collecting and analyzing oral and nasal breathing data, comprising S1: reducing noise interference by grayscale and filtering the oral and nasal breathing image, extracting features from the image using a texture analysis method, dividing the image into multiple parts using image edge information, collecting the parts of the image, obtaining the variation rate of oral and nasal breathing data, the intensity of respiratory frequency change, and texture complexity through an image edge detection algorithm, and simultaneously collecting network status information through network monitoring.

[0008] S2: Obtain the variation rate, respiratory frequency change intensity, and texture complexity of the oral and nasal breathing data of a portion of the image, establish a data analysis model, and obtain the image evaluation coefficients;

[0009] S3: Obtain the image evaluation coefficients and compare them with the image threshold to obtain the comparison results. Based on each comparison result and the number of comparison results, count the number of important parts of the image and their proportion.

[0010] S4; Obtain network status information and the proportion of important parts of the image within the image, and obtain the time-domain upload bandwidth and time-domain latency through data processing;

[0011] S5: Based on the temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image, a fuzzy Bayesian neural network is used to determine the transmission duration, which is then compared with the preset transmission duration. The transmitted image is then readjusted based on the comparison results to obtain the adjustment result.

[0012] In a preferred embodiment, specifically, the color image is converted to a grayscale image, the image is smoothed using a Gaussian filter G to reduce noise, and the smoothed image is divided into multiple image parts using the Canny edge detection algorithm. The specific steps are as follows:

[0013] Step A1: Apply the Canny edge detection algorithm to detect edges in the smoothed image;

[0014] Step A2: Detect contours from the edge detection results;

[0015] Step A3: For each contour, create a binary mask;

[0016] Step A4: Apply a mask to the original image to extract the corresponding portion of the image.

[0017] In a preferred embodiment, the original image is divided into multiple partial images, where c represents the c-th partial image;

[0018] By collecting inspiratory data below or above a high threshold and expiratory data below or above a low threshold, the frequency of these occurrences was obtained, and the ratio of this frequency to the total number of breaths was used to calculate the variability rate (Ar) of nasal and oropharyngeal breathing data. c ;

[0019] By defining a healthy respiratory cycle interval, respiratory rates greater than or less than this interval from a subset of images are collected. The absolute value of the difference between this value and the maximum or minimum value of the healthy respiratory cycle interval is used to obtain the relative deviation of a single respiratory rate. These deviations are then summed to obtain the intensity of the respiratory rate change, Dv. c ;

[0020] The texture complexity Te is calculated by extracting texture features using the gray-level co-occurrence matrix. c .

[0021] In a preferred embodiment, the variance rate of oral and nasal breathing data, the intensity of respiratory rate changes, and texture complexity are obtained; a data analysis model is established to obtain the image evaluation coefficient Pt. c The formula used is as follows:

[0022]

[0023] In the formula, Pt c For image evaluation coefficients, as well as These are preset scaling factors for texture complexity, the rate of variation in oral and nasal breathing data, and the intensity of changes in respiratory rate, respectively. as well as All are greater than 0.

[0024] In a preferred embodiment, image evaluation coefficients are obtained, and the image evaluation coefficients are compared and analyzed with iteratively increasing image thresholds;

[0025] If the image evaluation coefficient is greater than or equal to the image threshold, then that part of the image is marked as an important image and an important signal is generated;

[0026] If the image evaluation coefficient is less than the image threshold, then that part of the image is marked as an unimportant image and an end signal is generated;

[0027] Multiple image evaluation coefficient calculations will be performed to obtain the image evaluation coefficient values ​​of multiple parts of an image and two sets of comparison results of the multiple parts of the image. Based on the comparison results, the image evaluation coefficient values ​​of the parts of the image evaluated as important signals and their proportions will be statistically analyzed.

