Lung ventilation function detection system based on algorithm analysis

Abstract

The invention relates to the technical field of lung function detection, and particularly discloses a lung ventilation function detection system based on algorithm analysis, which adopts a data processing technology based on deep learning to carry out data analysis on breathing parameters of a patient, extracts semantic embedding feature representation of each breathing parameter data item, and obtains a semantic embedding feature representation of each breathing parameter data item; and on the basis of the semantic association topological relation among the breathing parameter data items, global depth association coding is performed on the semantic embedding features of the breathing parameter data items, so that global depth perception of the breathing state of the patient is realized, and whether the patient suffers from bronchial asthma or not is intelligently identified. In this way, the limitation of a traditional data analysis method can be effectively overcome, the dependence on professional knowledge of doctors is reduced, the accuracy and efficiency of pulmonary ventilation function detection are improved, and the risks of misdiagnosis and missed diagnosis are reduced.

Description

Technical Field

[0001] The present application relates to the technical field of pulmonary function detection, and more specifically, to a pulmonary ventilation function detection system based on algorithm analysis. Background Art

[0002] With the development of medical technology, the early diagnosis and treatment of lung diseases, especially chronic respiratory diseases, have received more and more attention. Bronchial asthma is a chronic inflammatory disease characterized by airway hyperresponsiveness and reversible airflow limitation. As a common chronic inflammatory disease, it affects the health of hundreds of millions of people worldwide.

[0003] Traditional pulmonary ventilation function testing relies on professional medical equipment, such as spirometers, to accurately measure the patient's respiratory parameters and analyze the data based on the doctor's clinical experience to evaluate the patient's lung function. However, this method has certain limitations, such as low diagnostic efficiency, high requirements for doctors' professional knowledge, and certain subjectivity and errors.

[0004] In recent years, with the rapid development of big data and artificial intelligence technology, algorithm-based medical health monitoring systems have gradually become a research hotspot. A series of important advances have been made by building artificial intelligence algorithms to process and analyze large-scale medical data. For example, in medical imaging, image analysis algorithms can effectively assist doctors in disease diagnosis, improve the accuracy and efficiency of diagnosis, and have been successfully applied to early screening of lung cancer.

[0005] However, for the detection of pulmonary ventilation function, some existing intelligent analysis algorithms mainly focus on the numerical changes of a single respiratory parameter, but lack in-depth analysis of the complex correlation between multiple parameters, which may ignore the potential interactions and joint influences between respiratory parameters, thereby limiting the accuracy and comprehensiveness of the diagnosis.

[0006] Therefore, an optimized pulmonary ventilation function detection system based on algorithm analysis is expected. Summary of the invention

[0007] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a pulmonary ventilation function detection system based on algorithm analysis, which uses a data processing technology based on deep learning to perform data analysis on the patient's respiratory parameters, and extracts the semantic embedded feature representation of each respiratory parameter data item respectively, and based on the semantic association topological relationship between each respiratory parameter data item, performs global deep association encoding on the semantic embedded features of each respiratory parameter data item, thereby achieving global deep perception of the patient's respiratory state, so as to intelligently identify whether the patient suffers from bronchial asthma. In this way, the limitations of traditional data analysis methods can be effectively overcome, the dependence on doctors' professional knowledge can be reduced, the accuracy and efficiency of pulmonary ventilation function detection can be improved, and the risk of misdiagnosis and missed diagnosis can be reduced.

[0008] Accordingly, according to one aspect of the present application, a pulmonary ventilation function detection system based on algorithm analysis is provided, comprising:

[0009] A respiratory parameter monitoring module, used to obtain respiratory parameters of the patient subject collected by the spirometer, wherein the respiratory parameters include tidal volume, inspiratory reserve volume, expiratory reserve volume, deep inspiratory volume and residual volume;

[0010] An embedded coding module, used for embedding and coding each data item in the breathing parameter to obtain a set of embedded coding vectors of the breathing parameter data items;

[0011] A feature enhancement module, used for performing feature enhancement processing on the set of breathing parameter data item embedded coding vectors to obtain a set of breathing parameter data item embedded enhanced coding vectors;

[0012] A global deep semantic association coding module, used for performing global deep semantic association coding on the set of embedded enhanced coding vectors of the breathing parameter data items based on the semantic association topological relationship between the data items in the breathing parameters to obtain a global deep semantic coding feature vector of the breathing state;

[0013] The pulmonary ventilation function detection module is used to determine whether the patient suffers from bronchial asthma based on the global deep semantic encoding feature vector of the respiratory state.

[0014] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the feature enhancement module includes: a feature significance calculation unit, which is used to calculate the semantic significance description factor of each respiratory parameter data item embedded coding vector in the set of respiratory parameter data item embedded coding vectors to obtain a set of respiratory parameter data item significance description factors; a feature modulation unit, which is used to perform hierarchical mask modulation on the set of respiratory parameter data item embedded coding vectors based on the set of respiratory parameter data item significance description factors to obtain a set of respiratory parameter data item embedded enhanced coding vectors.

