A method, system, storage medium and device for detecting end-tidal
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN COMEN MEDICAL INSTR
- Filing Date
- 2023-09-22
- Publication Date
- 2026-07-10
AI Technical Summary
End-tidal value detection is easily affected by false exhalation and respiratory waveform distortion, which reduces the accuracy of the test.
A respiratory model was constructed using gated convolution and residual networks. The optimal network weight matrix was obtained through training and testing, and the model was used to detect the end-tidal value of real breathing.
It enhances the anti-interference ability of end-tidal value detection and improves the accuracy of detection.
Smart Images
Figure CN117357095B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of respiratory monitoring, and particularly relates to a method, system, storage medium and device for end-tidal value detection. Background Technology
[0002] Monitoring carbon dioxide and anesthetic gas concentrations is particularly important in respiratory monitoring. End-tidal carbon dioxide and anesthetic gas concentrations are crucial parameters characterizing these concentrations during respiration. Therefore, accurately detecting end-tidal values is of great significance for respiratory monitoring.
[0003] However, end-expiratory value (EPV) detection is susceptible to many interferences. These include patient movements such as sham exhalation and distortions in the respiratory waveform, which significantly reduce the accuracy of EPV detection. Therefore, improving the accuracy of EPV monitoring requires enhancing its resistance to interference. Summary of the Invention
[0004] In view of the above problems, the present invention provides a method, system, storage medium and device for end-tidal value detection, which can significantly enhance the anti-interference ability of end-tidal value monitoring and improve the accuracy of end-tidal value detection.
[0005] The technical solution adopted in this invention is: a method for detecting end-tidal values, comprising the following steps:
[0006] Obtain the respiratory dataset;
[0007] Label the predicted end-tidal values in the respiratory dataset;
[0008] A breathing model constructed using gated convolution and residual networks was trained using the breathing dataset to obtain a breathing model with the optimal network weight matrix;
[0009] The respiratory model with the optimal network weight matrix was tested using the respiratory dataset to obtain the test end-tidal values;
[0010] If the difference between the tested end-expiratory value and the predicted end-expiratory value is within a preset range, the actual breathing is detected using the breathing model with the optimal network weight matrix, and the end-expiratory value of the actual breathing is obtained.
[0011] Preferably, the step of training a breathing model using the breathing dataset to construct a gated convolutional and residual network to obtain a breathing model with the optimal network weight matrix includes:
[0012] Input the respiratory dataset into the respiratory model;
[0013] The signal features of the training set of the respiratory dataset are obtained by performing gated convolution on the training set.
[0014] The residual network is used to convolve the signal features of the training set to obtain the end-tidal values of the model.
[0015] The accuracy and loss rate are obtained by comparing the model end-tidal value with the predicted end-tidal value.
[0016] If the accuracy is higher than the preset accuracy and the loss rate is lower than the preset loss rate, then the network weight matrix used by the breathing model is considered to be the optimal network weight matrix, and a breathing model with the optimal network weight matrix is obtained.
[0017] Preferably, obtaining the respiratory dataset includes:
[0018] The raw respiratory signal is acquired and then filtered to obtain a filtered respiratory signal.
[0019] The filtered respiratory signal is labeled based on historical experience to obtain a labeled respiratory signal;
[0020] The labeled respiratory signals are extracted at preset intervals to obtain the respiratory dataset.
[0021] Preferably, the preset cycle is set according to at least one of a preset number of respiratory cycles and a preset respiratory duration.
[0022] Preferably, the respiratory dataset is divided into a training set, a validation set, and a test set in a 7:2:1 ratio;
[0023] The breathing model constructed by training gated convolutional and residual networks using the breathing dataset to obtain a breathing model with the optimal network weight matrix includes:
[0024] The training set is input into the breathing model constructed by the gated convolution and residual network for training, resulting in a breathing model with a network weight matrix to be verified.
[0025] The validation set is input into the breathing model with the network weight matrix to be validated. By verifying the accuracy and loss rate, the breathing model with the optimal network weight matrix is obtained.
[0026] The step of testing the respiratory model with the optimal network weight matrix using the respiratory dataset to obtain the test end-tidal value includes:
[0027] The test set is input into the respiratory model with the optimal network weight matrix to obtain the test end-tidal values.
[0028] Preferably, the gated convolution includes a gated convolution formula, which is as follows:
[0029]
[0030] In the formula, H l H represents the output of the current layer. l-1 W indicates that the output of the previous layer is used as the input of this layer. g and W f σ represents different convolution filters, and σ represents the Sigmoid function.