[0028] In a preferred embodiment, the network status information includes time-domain upload bandwidth and time-domain latency:

[0029] By dividing the time domain into different sizes and labeling each time domain as i, and further dividing it into n time domains, the time domain upload bandwidth B(T) is obtained based on the amount of data transmitted within each time domain and the duration of each time domain. i );

[0030] After dividing the time domain, the latency is measured using the ping command or other network measurement tools to obtain the time domain latency D(T). i );

[0031] The importance percentage (Ze) of a partial image is obtained by comparing the total number of partial images within the image with the number of important images.

[0032] In a preferred embodiment, a fuzzy Bayesian neural network is used to determine the transmission duration. The temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image are defined as input nodes in the input layer, and the transmission duration is defined as the output layer.

[0033] A fuzzy Bayesian neural network was developed to describe the impact of temporal upload bandwidth, temporal delay duration, and the proportion of image importance within an image on transmission duration.

[0034] Based on the fuzzy Bayesian neural network, the transmission duration is determined by the temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image.

[0035] In a preferred embodiment, after the transmission duration is determined, it is compared with the preset transmission duration. If it is less than the preset transmission duration, the important images in the image are transmitted to the PC. Otherwise, an alarm is issued and the amount of resources required for one transmission is adjusted.

[0036] A system for acquiring and analyzing oral and nasal breathing data includes an acquisition module, a processing module, an image analysis module, a resource allocation module, and an output terminal.

[0037] The acquisition module is used to acquire images, divide the image into multiple parts based on image edge information, acquire partial images, and obtain the anomaly rate, respiratory frequency change intensity, and texture complexity of oral and nasal breathing data through image edge detection algorithms, which are then sent to the processing module. The network status information is acquired through network monitoring and sent to the resource allocation module.

[0038] The processing module is used to acquire the variance rate, respiratory rate change intensity, and texture complexity of some oral and nasal breathing data, establish a data analysis model, obtain image evaluation coefficients, and send them to the image analysis module.

[0039] The image analysis module is used to obtain image evaluation coefficients and compare them with image thresholds to obtain comparison results. Based on each comparison result and the number of comparison results, the module counts the number and proportion of important parts of the image and sends them to the resource allocation module.

[0040] The resource allocation module is used to obtain network status information and the proportion of important parts of the image within the image. It obtains the transmission time through a fuzzy Bayesian neural network and compares it with the preset transmission time. Based on the comparison result, it readjusts the transmitted image and sends the adjusted result to the output end.

[0041] The output terminal is used to receive the adjustment results and report them.

[0042] The technical effects and advantages of this invention are as follows:

[0043] 1. This invention divides an image into multiple parts using image edge information, collects the texture complexity, variability rate of nasal and oral breathing data, and intensity of respiratory frequency changes in these parts, establishes a data analysis model to obtain image evaluation coefficients, and compares these coefficients with image thresholds to obtain comparison results. Based on each comparison result and the number of comparison results, the number and proportion of important parts of the image are statistically determined. This invention overcomes the limitations of traditional image detection algorithms in detecting nasal and oral breathing data images, reduces network resources occupied by the image while maintaining image quality, and improves the overall efficiency of image transmission.

[0044] 2. This invention obtains network status information and the proportion of important parts of an image, establishes a fuzzy Bayes algorithm based on time-domain upload bandwidth, time-domain delay duration, and the proportion of important parts of an image to determine the transmission duration, compares and analyzes the transmission result with the preset transmission duration, and adjusts the amount of resources required for each transmission. This dynamic adjustment of resources reduces the impact of network congestion on transmission and ensures the timely transmission and processing of critical image information. Attached Figure Description

[0045] Figure 1 This is a flowchart of a method for collecting and analyzing oral and nasal breathing data according to the present invention.

[0046] Figure 2 This is a schematic diagram of a nasal and oral respiration data acquisition and analysis system according to the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Example 1

[0049] Nasal and oral breathing data exhibit regularity, with the time interval between each inhalation and exhalation forming a recurring pattern in the spectrogram. Existing image edge detection methods segment regions with significant frequency intensity changes, but they cannot effectively detect subtle variations in recurring waveform images. Therefore, further refinement of image edge detection algorithms is needed to determine the significant proportion of nasal and oral breathing image data.