[0015] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the embedded coding module is used to: use an embedded encoder based on a fully connected layer to perform low-dimensional embedded coding on each data item in the respiratory parameter to obtain a set of embedded coding vectors of the respiratory parameter data items.

[0016] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the feature significance calculation unit includes: a hyperbolic space mapping subunit, which is used to map the individual respiratory parameter data item embedded coding vectors in the set of respiratory parameter data item embedded coding vectors to the hyperbolic space to obtain a set of mapped respiratory parameter data item embedded coding vectors; a feature significance quantification description subunit, which is used to calculate the semantic significance description factor of each mapped respiratory parameter data item embedded coding vector in the set of mapped respiratory parameter data item embedded coding vectors to obtain a set of significance description factors of the respiratory parameter data items.

[0017] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the feature significance quantification description subunit is used to: calculate the square Poincare norm of each mapped respiratory parameter data item embedding coding vector in the set of mapped respiratory parameter data item embedding coding vectors as the semantic significance description factor to obtain the set of respiratory parameter data item significance description factors.

[0018] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the feature modulation unit includes: a hierarchical mask subunit, which is used to perform hierarchical mask modulation on the set of significance descriptors of the respiratory parameter data items to obtain a set of significance modulation weights of the respiratory parameter data items; a weighted modulation subunit, which is used to use the significance modulation weights of each respiratory parameter data item in the set of significance modulation weights of the respiratory parameter data items as weights, and respectively perform weighted modulation on the corresponding respiratory parameter data item embedding coding vectors in the set of respiratory parameter data item embedding coding vectors to obtain a set of respiratory parameter data item embedding enhancement coding vectors.

[0019] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the hierarchical mask subunit is used to: input the set of significance description factors of the respiratory parameter data items into a normalization process based on a softmax function to obtain a set of normalized significance description factors of the respiratory parameter data items; input the set of normalized significance description factors of the respiratory parameter data items into a multi-level mask function to obtain a set of significance modulation weights of the respiratory parameter data items.

[0020] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the global deep semantic association coding module includes: a respiratory parameter adjacency topology matrix construction unit, which is used to construct a respiratory parameter adjacency topology matrix based on the semantic association topological relationship between each data item in the respiratory parameter; a respiratory state global feature extraction unit, which is used to embed the respiratory parameter data items into a set of enhanced coding vectors and the respiratory parameter adjacency topology matrix and input them into a respiratory state global deep semantic feature extractor based on a graph convolutional neural network model to obtain the respiratory state global deep semantic coding feature vector.

[0021] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the respiratory parameter adjacency topology matrix construction unit includes: an adjacency matrix construction subunit, which is used to construct an adjacency matrix of a set of respiratory parameter data items embedded in coding vectors, and the eigenvalues ​​of each non-diagonal position in the adjacency matrix are the respiratory state semantic association coefficients between the corresponding two respiratory parameter data items embedded in the coding vectors; an adjacency topology feature extraction subunit, which is used to input the adjacency matrix of the set of respiratory parameter data items embedded in coding vectors into a respiratory parameter adjacency topology feature extractor based on a dilated convolutional layer to obtain the respiratory parameter adjacency topology matrix.

[0022] In the above-mentioned pulmonary ventilation function detection system based on algorithm analysis, the pulmonary ventilation function detection module is used to: input the global deep semantic encoding feature vector of the respiratory state into a classifier-based respiratory state evaluator to obtain an evaluation result, and the evaluation result is used to indicate whether the patient suffers from bronchial asthma.

[0023] Compared with the prior art, the pulmonary ventilation function detection system based on algorithm analysis provided by the present application uses data processing technology based on deep learning to perform data analysis on the patient's respiratory parameters, extracts the semantic embedding feature representation of each respiratory parameter data item, and performs global deep association encoding on the semantic embedding features of each respiratory parameter data item based on the semantic association topological relationship between each respiratory parameter data item, thereby realizing global deep perception of the patient's respiratory state, so as to intelligently identify whether the patient suffers from bronchial asthma. In this way, the limitations of traditional data analysis methods can be effectively overcome, the dependence on doctors' professional knowledge can be reduced, the accuracy and efficiency of pulmonary ventilation function detection can be improved, and the risk of misdiagnosis and missed diagnosis can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0025] Figure 1 It is a block diagram of a pulmonary ventilation function detection system based on algorithm analysis according to an embodiment of the present application.

[0026] Figure 2 Schematic diagram of data flow of a pulmonary ventilation function detection system based on algorithm analysis according to an embodiment of the present application.

[0027] Figure 3 It is a block diagram of a feature enhancement module in a pulmonary ventilation function detection system based on algorithm analysis according to an embodiment of the present application.

[0028] Figure 4 It is a block diagram of a global deep semantic association encoding module in a pulmonary ventilation function detection system based on algorithm analysis according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above and other purposes, features and advantages of the present application will become more apparent. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.