[0031] Preferably, the residual network includes a residual formula, which is as follows;
[0032] Z(x) = F(x, W) + x
[0033] In the formula, Z(x) represents the output of the residual network, F(x,W) represents the output of the convolution in the residual network, x is the breathing dataset, and W represents the convolution filter.
[0034] An end-tidal value detection system includes:
[0035] The data module is used to acquire respiratory datasets;
[0036] The model training module is used to label the predicted end-tidal values in the respiratory dataset; train a respiratory model constructed by gated convolution and residual networks using the respiratory dataset to obtain a respiratory model with the optimal network weight matrix; and test the respiratory model with the optimal network weight matrix using the respiratory dataset to obtain test end-tidal values.
[0037] The model detection module is used to detect the actual breathing using the breathing model with the optimal network weight matrix if the difference between the tested end-expiratory value and the predicted end-expiratory value is within a preset range, thereby obtaining the end-expiratory value of the actual breathing.
[0038] A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the method as described above.
[0039] A computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described above.
[0040] The embodiments of the present invention have the following beneficial effects:
[0041] This invention labels the predicted end-tidal values in the respiratory dataset and trains a respiratory model using gated convolutions and residual networks to obtain a respiratory model with an optimal network weight matrix. If the difference between the end-tidal value obtained by the respiratory model and the predicted end-tidal value is within a preset range when tested using the respiratory dataset, the actual breathing is detected using the respiratory model to obtain the actual end-tidal value. Therefore, this invention enhances the anti-interference ability of the respiratory model constructed using gated convolutions and residual networks, and improves the detection accuracy. Attached Figure Description
[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0043] in:
[0044] Figure 1 Here is a flowchart of an end-tidal value detection method in one embodiment;
[0045] Figure 2 This is a diagram of the internal structure of a computer device in one embodiment.
[0046] The present invention will be further explained below with reference to the accompanying drawings and embodiments. 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] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.
[0049] The end-tidal value detection method provided in this application will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.
[0050] This application provides a method for detecting end-tidal values. Figure 1 This is a flowchart of an end-tidal value detection method in one embodiment. For example... Figure 1 As shown, in one embodiment, the end-tidal value detection method includes the following steps:
[0051] Step S101: Obtain the respiratory dataset.
[0052] First, it's necessary to specify what type of respiratory dataset you want to obtain. For example, you might need a respiratory dataset containing people of different ages, genders, and health conditions. Second, you can also obtain free respiratory datasets that are publicly available online.
[0053] After obtaining the respiratory dataset, some preprocessing work may be required, such as cleaning the data, filling in missing values, and converting data types.
[0054] Step S102: Label the predicted end-tidal values in the respiratory dataset.
[0055] Specifically, by observing the signals in the respiratory dataset, predicted end-tidal values can be manually labeled based on expertise and experience.
[0056] Step S103: Train a breathing model constructed using a gated convolutional and residual network using the breathing dataset to obtain a breathing model with the optimal network weight matrix.
[0057] Furthermore, the residual network can be a deep residual network. Features of the breathing pattern are extracted using gated convolutional layers, which control the transmission of information across different frequency ranges. A residual connection is added after each gated convolutional layer to maintain the consistency of the feature maps. The final layer can be a fully connected layer or a regression layer, selecting the appropriate output data based on the end-tidal value task requirements.
[0058] The breathing model is then trained using the breathing dataset, and the network weights are updated via backpropagation. The model's performance is evaluated using a subset of data from the breathing dataset, and adjustments and optimizations are made as needed. When the model converges or reaches a predetermined number of training epochs, the current optimal network weight matrix is saved, and the current breathing model is considered to have the optimal network weight matrix.
[0059] Step S104: Use the respiratory dataset to test the respiratory model with the optimal network weight matrix and obtain the test end-tidal value.
[0060] Specifically, after obtaining the optimal network weight matrix for the respiratory model, a final evaluation is performed using unused data from the respiratory dataset. This final evaluation determines whether the end-tidal values obtained by inputting unused data from the respiratory dataset into the model meet the requirements. This ensures that the model does not overfit during training and also allows for further testing of the respiratory model's performance using unused data, thereby improving the model's usability.
[0061] Furthermore, a respiratory model with the optimal network weight matrix can output the index positions of the respiratory signals (an array of positions), and the test end-tidal value can be obtained based on the index positions.
[0062] In this embodiment, the breathing model with the optimal network weight matrix that is being tested can be released in a grayscale and subjected to multiple tests to improve the practicality and stability of the breathing model with the optimal network weight matrix.