[0050] Please see Figure 1 A method for collecting and analyzing oral and nasal breathing data, the specific operation process is as follows:

[0051] S1: Reduce noise interference by grayscale and filtering the mouth and nose breathing images, extract features from the images using texture analysis, divide the images into multiple parts using image edge information, collect partial images, and obtain the variation rate of mouth and nose breathing data, the intensity of breathing frequency change, and texture complexity through image edge detection algorithm. At the same time, collect network status information through network monitoring.

[0052] The image edge information includes the variation rate of oral and nasal breathing data, the intensity of changes in breathing frequency, and texture complexity.

[0053] Among them, the image edge detection algorithm is used to identify regions in an image where the gray level changes drastically, and these regions correspond to the boundaries of objects in the image.

[0054] Specifically, the color image is converted to a grayscale image, and a Gaussian filter G is used to smooth the image and reduce noise. The formula is expressed as follows: Among them, I blur (x, y) is the pixel value at position (x, y) of the smoothed image, I gra y(x+u, y+v) is the pixel value of the grayscale image at position (x+u, y+v), G(u, k) is the weight value of the Gaussian filter at offset (u, k), and u, k is the offset range of the filter kernel, usually the filter radius.

[0055] The smoothed image is divided into multiple parts using the Canny edge detection algorithm. The specific steps are as follows:

[0056] Step A1: Apply the Canny edge detection algorithm to detect edges in the smoothed image. The formula is as follows: E(x, y) = Canny(I blur (x, y), T low T high ), where T low It is a low threshold used for edge connections; T high It uses a high threshold for strong edge detection;

[0057] Step A2: Detect contours from the edge detection results: C = findContours(E), where C is the set of detected contours;

[0058] Step A3: For each contour C∈C, create a binary mask M c :

[0059]

[0060] Among them, M c (x, y) is the mask value of the pixel at (x, y), which is 255 inside the outline and 0 outside the outline;

[0061] Step A4: Set the mask M c Applied to the original image I to extract the corresponding portion of image I c ;

[0062]

[0063] Among them, I c (x, y) represents the value at the pixel point of the extracted portion of the image, M c When (x, y) = 255, I c (x, y) = I(x, y), otherwise I c (x, y) = 0.

[0064] Through the above steps, the original image is divided into multiple partial images, where c represents the c-th partial image;

[0065] In step A1, the low and high thresholds are further defined. Since the oral and nasal breathing image data are relatively repetitive waveform images, the peaks and valleys that are repeated most frequently in the waveform images are analyzed and defined as the inhalation and exhalation phases. At the same time, the high threshold is defined based on the height of the peak and the low threshold is defined based on the depth of the valley. Therefore, the high and low thresholds of the Canny edge detection algorithm change with the user.

[0066] The logic for obtaining the variability rate of nasal and oropharyngeal breathing data involves collecting inspiratory data below or above a high threshold and expiratory data below or above a low threshold, obtaining the frequency of these occurrences, and then dividing this frequency by the total number of breaths to obtain the variability rate Ar of nasal and oropharyngeal breathing data. c .

[0067] The logic for obtaining the intensity of respiratory rate variation is as follows: First, a healthy respiratory cycle interval is defined. Then, respiratory rates in a subset of images that are greater than or less than this healthy interval are collected. The absolute value of the difference between this value and the maximum or minimum value of the healthy respiratory cycle interval is then compared with the difference between the maximum and minimum values ​​of the healthy respiratory cycle interval to obtain the relative deviation of a single respiratory rate. These relative deviations are then summed to obtain the intensity of the respiratory rate variation, Dv. c .

[0068] It should be noted that the healthy respiratory cycle interval is based on a cross-sectional study mentioned in BMC Public Health, which includes the relationship between breathing patterns and physical health during spontaneous breathing. The parameters of the breathing pattern mentioned include respiratory rate (RR), which refers to the number of breaths per minute.