[0030] In response to the technical problems described in the above background technology, the technical concept of this application is: using data processing technology based on deep learning to analyze the patient's respiratory parameters, extracting the semantic embedding feature representation of each respiratory parameter data item, and based on the semantic association topological relationship between each respiratory parameter data item, performing global deep association encoding on the semantic embedding features of each respiratory parameter data item, thereby achieving global deep perception of the patient's respiratory state, so as to intelligently identify whether the patient suffers from bronchial asthma. In this way, the limitations of traditional data analysis methods can be effectively overcome, the dependence on doctors' professional knowledge can be reduced, the accuracy and efficiency of pulmonary ventilation function testing can be improved, and the risk of misdiagnosis and missed diagnosis can be reduced.

[0031] Figure 1 It is a block diagram of a pulmonary ventilation function detection system based on algorithm analysis according to an embodiment of the present application. Figure 2 FIG. 1 is a data flow diagram of a pulmonary ventilation function detection system based on algorithm analysis according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the embodiment of the present application, the pulmonary ventilation function detection system 100 based on algorithm analysis includes: a respiratory parameter monitoring module 110, which is used to obtain the respiratory parameters of the patient subject collected by the spirometer, and the respiratory parameters include tidal volume, supplementary inspiratory volume, supplementary expiratory volume, deep inspiratory volume and residual volume; an embedded coding module 120, which is used to embed and encode each data item in the respiratory parameters to obtain a set of respiratory parameter data item embedded coding vectors; a feature enhancement module 130, which is used to perform feature enhancement processing on the set of respiratory parameter data item embedded coding vectors to obtain a set of respiratory parameter data item embedded enhanced coding vectors; a global deep semantic association coding module 140, which is used to perform global deep semantic association coding on the set of respiratory parameter data item embedded enhanced coding vectors based on the semantic association topological relationship between each data item in the respiratory parameters to obtain a global deep semantic coding feature vector of the respiratory state; a pulmonary ventilation function detection module 150, which is used to determine whether the patient suffers from bronchial asthma based on the global deep semantic coding feature vector of the respiratory state.

[0032] In the above-mentioned pulmonary ventilation function detection system 100 based on algorithm analysis, the respiratory parameter monitoring module 110 is used to obtain the respiratory parameters of the patient subject collected by the spirometer, and the respiratory parameters include tidal volume, supplementary inspiratory volume, supplementary expiratory volume, deep inspiratory volume and residual volume. It should be understood that tidal volume (Tidal Volume, VT) refers to the amount of air inhaled or exhaled each time during quiet breathing, reflecting the basic ventilation of the lungs during normal breathing. Supplementary inspiratory volume (Inspiratory Reserve Volume, IRV) and supplementary expiratory volume (Expiratory Reserve Volume, ERV) refer to the maximum amount of air that can be inhaled and exhaled after quiet inhalation, reflecting the reserve capacity of the lungs. Deep inspiratory capacity (Inspiratory Capacity, IC) is the sum of tidal volume and supplementary inspiratory volume, representing the total capacity of the lungs under maximum inspiratory state. Residual volume (Residual Volume, RV) refers to the amount of air still retained in the lungs after maximum exhalation, which helps to determine whether there are problems such as airway obstruction. These respiratory parameters collectively reflect the changes in lung volume, the strength of the respiratory muscles, and the patency of the airways, constituting a comprehensive assessment of lung ventilation function and serving as an important basis for judging the health of the respiratory system.

[0033] In the above-mentioned pulmonary ventilation function detection system 100 based on algorithm analysis, the embedded coding module 120 is used to embed the various data items in the respiratory parameters to obtain a set of respiratory parameter data item embedded coding vectors. It should be understood that since the various data items in the respiratory parameters have different physical meanings and dimensions, directly performing numerical analysis may cause information loss or inaccurate analysis results. Therefore, in order to unify the data dimension, the present application uses an embedded encoder based on a fully connected layer to perform low-dimensional embedded coding on the various data items in the respiratory parameters, so as to map the respiratory parameter data of different dimensions to a unified low-dimensional vector space, while reducing the data dimension, retaining the core semantic information of the data, so as to generate a set of respiratory parameter data item embedded coding vectors with semantic information, thereby laying the foundation for subsequent data association analysis.

[0034] In the above-mentioned pulmonary ventilation function detection system 100 based on algorithm analysis, the feature enhancement module 130 is used to perform feature enhancement processing on the set of respiratory parameter data item embedded coding vectors to obtain a set of respiratory parameter data item embedded enhanced coding vectors. It should be understood that, considering that the various data items in the respiratory parameters usually have different importance, for example, in most cases, the change of tidal volume may better reflect the patient's respiratory state than the supplementary inspiratory volume or supplementary expiratory volume. Therefore, in order to improve the accuracy of the patient's respiratory state analysis, the present application further performs explicit modeling on the set of respiratory parameter data item embedded coding vectors, and introduces an attention mechanism to strengthen the key information that can better reflect the patient's respiratory state, thereby improving the overall diagnostic performance. It is worth mentioning that the attention mechanism is an important technology in the field of deep learning, which is realized by calculating the importance weights of each part of the input data, and the part with a higher weight will receive more attention. The attention score is obtained by calculating the similarity between the query and the key, and then the attention weight is obtained by normalization. Finally, the values ​​are weighted and summed according to these weights to obtain the final output result.