[0063] Step S105: If the difference between the tested end-expiratory value and the predicted end-expiratory value is within a preset range, then the actual breathing is detected using a breathing model with the optimal network weight matrix to obtain the actual end-expiratory value of the breathing.
[0064] It should be noted that the tested end-expiratory value is compared with the pre-labeled end-expiratory value of the current data, and the difference between them is calculated. If the difference is within a preset range, it indicates that the respiratory model with the optimal network weight matrix is effective in detecting end-expiratory values. In this case, applying the respiratory model with the optimal network weight matrix to the actual patient data to predict their end-expiratory values, we can consider the end-expiratory values obtained by the respiratory model with the optimal network weight matrix to be the true end-expiratory values.
[0065] This application replaces the recursive connections commonly used in recursive networks with gated convolutions, a mechanism that mitigates gradient propagation. Furthermore, a deep residual network is used to redirect input information to the output, preserving information integrity, simplifying the learning objective and difficulty, and ensuring the network is less prone to gradient vanishing or exploding problems. By using gated convolutions and deep residual networks to detect end-tidal values, the robustness of the end-tidal detection algorithm is enhanced, and detection accuracy is improved.
[0066] In one executable embodiment, a breathing model constructed using gated convolutional and residual networks is trained using a breathing dataset to obtain a breathing model with the optimal network weight matrix, including:
[0067] Input the breathing dataset into the breathing model.
[0068] Furthermore, the breathing dataset is split into training, validation, and test sets. Typically, 80% of the data is used as the training set, 10% as the validation set, and the remaining 10% as the test set. This allows for evaluation of the model's performance on unknown data and avoids overfitting or underfitting issues.
[0069] The signal features of the training set of the respiratory dataset are obtained by performing gated convolution on the training set.
[0070] Specifically, gated convolution is a special type of convolution operation that only considers signals within a specific time window during the convolution process.
[0071] Furthermore, the training set of the respiratory dataset is subjected to gated convolution, and signal features of important information about the respiratory signals in the dataset are extracted based on 1D convolution operations of K filters. The filters can extract the correlation between K adjacent values in the input data. These correlations are used as signal features of the training set.
[0072] The residual network is used to convolve the signal features of the training set to obtain the end-tidal values of the model.
[0073] It's important to note that by combining deep residual networks, the original features of the input respiratory signal and features lost during convolution are preserved, resulting in a feature matrix that combines the input respiratory signal after convolution with the original temporal features. By convolving the signal features, residual networks can better capture complex patterns and features in the data. This helps improve the model's predictive and generalization abilities, thereby enhancing overall performance.
[0074] The accuracy and loss rate are obtained by comparing the model's end-tidal values with the predicted end-tidal values.
[0075] Specifically, accuracy represents the proportion of correctly predicted samples out of the total number of samples. The calculation formula is: (TP+TN) / (TP+FP+TN+FN), where TP is the true positives, TN is the true negatives, FP is the false positives, and FN is the false negatives. Loss rate is typically used to measure the difference between the model's predictions and the actual results. Common loss functions include mean squared error (MSE) and cross-entropy loss. The smaller the value of the loss function, the better the model's performance.
[0076] Evaluating the results of respiratory models by assessing accuracy and loss rate can improve the accuracy of respiratory models.
[0077] If the accuracy is higher than the preset accuracy and the loss rate is lower than the preset loss rate, then the network weight matrix used by the breathing model is considered to be the optimal network weight matrix, and a breathing model with the optimal network weight matrix is obtained.
[0078] Furthermore, if a breathing model's prediction accuracy exceeds a preset accuracy threshold and its loss rate is less than a preset loss rate threshold, then the network weight matrix used in this model can be considered optimal. In this case, a breathing model with optimal performance (i.e., high accuracy and low loss rate) can be obtained.
[0079] In one executable embodiment, obtaining a respiratory dataset includes:
[0080] Raw respiratory signals are collected and filtered to obtain filtered respiratory signals.
[0081] Specifically, the raw respiratory signal is preprocessed, such as through filtering and noise reduction, to reduce noise interference.
[0082] Based on historical experience, the filtered respiratory signals are labeled to obtain labeled respiratory signals.
[0083] It should be noted that historical experience mainly includes normal breathing patterns and respiratory characteristics of respiratory diseases.
[0084] Understanding and recognizing normal breathing patterns is fundamental to respiratory signal labeling. This requires an understanding of various aspects, including the depth, frequency, and rhythm of breathing.