[0069] If the respiratory rate is lower than the minimum value of the healthy respiratory cycle range, the respiratory rate is subtracted from the minimum value of the healthy respiratory cycle range to indicate the degree of rapid breathing; if the respiratory rate is higher than the maximum value of the healthy respiratory cycle range, the respiratory rate is subtracted from the maximum value of the healthy respiratory cycle range to indicate the degree of slow breathing.

[0070] The logic for obtaining texture complexity is to extract texture features from GLCM, use them to calculate texture complexity, and calculate image contrast, entropy, and energy according to the formula.

[0071] It should be noted that GLCM stands for Gray Co-occurrence Matrix. Specifically, GLCM describes the frequency distribution of the co-occurrence of pixel gray-level pairs at a given distance and direction, thus reflecting the texture features of the image. Each element P(i,j) represents the frequency of co-occurrence of a pixel pair with gray-level value i and gray-level value j at a specified distance d and direction θ. The size of the matrix is ​​L×L, where L is the number of gray levels in the image.

[0072] Wherein, the contrast formula is: It describes the degree of local variation in gray levels in an image;

[0073] The entropy formula is as follows: It measures the randomness of the image's grayscale distribution;

[0074] The energy formula is as follows: It describes the uniformity of the image's gray levels;

[0075] Specifically, the higher the contrast and entropy values, the higher the texture randomness and complexity; conversely, the higher the energy, the higher the texture uniformity and the lower the texture complexity.

[0076] Specifically, the texture complexity calculation formula is obtained through weighted average calculation, based on the following formula:

[0077] Te c =α1*Co c +α2*Et c -α3*En c

[0078] In the formula, Te c For texture complexity, α1, α2, and α3 are preset scaling factors for contrast, entropy, and energy, and α1, α2, and α3 are all greater than 0.

[0079] S2: Obtain the variation rate, respiratory frequency change intensity, and texture complexity of the oral and nasal breathing data of a portion of the image, establish a data analysis model, and obtain the image evaluation coefficients;

[0080] To obtain the variability rate, respiratory rate variation intensity, and texture complexity of oral and nasal breathing data, a data analysis model was established to obtain the image evaluation coefficient Pt. c The formula used is as follows:

[0081]

[0082] In the formula, Pt c For image evaluation coefficients, as well as These are preset scaling factors for texture complexity, the rate of variation in oral and nasal breathing data, and the intensity of changes in respiratory rate, respectively. as well as All are greater than 0.

[0083] Among them, texture complexity, the rate of variation of oral and nasal breathing data, and the intensity of respiratory frequency change are all data representations that directly express the image evaluation coefficient values ​​and the current transmission priority of some images.

[0084] As can be seen from the above formula, when the texture complexity, the rate of variation of oral and nasal breathing data and the intensity of respiratory frequency change are higher, it means that when some images are transmitted earlier, the image evaluation coefficient is larger. Conversely, when some images are transmitted later, the image evaluation coefficient is smaller.

[0085] S3: Obtain the image evaluation coefficients and compare them with the image threshold to obtain the comparison results. Based on each comparison result and the number of comparison results, count the number of important parts of the image and their proportion.

[0086] The logic for obtaining the image threshold is to collect a set of distributed samples from a historical partial image transmission database, then divide the dataset into a training set and a test set, set evaluation metrics and clustering algorithms, train the model on the training set in each iteration of cross-validation, evaluate the model performance on the test set, and then adjust the image threshold based on the performance of the validation set. Therefore, the image threshold is constantly iterated and updated.

[0087] In this invention, clustering algorithm is a type of unsupervised learning algorithm used to standardize all image feature data to ensure that each feature is compared on the same scale. K initial centroids are randomly selected, and each part of the image is assigned to the cluster containing the nearest centroid according to Euclidean distance. The new centroid of each cluster is calculated and used as the centroid of the next iteration. The iteration is repeated until the centroid no longer changes or the change is less than a set threshold, and the final image threshold is determined.

[0088] Obtain image evaluation coefficients and compare and analyze them with iteratively selected image thresholds.