[0035] Figure 3 FIG. 1 is a block diagram of a feature enhancement module in a pulmonary ventilation function detection system based on algorithm analysis according to an embodiment of the present application. Figure 3 As shown, the feature enhancement module 130 includes: a feature significance calculation unit 131, which is used to calculate the semantic significance description factor of each breathing parameter data item embedded coding vector in the set of breathing parameter data item embedded coding vectors to obtain a set of breathing parameter data item significance description factors; a feature modulation unit 132, which is used to perform hierarchical mask modulation on the set of breathing parameter data item embedded coding vectors based on the set of breathing parameter data item significance description factors to obtain the set of breathing parameter data item embedded enhanced coding vectors.

[0036] Specifically, the feature significance calculation unit 131 is used to calculate the semantic significance description factor of each breathing parameter data item embedded coding vector in the set of the breathing parameter data item embedded coding vector to obtain a set of breathing parameter data item significance description factors, including: a hyperbolic space mapping subunit, used to map each breathing parameter data item embedded coding vector in the set of the breathing parameter data item embedded coding vector to a hyperbolic space to obtain a set of mapped breathing parameter data item embedded coding vectors; a feature significance quantization description subunit, used to calculate the semantic significance description factor of each mapped breathing parameter data item embedded coding vector in the set of the mapped breathing parameter data item embedded coding vector to obtain a set of breathing parameter data item significance description factors. In a specific example of the present application, the squared Poincare norm of each mapped breathing parameter data item embedded coding vector in the set of the mapped breathing parameter data item embedded coding vector is calculated as the semantic significance description factor to obtain the set of breathing parameter data item significance description factors.

[0037] Specifically, the feature modulation unit 132 is used to perform hierarchical mask modulation on the set of breathing parameter data item embedding coding vectors based on the set of breathing parameter data item significance description factors to obtain the set of breathing parameter data item embedding enhancement coding vectors, including: a hierarchical mask sub-unit, used to perform hierarchical mask modulation on the set of breathing parameter data item significance description factors to obtain a set of breathing parameter data item significance modulation weights; a weighted modulation sub-unit, used to use each breathing parameter data item significance modulation weight in the set of breathing parameter data item significance modulation weights as a weight, and respectively perform weighted modulation on the corresponding breathing parameter data item embedding coding vectors in the set of breathing parameter data item embedding coding vectors to obtain the set of breathing parameter data item embedding enhancement coding vectors. Among them, the hierarchical mask sub-unit is used to: input the set of significance description factors of the breathing parameter data items into a normalized process based on a softmax function to obtain a set of normalized significance description factors of the breathing parameter data items; input the set of normalized significance description factors of the breathing parameter data items into a multi-level mask function to obtain a set of significance modulation weights of the breathing parameter data items. More specifically, the set of normalized respiratory parameter data item significance description factors is input into a multi-level mask function to obtain a set of significance modulation weights of the respiratory parameter data items, including: setting a first mask threshold and a second mask threshold, wherein the second mask threshold is twice the first mask threshold; comparing each normalized respiratory parameter data item significance description factor in the set of normalized respiratory parameter data item significance description factors with the first mask threshold and the second mask threshold, if the normalized respiratory parameter data item significance description factor is greater than the second mask threshold, weighted amplifying it; if the normalized respiratory parameter data item significance description factor is greater than the first mask threshold and less than or equal to the second mask threshold, keeping it unchanged; if the normalized respiratory parameter data item significance description factor is less than or equal to the first mask threshold, weighted suppressing it, thereby obtaining a set of significance modulation weights of the respiratory parameter data items.

[0038] Accordingly, the set of breathing parameter data item embedded coding vectors is processed by the following feature enhancement formula to obtain the set of breathing parameter data item embedded enhanced coding vectors, wherein the feature enhancement formula is:

[0039] X={x1,x2,…,x i , …, x n}

[0040] h i =W1x i W2

[0041]

[0042] w i =mask(o i ′)

[0043]