[0085] Characteristics of respiratory diseases, such as chronic obstructive pulmonary disease (COPD), often include a longer exhalation time during inhalation, while asthma patients may have more frequent wheezing.
[0086] The labeled respiratory signals are extracted according to a preset period to obtain a respiratory dataset.
[0087] Furthermore, gated convolution only considers signals within a specific time window. Therefore, in a long string of respiratory signals, the signals need to be truncated according to a certain period to finally obtain a usable respiratory dataset.
[0088] In one executable embodiment, the preset cycle is set based on at least one of a preset number of respiratory cycles and a preset respiratory duration.
[0089] Normally, the respiratory signal is captured every 20 cycles. If the respiratory rate is too fast, the signal is captured in 70-second intervals.
[0090] In one executable embodiment, the breathing dataset is divided into a training set, a validation set, and a test set in a 7:2:1 ratio.
[0091] This ratio ensures that the model is trained and evaluated on different datasets, thus better generalizing to new data. The training set comprises 70% of the entire dataset, the validation set 20%, and the test set 10%. This division ensures effective model training while avoiding overfitting and underfitting.
[0092] A breathing model constructed using gated convolutional and residual networks was trained using a breathing dataset to obtain a breathing model with the optimal network weight matrix, including:
[0093] The training set is input into a breathing model constructed using gated convolution and residual networks for training, resulting in a breathing model with a network weight matrix to be validated.
[0094] Specifically, the training set is input into the selected model, and the network weight matrix is adjusted to minimize the prediction error.
[0095] The validation set is input into the breathing model with the network weight matrix to be validated. By verifying the accuracy and loss rate, the breathing model with the optimal network weight matrix is obtained.
[0096] Furthermore, the validation set is fed into the trained model, and various metrics are calculated and recorded to evaluate the model's performance. These metrics should differ from those used during training. Suitable metrics include cross-validation, leave-one-out, confusion matrix, ROC curve, and mean absolute error.
[0097] The respiratory model with the optimal network weight matrix was tested using a respiratory dataset to obtain test end-tidal values, including:
[0098] Input the test set into the respiratory model with the optimal network weight matrix to obtain the test end-tidal values.
[0099] By inputting the respiratory model with the optimal network weight matrix into the test set and obtaining test end-tidal values, overfitting can be prevented. Overfitting occurs when a model performs well on the training set but poorly on new, unseen data. If we only use the training set to train and evaluate the model, we may overestimate its performance on new data. By inputting the test set into the model, we can more accurately assess its generalization ability. The test set provides feedback on the model's performance. If the respiratory model performs poorly on the test set, the problem needs to be identified, and the network weight matrix of the respiratory model needs to be adjusted.
[0100] In one executable embodiment, gated convolution includes a gated convolution formula, which is as follows:
[0101]
[0102] In the formula, Hl H represents the output of the current layer. l-1 W indicates that the output of the previous layer is used as the input of this layer. g and W f σ represents different convolution filters, and σ represents the Sigmoid function.
[0103] In one executable embodiment, the residual network includes a residual formula, as follows;
[0104] Z(x) = F(x, W) + x
[0105] In the formula, Z(x) represents the output of the residual network, F(x,W) represents the output of the convolution in the residual network, x is the breathing dataset, and W represents the convolution filter.
[0106] The present invention also provides an end-tidal value detection system, comprising:
[0107] The data module is used to acquire respiratory datasets.
[0108] The model training module is used to label the predicted end-tidal values in the respiratory dataset; train a respiratory model constructed by gated convolution and residual networks using the respiratory dataset to obtain a respiratory model with the optimal network weight matrix; and test the respiratory model with the optimal network weight matrix using the respiratory dataset to obtain the test end-tidal values.
[0109] The model detection module is used to detect the actual breathing using a breathing model with the best network weight matrix if the difference between the tested end-expiratory value and the predicted end-expiratory value is within a preset range, thereby obtaining the actual end-expiratory value of the breathing.
[0110] The aforementioned end-tidal value detection system acquires a respiratory dataset, labels the predicted end-tidal values in the dataset, trains a respiratory model using gated convolutional and residual networks based on the dataset to obtain a respiratory model with the optimal network weight matrix, then tests the model with the optimal network weight matrix using the dataset to obtain the test end-tidal value, and finally detects the actual breathing using the respiratory model with the optimal network weight matrix when the difference between the test end-tidal value and the predicted end-tidal value is within a preset range, thus obtaining the actual end-tidal value of the breathing.