[0089] If the image evaluation coefficient is greater than or equal to the image threshold, then that part of the image is marked as an important image and an important signal is generated;

[0090] If the image evaluation coefficient is less than the image threshold, then that part of the image is marked as an unimportant image and an end signal is generated;

[0091] Multiple image evaluation coefficient calculations will be performed to obtain image evaluation coefficient values ​​for multiple parts of an image and two sets of comparison results for the multiple parts of the image. Based on the comparison results, the image evaluation coefficient values ​​of the parts of the image evaluated as important signals and their proportions will be statistically analyzed.

[0092] This invention divides an image into multiple parts using image edge information, collects the texture complexity, variability rate of nasal and oral breathing data, and intensity of respiratory frequency changes in these parts, establishes a data analysis model to obtain image evaluation coefficients, and compares these coefficients with image thresholds to obtain comparison results. Based on each comparison result and the number of comparison results, the number and proportion of important parts of the image are statistically determined. This invention overcomes the limitations of traditional image detection algorithms in detecting nasal and oral breathing data images, reduces network resources occupied by the image while maintaining image quality, and improves the overall efficiency of image transmission.

[0093] Example 2

[0094] In Embodiment 1 of this invention, a key example is given of dividing an image into multiple parts using image edge information, collecting the texture complexity, nasal and oral breathing data variation rate, and respiratory frequency change intensity of these parts, establishing a data analysis model to obtain image evaluation coefficients, and comparing them with image thresholds to obtain comparison results. Based on each comparison result and the number of comparison results, the number and value of important parts of the image are statistically determined. However, Embodiment 1 only divides the image into multiple parts and obtains the image importance based on the proportion of importance of each part. Obviously, although this allows transmission based on image importance, it cannot transmit all important images when network resources are scarce or in scenarios with high immediacy. To address the above problems, Embodiment 2 of this invention further refines the process.

[0095] S4: Obtain network status information and the proportion of important parts of the image within the image, and obtain the time-domain upload bandwidth and time-domain latency through data processing;

[0096] The network status information includes time-domain upload bandwidth and time-domain latency.

[0097] The logic for obtaining time-domain upload bandwidth is to divide the time domain into different sizes. For peak traffic periods or periods of high network usage, time domains are planned, labeled as i based on peak and off-peak periods, and then divided into n time domains, according to the formula: Wherein, Data Size(T) i Time(T) represents the amount of data transmitted in the i-th time domain. i Let B(T) be the duration of the i-th time domain. i This refers to the upload bandwidth in the time domain.

[0098] It should be noted that the analysis of bandwidth usage patterns is based on the formula: If B(T) i )>μ b Then it is the peak period, if B(T) i )≤μ b This is the off-peak period; the number of time-domain divisions is determined by the researchers based on the specific implementation situation and is not limited here.

[0099] The logic for obtaining time-domain latency is as follows: after dividing the time domain, the latency is measured using the ping command or other network measurement tools, based on the formula... Among them, D(T) i ) represents the time period T i The time-domain delay duration within, For time period T i The delay value of the j-th measurement, N i For time period T i Number of measurements within;

[0100] The proportion of important images within an image is obtained by comparing the total number of important images within the image to the proportion of important images.

[0101] S5: Based on the temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image, a fuzzy Bayesian neural network is used to determine the transmission duration, which is then compared with the preset transmission duration. The transmitted image is then readjusted based on the comparison results to obtain the adjustment result.

[0102] Fuzzy Bayesian neural networks are a type of neural network model that combines fuzzy logic and Bayesian inference. Its network structure mainly includes an input layer, a hidden layer, and an output layer.

[0103] The input layer contains feature vectors of the input data's temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image. Each feature vector corresponds to an input node of the network.