[0044] Y={y1,y2,...,y i , ..., y n}

[0045] y i =w i ·x i

[0046] Wherein, X represents the set of embedding coding vectors of the respiratory parameter data items, x1, x2, x i and x n represents the first, second, i-th and n-th breathing parameter data item embedded coding vectors in the set of breathing parameter data item embedded coding vectors, respectively, n is the number of breathing parameter data item embedded coding vectors, W1 and W2 are different weight matrices used to project the breathing parameter data item embedded coding vector from Euclidean space to hyperbolic space, h i Embedding encoding vector for the i-th mapped respiratory parameter data item, ||·|| 2 represents the square of the norm of the eigenvector, log represents the logarithmic function with base 2, exp(·) represents the exponential function with base e, o i represents the significant descriptive factor of the i-th respiratory parameter data item, o i ′ represents the significant descriptor of the i-th normalized respiratory parameter data item, θ represents the first mask threshold, λ and η are different weight parameters, and λ>1, η<1, mask(·) is a multi-level mask function, w i is the significance modulation weight of the i-th respiratory parameter data item, Y is the set of embedded enhanced coding vectors of the respiratory parameter data item, y1, y2, y i and n They respectively represent the first, second, i-th and n-th breathing parameter data item embedded enhanced coding vectors in the set of breathing parameter data item embedded enhanced coding vectors.

[0047] It should be understood that, considering that compared with Euclidean space, hyperbolic space can embed more hierarchical information more effectively, which helps to better understand the semantic hierarchy between each data item. Therefore, in the present application, each breathing parameter data item embedding coding vector is first converted from Euclidean space to hyperbolic space. In hyperbolic space, as the distance increases, the spatial volume grows exponentially. Based on this characteristic, the semantic hierarchy between each data item can be better maintained. Then, in the hyperbolic space, the square Poincare norm of the embedding coding vector of each breathing parameter data item is further calculated to measure the importance or contribution of the feature, so as to obtain a set of significant description factors of the breathing parameter data item. Then, the set of significant description factors of the breathing parameter data item is subjected to three-layer hierarchical mask modulation, and the semantic significance description factors of different levels are subjected to feature scaling based on different scales, so as to retain the semantic features of different data items while paying attention to different degrees of features, so as to better highlight important features and suppress irrelevant features. Finally, the obtained breathing parameter data item significance modulation weight is used to perform weighted modulation on the original breathing parameter data item embedding coding vector, thereby obtaining a set of more representative and discriminative breathing parameter data item embedding enhanced coding vectors.

[0048] In the above-mentioned pulmonary ventilation function detection system 100 based on algorithm analysis, the global deep semantic association coding module 140 is used to perform global deep semantic association coding on the set of respiratory parameter data items embedded in the enhanced coding vector based on the semantic association topological relationship between the various data items in the respiratory parameters to obtain a global deep semantic coding feature vector of the respiratory state. It should be understood that, considering that in pulmonary ventilation function detection, the association between the various respiratory parameter data items has an important indicative significance for the diagnostic result. For example, the joint feature of tidal volume and inspiratory supplement volume can indicate the state of the patient's respiratory muscle strength, while the combination of expiratory supplement volume and residual volume may reveal the degree of airway obstruction. Therefore, in order to achieve a comprehensive analysis of the patient's respiratory state, the present application further performs semantic association analysis on the various data items in the respiratory parameters.

[0049] Figure 4 FIG. 1 is a block diagram of a global deep semantic association encoding module in a pulmonary ventilation function detection system based on algorithm analysis according to an embodiment of the present application. Figure 4As shown, the global deep semantic association coding module 140 includes: a breathing parameter adjacency topology matrix construction unit 141, which is used to construct a breathing parameter adjacency topology matrix based on the semantic association topological relationship between each data item in the breathing parameter; a breathing state global feature extraction unit 142, which is used to embed the breathing parameter data items into a set of enhanced coding vectors and the breathing parameter adjacency topology matrix and input them into a breathing state global deep semantic feature extractor based on a graph convolutional neural network model to obtain the breathing state global deep semantic coding feature vector.

[0050] Specifically, the breathing parameter adjacency topology matrix construction unit 141 includes: an adjacency matrix construction subunit, which is used to construct an adjacency matrix of the set of breathing parameter data items embedded in the coding vectors, and the eigenvalues ​​of each non-diagonal position in the adjacency matrix are the breathing state semantic association coefficients between the corresponding two breathing parameter data items embedded in the coding vectors; an adjacency topology feature extraction subunit, which is used to input the adjacency matrix of the set of breathing parameter data items embedded in the coding vectors into a breathing parameter adjacency topology feature extractor based on a void convolution layer to obtain the breathing parameter adjacency topology matrix.

[0051] That is, first, construct the adjacency matrix of the set of the breathing parameter data items embedded in the coding vector, and reveal the potential association pattern between each breathing parameter from a global perspective. Wherein, the eigenvalues ​​of each position on the non-diagonal position in the adjacency matrix are the semantic association coefficients of the breathing state between the corresponding two breathing parameter data items embedded in the coding vector, reflecting the semantic association strength between the two. In a specific example of the present application, cosine similarity is used as a calculation method for the semantic association coefficient of the breathing state to measure the semantic association between different breathing parameters. Secondly, in order to further extract the deeper topological structure information in the adjacency matrix, so as to enhance the understanding of the association pattern of each breathing parameter data item, the present application further uses a hole convolution network to process the adjacency matrix of the set of the breathing parameter data item embedded in the coding vector. The hole convolution network can expand the receptive field while retaining the local topological structure information by introducing holes in the convolution kernel, so as to capture a wider range of contextual information, learn more global breathing parameter association structure features, and generate a breathing parameter adjacency topological matrix, so as to achieve a deep understanding of the nonlinear association pattern between each breathing parameter data item.