[0111] In one embodiment, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described end-tidal value detection method embodiment and achieves the same technical effect. To avoid repetition, it will not be described again here.
[0112] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, it implements the various processes of the above-described end-tidal value detection method embodiment and achieves the same technical effect. To avoid repetition, it will not be described again here.
[0113] Figure 2 This is a diagram illustrating the internal structure of a computer device in one embodiment. This computer device can specifically be a terminal or a server. Figure 2 As shown, the computer device includes a processor, memory, and network interface connected via a system bus. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and may also store a computer program. When executed by the processor, this computer program enables the processor to implement the end-tidal value detection method. The internal memory may also store a computer program, which, when executed by the processor, enables the processor to perform the end-tidal value detection method. Those skilled in the art will understand that... Figure 2 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0114] Those skilled in the art will understand that implementing all or part of the processes in the above embodiments can be accomplished by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can 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 various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.
[0115] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0116] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A method for detecting end-tidal values, characterized in that, Includes the following steps: Obtain the respiratory dataset; Label the predicted end-tidal values in the respiratory dataset; A breathing model constructed using gated convolution and residual networks was trained using the breathing dataset to obtain a breathing model with the optimal network weight matrix; The respiratory model with the optimal network weight matrix was tested using the respiratory dataset to obtain the test end-tidal values; If the difference between the tested end-expiratory value and the predicted end-expiratory value is within a preset range, then the actual breathing is detected using the breathing model with the optimal network weight matrix, and the end-expiratory value of the actual breathing is obtained. The breathing model constructed by training gated convolutional and residual networks using the breathing dataset to obtain a breathing model with the optimal network weight matrix includes: Input the respiratory dataset into the respiratory model; The signal features of the training set of the respiratory dataset are obtained by performing gated convolution on the training set. The residual network is used to convolve the signal features of the training set to obtain the end-tidal values of the model. The accuracy and loss rate are obtained by comparing the model end-tidal value with the predicted end-tidal value. If the accuracy is higher than the preset accuracy and the loss rate is lower than the preset loss rate, then the network weight matrix used by the breathing model is considered to be the optimal network weight matrix, and a breathing model with the optimal network weight matrix is obtained. The respiratory dataset was divided into a training set, a validation set, and a test set in a 7:2:1 ratio. The breathing model constructed by training gated convolutional and residual networks using the breathing dataset to obtain a breathing model with the optimal network weight matrix includes: The training set is input into the breathing model constructed by the gated convolution and residual network for training, resulting in a breathing model with a network weight matrix to be verified. The validation set is input into the breathing model with the network weight matrix to be validated. By verifying the accuracy and loss rate, the breathing model with the optimal network weight matrix is obtained. The step of testing the respiratory model with the optimal network weight matrix using the respiratory dataset to obtain the test end-tidal value includes: The test set is input into the respiratory model with the optimal network weight matrix to obtain the test end-tidal value; The gated convolution includes a gated convolution formula, which is as follows: In the formula, This indicates the output of the current layer. This means that the output of the previous layer is used as the input of this layer. and Representing different convolution filters, This refers to the Sigmod function; The residual network includes a residual formula, which is as follows; In the formula, This represents the output of the residual network. This represents the output of the convolution in the residual network. For the respiratory dataset, This represents a convolution filter.
2. The method according to claim 1, characterized in that, The acquisition of the respiratory dataset includes: The raw respiratory signal is acquired and then filtered to obtain a filtered respiratory signal. The filtered respiratory signal is labeled based on historical experience to obtain a labeled respiratory signal; The labeled respiratory signals are extracted at preset intervals to obtain the respiratory dataset.
3. The method according to claim 2, characterized in that, The preset cycle is set based on at least one of a preset number of breathing cycles and a preset breathing time length.
4. An end-tidal value detection system, applied to the end-tidal value detection method as described in claim 1, characterized in that, include: The data module is used to acquire respiratory datasets; The model training module is used to label the predicted end-tidal values in the respiratory dataset; A respiratory model constructed using gated convolutional and residual networks is trained using the respiratory dataset to obtain a respiratory model with the optimal network weight matrix; the respiratory model with the optimal network weight matrix is then tested using the respiratory dataset to obtain test end-tidal values. The model detection module is used to detect the actual breathing using the breathing model with the optimal network weight matrix if the difference between the tested end-expiratory value and the predicted end-expiratory value is within a preset range, thereby obtaining the end-expiratory value of the actual breathing.
5. A computer-readable storage medium, characterized in that, The system contains a computer program that, when executed by a processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 3.
6. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 3.