[0104] As can be seen from the above, the temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image are all single features for evaluating transmission duration, so the input layer contains three input nodes;

[0105] The output layer produces the final output of the network, and each output node corresponds to the transmission duration T evaluated by the network:

[0106] The training method for hidden layers will be explained in detail below:

[0107] (1) By collecting historical data, the hidden layer is trained by continuously inputting the temporal upload bandwidth, temporal delay duration, the proportion of important parts of the image and the output transmission duration.

[0108] (2) Using the gradient descent method, the model parameters are initialized first. The weight coefficients β1, β2, and β3 are set, and the input time-domain upload bandwidth, time-domain delay duration, and the proportion of image importance in the image are substituted to obtain: β1*B(T i ), β2*D(T i ), β3*Ze; For each subsequent iteration, the gradient of the loss function with respect to the parameters needs to be calculated. The gradient shows the rate of change of the loss function at the current parameter value, and in which direction the parameters should be adjusted to reduce the loss; Typically, the mean squared error loss function is used as the loss function, and its mathematical expression is:

[0109]

[0110] In the formula, J(θ) is the mean squared error loss function, θ represents the model parameters, N is the sample size, and y i It is the true value of the i-th sample. This is the value of the i-th sample predicted by the model; the sample refers to the drying time of the region, thus calculating the loss; the gradient of the loss function with respect to the model parameters is calculated using the backpropagation algorithm. The backpropagation algorithm calculates the gradient layer by layer using the chain rule, propagating the gradient from the output layer to the input layer; now that the output is T and the loss function is J(θ), the gradient of the loss function with respect to the model parameters θ can be expressed as:

[0111]

[0112] (3) Update the parameters of the neural network based on the calculated gradient. The formula for updating the parameters is usually as follows:

[0113] θ new =θ old -Learning rate * gradient;

[0114] Where, θ old These are the current parameter values, the learning rate is a pre-set hyperparameter, and the gradient is calculated using the backpropagation algorithm described above.

[0115] The process of setting a learning rate is usually based on historical data experience or by trying different values. Generally speaking, the choice of learning rate depends on many factors, including the characteristics of the dataset, the complexity of the model, the choice of optimization algorithm, and the convergence speed during training. The learning rate of gradient descent is usually in the range of 0.1, 0.01, 0.001, etc.

[0116] It's important to note that a large learning rate can lead to excessively large parameter updates, making the optimization process unstable or even unable to converge; while a small learning rate can result in a slow convergence speed. During training, the learning rate typically needs to be adjusted to improve model performance and convergence speed. A common strategy is to use learning rate decay, which involves gradually reducing the learning rate as training progresses to allow for a more detailed search of the parameter space when approaching the optimal solution.

[0117] The training and validation sets are divided into different sizes. Generally, the training set contains enough samples to ensure that the model has enough data to learn the patterns and features of the task, while the validation set is used to evaluate the model's generalization ability and adjust the learning rate. Usually, a time-order partitioning method is used, that is, earlier data is used as the training set and newer data is used as the validation set. This can better simulate the model's generalization ability to future data in real applications. Of course, this embodiment is only an example of a detailed analysis of a certain method and does not represent the uniqueness of this method, so it will not be elaborated further here.

[0118] (4) Repeat steps (1) to (3) above for iteration. By setting a tolerance parameter, training can be stopped when the training performance does not improve in several consecutive iterations, which can avoid overfitting and save resources.

[0119] Once the transmission duration is determined, it is compared with the preset transmission duration. If it is less than the preset transmission duration, important images within the image are transmitted to the PC. Otherwise, an alarm is issued, and the amount of resources required for a single transmission is adjusted.

[0120] It should be noted that important images within the transmitted image, i.e. important parts of the image, require the allocation and use of different amounts of resources. The amount of resources remaining in the current time domain can be reflected by the upload bandwidth and time domain delay duration. The fuzzy Bayes algorithm is used to determine its transmission duration.

[0121] This invention obtains network status information and the proportion of important parts of an image. It establishes a fuzzy Bayes algorithm based on the temporal upload bandwidth, temporal delay duration, and the proportion of important parts of the image to determine the transmission duration. The transmission result is compared with the preset transmission duration to obtain the transmission result. The amount of resources required for a single transmission is adjusted, and the resources are dynamically adjusted to reduce the impact of network congestion on transmission, thus ensuring the timely transmission and processing of critical image information.