[0052] Specifically, the respiratory state global feature extraction unit 142 is used to embed the set of enhanced coding vectors and the respiratory parameter adjacency topology matrix into the respiratory state global deep semantic feature extractor based on the graph convolutional neural network model to obtain the respiratory state global deep semantic coding feature vector. That is, the graph convolutional neural network model is used to embed the set of enhanced coding vectors and the respiratory parameter adjacency topology matrix into the respiratory parameter data item to perform graph structure learning to achieve a comprehensive perception of the patient's respiratory state. Here, the enhanced coding vectors embedded in each respiratory parameter data item are used as node features, and the eigenvalues ​​in the respiratory parameter adjacency topology matrix are used as the connection weights between the corresponding nodes to construct a graph structure. The graph convolutional neural network can effectively capture the complex relationships between nodes by performing convolution operations on the graph structure, learn the deep associations between the various respiratory parameter data items, and make full use of the topological information in the respiratory parameter adjacency topology matrix to gradually aggregate node features, thereby achieving a global perception of the patient's respiratory state and generating a global deep semantic coding feature vector for the respiratory state.

[0053] In the above-mentioned pulmonary ventilation function detection system 100 based on algorithm analysis, the pulmonary ventilation function detection module 150 is used to determine whether the patient suffers from bronchial asthma based on the global deep semantic coding feature vector of the respiratory state, and is further used to: input the global deep semantic coding feature vector of the respiratory state into a respiratory state evaluator based on a classifier to obtain an evaluation result, and the evaluation result is used to indicate whether the patient suffers from bronchial asthma. It is worth mentioning that a classifier is an algorithm or model used in the field of machine learning to classify data into known categories, which adjusts its internal parameters through back propagation and optimization algorithms (such as Adam, SGD, etc.) to improve the accuracy of classification judgment. In the actual application stage, the global deep semantic coding feature vector of the respiratory state is input into the trained classifier, and the classifier can parse and classify it according to the learned classification rules to determine whether the patient suffers from bronchial asthma.

[0054] In the technical scheme of the present application, each breathing parameter data item embedded in the set of the breathing parameter data item enhanced coding vector is embedded in the enhanced coding vector to represent the enhanced embedded coding feature of the breathing parameter data item. The breathing parameter adjacency topology matrix represents the adjacency matrix of the set of the breathing parameter data item embedded coding vector, that is, the adjacency topology distribution of the sparse semantic association of the breathing state between any two breathing parameter data item embedded coding vectors. When it is input into the breathing state global deep semantic feature extractor based on the graph convolutional neural network model, the obtained breathing state global deep semantic coding feature vector not only includes the breathing state semantic adjacency topology feature, but also includes the high-dimensional implicit feature of the breathing parameter data item under the modulation of the adjacency topology feature. However, there are spatial structure and depth differences between the adjacency topology feature and the embedded coding feature, which makes the feature manifold of the obtained breathing state global deep semantic coding feature vector have expression sparsity and uneven distribution in the high-dimensional feature space, thereby affecting the accuracy of the evaluation result obtained by the breathing state evaluator based on the classifier.

[0055] Preferably, in an example of the present application, the global deep semantic coding feature vector of the respiratory state is input into a classifier-based respiratory state evaluator to obtain an evaluation result, including: performing L random sampling on the global deep semantic coding feature vector of the respiratory state to obtain L respiratory state evaluation random perturbation eigenvalues; arranging the L respiratory state evaluation random perturbation eigenvalues ​​in order from large to small to obtain a respiratory state evaluation random perturbation ordered vector; calculating the self-inner product of the global deep semantic coding feature vector of the respiratory state to obtain a respiratory state evaluation energy modulation eigenvalue; arranging the respiratory state evaluation random perturbation ordered vector based on the respiratory state evaluation energy modulation eigenvalue; The vector is semantically scaled to obtain a semantically enhanced randomly perturbed ordered vector for respiratory state evaluation; the product between the semantically enhanced randomly perturbed ordered vector for respiratory state evaluation and the transposed vector of the respiratory state global deep semantic coding feature vector is calculated to obtain an orthogonal modulation matrix for best sample prompt for respiratory state evaluation; the matrix product between the orthogonal modulation matrix for best sample prompt for respiratory state evaluation and the respiratory state global deep semantic coding feature vector is calculated to obtain an optimized global deep semantic coding feature vector for respiratory state; the optimized global deep semantic coding feature vector for respiratory state is input into the classifier-based respiratory state evaluator to obtain the evaluation result.

[0056] Accordingly, the optimization process of the global deep semantic coding feature vector of the respiratory state is expressed by the following optimization formula:

[0057] S i =∈ i , i = 1, 2, ..., L

[0058]

[0059] u=average(V)

[0060]

[0061] V order =Order{S i}

[0062]

[0063] Wherein, V is the global deep semantic encoding feature vector of the respiratory state, represents the normal distribution, ∈ i represents a random variable that follows a normal distribution, L represents the number of random sampling, u and σ 2 They represent the feature mean and feature variance of the global deep semantic encoding feature vector of the respiratory state, S i represents the random perturbation eigenvalues ​​of each respiratory state evaluation obtained by L random samplings of the global deep semantic encoding feature vector of the respiratory state, V order is the ordered vector of random disturbances in the respiratory state assessment, represents the self-inner product of the calculated eigenvector, e is the energy modulation eigenvalue of the respiratory state assessment, M t is the best sample prompt orthogonal modulation matrix for respiratory state assessment, ⊙ represents the point multiplication by position, represents vector multiplication, V′ is the optimized global deep semantic encoding feature vector of the respiratory state.

[0064] That is, by randomly sampling the global deep semantic encoding feature vector of the respiratory state L times to obtain a set of new feature values ​​based on the original features but containing slight changes, randomness is introduced to explore different aspects of the feature space, and by simulating the noise or variation in the actual data, it helps the model to better generalize to unseen data. Then, these respiratory state evaluation random perturbation feature values ​​are arranged in order from large to small to form a respiratory state evaluation random perturbation ordered vector to identify which perturbation directions may be the most important for the current problem. Such an ordering not only helps the algorithm focus on the feature differences that can best distinguish different categories, but also helps to reveal hidden data structures or patterns. Subsequently, the inner product of the global deep semantic encoding feature vector of the respiratory state is calculated to obtain the energy modulation feature value of the respiratory state assessment. The energy modulation feature value of the respiratory state assessment reflects the overall strength or "energy" of the original feature vector, and provides a benchmark for measuring the importance of the feature. Using this energy modulation feature value of the respiratory state assessment as a reference point, the previously formed random perturbation ordered vector of the respiratory state assessment is semantically scaled to construct a semantically enhanced random perturbation ordered vector of the respiratory state assessment. In this way, those disturbances that appear to be more important in the original feature space are amplified, thereby enhancing the model's sensitivity to changes in key features, while also retaining relatively small but potentially useful disturbance information.

[0065] Then, the product of the semantically enhanced random perturbation ordered vector of respiratory state assessment and the transpose of the global deep semantic encoding feature vector of respiratory state is calculated to obtain the best sample prompt orthogonal modulation matrix of respiratory state assessment. The purpose is to find the interaction between the features under two different perspectives. The formed matrix can be regarded as a conversion tool for capturing and representing the complex linear relationship between features. Finally, the final feature structure modulation is completed by applying the best sample prompt orthogonal modulation matrix of respiratory state assessment constructed above to the original global deep semantic encoding feature vector of respiratory state, so that the features are more in line with the requirements of the objective function, reducing redundant information, while maximizing the retention of useful information, so as to improve learning efficiency and prediction performance.

[0066] Therefore, in a preferred example, the feature manifold optimization process of the global deep semantic encoding feature vector of the respiratory state is explained, which extracts the best prompt sample based on the fine-grained target retrieval of random perturbations, and then guides adversarial robustness learning based on the best prompt sample to complete the structure of the high-dimensional manifold of the feature vector in the feature space to improve the detail and structure of the feature expression. In this way, the accuracy of the evaluation result obtained by the classifier-based respiratory state evaluator is improved.

[0067] In summary, a pulmonary ventilation function detection system based on algorithm analysis is explained, which uses data processing technology based on deep learning to analyze the patient's respiratory parameters, extracts the semantic embedding feature representation of each respiratory parameter data item, and performs global deep association encoding on the semantic embedding features of each respiratory parameter data item based on the semantic association topological relationship between each respiratory parameter data item, thereby achieving global deep perception of the patient's respiratory state, so as to intelligently identify whether the patient suffers from bronchial asthma. In this way, the limitations of traditional data analysis methods can be effectively overcome, the dependence on doctors' professional knowledge can be reduced, the accuracy and efficiency of pulmonary ventilation function detection can be improved, and the risk of misdiagnosis and missed diagnosis can be reduced.