[0122] Example 3

[0123] Please see Figure 2 A system for acquiring and analyzing oral and nasal breathing data includes an acquisition module, a processing module, an image analysis module, a resource allocation module, and an output terminal.

[0124] The acquisition module is used to acquire images, divide the image into multiple parts based on image edge information, acquire partial images, and obtain the anomaly rate, respiratory frequency change intensity, and texture complexity of oral and nasal breathing data through image edge detection algorithms, which are then sent to the processing module. The network status information is acquired through network monitoring and sent to the resource allocation module.

[0125] The processing module is used to acquire the variance rate, respiratory rate change intensity, and texture complexity of some oral and nasal breathing data, establish a data analysis model, obtain image evaluation coefficients, and send them to the image analysis module.

[0126] The image analysis module is used to obtain image evaluation coefficients and compare them with image thresholds to obtain comparison results. Based on each comparison result and the number of comparison results, the module counts the number and proportion of important parts of the image and sends them to the resource allocation module.

[0127] The resource allocation module is used to obtain network status information and the proportion of important parts of the image within the image. It obtains the transmission time through a fuzzy Bayesian neural network and compares it with the preset transmission time. Based on the comparison result, it readjusts the transmitted image and sends the adjusted result to the output end.

[0128] The output terminal is used to receive the adjustment results and report them.

[0129] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0130] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0131] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply 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.

[0132] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0133] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0134] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0135] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0136] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0137] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0138] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for collecting and analyzing oral and nasal breathing data, characterized in that: include: S1: By grayscale and filtering the mouth and nose breathing images, noise interference is reduced. Texture analysis is used to extract features from the images. The images are divided into multiple parts by image edge information. Partial images are collected. The mouth and nose breathing data variation rate, breathing frequency change intensity and texture complexity are obtained by image edge detection algorithm. At the same time, network status information is collected by network monitoring. S2: Obtain the variation rate, respiratory frequency change intensity, and texture complexity of the oral and nasal breathing data of a portion of the image, establish a data analysis model, and obtain the image evaluation coefficients; S3: Obtain the image evaluation coefficients and compare them with the image threshold to obtain the comparison results. Based on each comparison result and the number of comparison results, count the number of important parts of the image and their proportion. S4: Obtain network status information and the proportion of important parts of the image within the image, and obtain the time-domain upload bandwidth and time-domain latency through data processing; S5: Based on the temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image, a fuzzy Bayesian neural network is used to determine the transmission duration, which is then compared with the preset transmission duration. The transmitted image is then readjusted based on the comparison results to obtain the adjustment result. Specifically, the color image is converted to a grayscale image, a Gaussian filter G is used to smooth the image to reduce noise, and the smoothed image is divided into multiple parts using the Canny edge detection algorithm. The specific steps are as follows: Step A1: Apply the Canny edge detection algorithm to detect edges in the smoothed image; Step A2: Detect contours from the edge detection results; Step A3: For each contour, create a binary mask; Step A4: Apply a mask to the original image to extract the corresponding portion of the image; The original image is divided into multiple partial images, where c represents the c-th partial image; By collecting inspiratory data below or above the high threshold and expiratory data below or above the low threshold, the frequency of occurrence is obtained, and the ratio of this frequency to the total number of breaths is used to obtain the variability rate (Arc) of oral and nasal breathing data. By determining the healthy respiratory cycle interval, respiratory frequencies in some images that are greater than or less than the healthy respiratory cycle interval are collected. The absolute value of the difference between the difference between the maximum and minimum values ​​of the respiratory cycle interval and the difference between the maximum and minimum values ​​of the healthy respiratory cycle interval is used to obtain the relative deviation of the single respiratory frequency. The respiratory frequency change intensity Dvc is obtained by adding these values ​​together. The texture complexity Tec is calculated by extracting texture features from the gray-level co-occurrence matrix. The variance rate, respiratory rate variation intensity, and texture complexity of nasal and oral breathing data were obtained. A data analysis model was established to obtain the image evaluation coefficient Ptc, based on the following formula: Ptc=φ1*Tec+φ2*Arc+φ3*Dvc In the formula, Ptc is the image evaluation coefficient, φ1, φ2 and φ3 are the preset proportional coefficients of texture complexity, the rate of variation of oral and nasal breathing data and the intensity of change in breathing frequency, respectively, and φ1, φ2 and φ3 are all greater than 0. Obtain image evaluation coefficients and compare and analyze them with iteratively selected image thresholds. If the image evaluation coefficient is greater than or equal to the image threshold, then that part of the image is marked as an important image and an important signal is generated; If the image evaluation coefficient is less than the image threshold, then that part of the image is marked as an unimportant image and an end signal is generated; Multiple image evaluation coefficient calculations will be performed to obtain the image evaluation coefficient values ​​of multiple parts of an image and two sets of comparison results of the multiple parts of the image. Based on the comparison results, the image evaluation coefficient values ​​of the parts of the image evaluated as important signals and their proportions will be statistically analyzed.