[0068] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0069] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0070] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0071] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0072] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A pulmonary ventilation function detection system based on algorithm analysis, characterized in that: include: A respiratory parameter monitoring module, used to obtain respiratory parameters of the patient subject collected by the spirometer, wherein the respiratory parameters include tidal volume, inspiratory reserve volume, expiratory reserve volume, deep inspiratory volume and residual volume; An embedded coding module, used for embedding and coding each data item in the breathing parameter to obtain a set of embedded coding vectors of the breathing parameter data items; A feature enhancement module, used for performing feature enhancement processing on the set of breathing parameter data item embedded coding vectors to obtain a set of breathing parameter data item embedded enhanced coding vectors; A global deep semantic association coding module, used for performing global deep semantic association coding on the set of embedded enhanced coding vectors of the breathing parameter data items based on the semantic association topological relationship between the data items in the breathing parameters to obtain a global deep semantic coding feature vector of the breathing state; A pulmonary ventilation function detection module, used to determine whether the patient suffers from bronchial asthma based on the global deep semantic encoding feature vector of the respiratory state; Among them, the feature enhancement module includes: a feature significance calculation unit, which is used to calculate the semantic significance description factor of each breathing parameter data item embedded coding vector in the set of breathing parameter data item embedded coding vectors to obtain a set of breathing parameter data item significance description factors; a feature modulation unit, which is used to perform hierarchical mask modulation on the set of breathing parameter data item embedded coding vectors based on the set of breathing parameter data item significance description factors to obtain a set of breathing parameter data item embedded enhanced coding vectors.

2. The pulmonary ventilation function detection system based on algorithm analysis according to claim 1, characterized in that: The embedded coding module is used for: A fully connected layer-based embedding encoder is used to perform low-dimensional embedding encoding on each data item in the breathing parameter to obtain a set of embedded coding vectors of the breathing parameter data items.

3. The pulmonary ventilation function detection system based on algorithm analysis according to claim 2, characterized in that: The feature significance calculation unit comprises: A hyperbolic space mapping subunit, used for mapping each breathing parameter data item embedded coding vector in the set of breathing parameter data item embedded coding vectors to a hyperbolic space to obtain a set of mapped breathing parameter data item embedded coding vectors; The feature significance quantization description subunit is used to calculate the semantic significance description factor of each mapped breathing parameter data item embedded coding vector in the set of mapped breathing parameter data item embedded coding vectors to obtain the set of breathing parameter data item significance description factors.

4. The pulmonary ventilation function detection system based on algorithm analysis according to claim 3, characterized in that: The feature significance quantization description subunit is used to: The square Poincare norm of each mapped breathing parameter data item embedding coding vector in the set of the mapped breathing parameter data item embedding coding vector is calculated as the semantic significance description factor to obtain the set of the breathing parameter data item significance description factors.

5. The pulmonary ventilation function detection system based on algorithm analysis according to claim 4, characterized in that: The characteristic modulation unit comprises: A hierarchical mask subunit, configured to perform hierarchical mask modulation on the set of significance description factors of the respiratory parameter data item to obtain a set of significance modulation weights of the respiratory parameter data item; The weighted modulation subunit is used to use the significance modulation weight of each breathing parameter data item in the set of the breathing parameter data item significance modulation weights as a weight, and respectively perform weighted modulation on the corresponding breathing parameter data item embedding coding vector in the set of the breathing parameter data item embedding coding vector to obtain the set of the breathing parameter data item embedding enhanced coding vector.

6. The pulmonary ventilation function detection system based on algorithm analysis according to claim 5, characterized in that: The hierarchical mask subunit is used for: The set of significance description factors of the respiratory parameter data items is input and normalized based on a softmax function to obtain a set of normalized significance description factors of the respiratory parameter data items; The set of normalized respiratory parameter data item significance description factors is input into a multi-level mask function to obtain a set of respiratory parameter data item significance modulation weights.

7. The pulmonary ventilation function detection system based on algorithm analysis according to claim 6, characterized in that: The global deep semantic association encoding module includes: A breathing parameter adjacency topology matrix construction unit, used to construct a breathing parameter adjacency topology matrix based on the semantic association topological relationship between each data item in the breathing parameter; The respiratory state global feature extraction unit is used to embed the respiratory parameter data item into a set of enhanced coding vectors and the respiratory parameter adjacency topology matrix and input them into a respiratory state global deep semantic feature extractor based on a graph convolutional neural network model to obtain the respiratory state global deep semantic coding feature vector.

8. The pulmonary ventilation function detection system based on algorithm analysis according to claim 7, characterized in that: The breathing parameter adjacency topology matrix construction unit comprises: An adjacency matrix construction subunit is used to construct an adjacency matrix of the set of breathing parameter data items embedded in the coding vectors, wherein the eigenvalue of each non-diagonal position in the adjacency matrix is ​​a breathing state semantic association coefficient between the corresponding two breathing parameter data items embedded in the coding vectors; The adjacency topology feature extraction subunit is used to input the adjacency matrix of the set of the breathing parameter data items embedded in the encoding vector into the breathing parameter adjacency topology feature extractor based on the void convolution layer to obtain the breathing parameter adjacency topology matrix.

9. The pulmonary ventilation function detection system based on algorithm analysis according to claim 8, characterized in that: The pulmonary ventilation function detection module is used for: The global deep semantic encoding feature vector of the respiratory state is input into a classifier-based respiratory state evaluator to obtain an evaluation result, and the evaluation result is used to indicate whether the patient suffers from bronchial asthma.

Images