2. The method for collecting and analyzing oral and nasal breathing data according to claim 1, characterized in that: Network status information includes time-domain upload bandwidth and time-domain latency. By dividing the time domain into different sizes and labeling the time domain as i, and dividing it into n time domains, the time domain upload bandwidth BTi is obtained based on the amount of data transmitted in the time domain and the duration of the time domain. After dividing the time domain, the latency is measured using the ping command or other network measurement tools to obtain the time domain latency duration DTi; The importance percentage (Ze) of a partial image is obtained by comparing the total number of partial images within the image with the number of important images.

3. The method for collecting and analyzing oral and nasal breathing data according to claim 2, characterized in that: The transmission duration is determined using a fuzzy Bayesian neural network. The temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image are defined as input nodes in the input layer, and the transmission duration is defined as the output layer. A fuzzy Bayesian neural network was developed to describe the impact of temporal upload bandwidth, temporal delay duration, and the proportion of image importance within an image on transmission duration. Based on the fuzzy Bayesian neural network, the transmission duration is determined by the temporal upload bandwidth, temporal delay duration, and the proportion of image importance within the image.

4. The method for collecting and analyzing oral and nasal breathing data according to claim 3, characterized in that: Once the transmission duration is determined, it is compared with the preset transmission duration. If it is less than the preset transmission duration, important images within the image are transmitted to the PC. Otherwise, an alarm is issued, and the amount of resources required for a single transmission is adjusted.

5. A system for acquiring and analyzing oral and nasal breathing data, used to implement the method for acquiring and analyzing oral and nasal breathing data according to any one of claims 1-4, characterized in that: It includes an acquisition module, a processing module, an image analysis module, a resource allocation module, and an output terminal; The acquisition module is used to acquire images, divide the image into multiple parts based on image edge information, acquire partial images, and obtain the anomaly rate, respiratory frequency change intensity, and texture complexity of oral and nasal breathing data through image edge detection algorithms, which are then sent to the processing module. The network status information is acquired through network monitoring and sent to the resource allocation module. The processing module is used to acquire the variance rate, respiratory rate change intensity, and texture complexity of some oral and nasal breathing data, establish a data analysis model, obtain image evaluation coefficients, and send them to the image analysis module. The image analysis module is used to obtain image evaluation coefficients and compare them with image thresholds to obtain comparison results. Based on each comparison result and the number of comparison results, the module counts the number and proportion of important parts of the image and sends them to the resource allocation module. The resource allocation module is used to obtain network status information and the proportion of important parts of the image within the image. It obtains the transmission time through a fuzzy Bayesian neural network and compares it with the preset transmission time. Based on the comparison result, it readjusts the transmitted image and sends the adjusted result to the output end. The output terminal is used to receive the adjustment results and report them.

Images