A bronchial asthma phenotype identification index screening method based on exosomeomics

Abstract

The application discloses a bronchial asthma phenotype identification index screening method based on exosomeomics, comprising the following steps: acquiring bronchial asthma medical record data to construct a medical record data subset, and extracting multi-dimensional information to obtain diagnostic characteristics; generating an original feature set by using common diagnostic characteristics of different bronchial asthma phenotypes; constructing a knowledge graph according to exosomeomics related knowledge and the action mechanism of exosomes of different cell sources on bronchial asthma, screening exosome characteristics, and expanding the original feature set; constructing a feature selection task based on multi-objective optimization, obtaining exosome feature combinations of different bronchial asthma on the basis of the common diagnostic characteristics, and determining exosome identification indexes of bronchial asthma phenotypes after comparison and verification of the identification accuracy of the original common diagnostic characteristics. The exosomeomics indexes highly related to bronchial asthma diagnosis are screened, biomarkers are obtained for predicting allergic and asthma conditions, and the identification accuracy of bronchial asthma phenotypes is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of feature selection, and more particularly to an exosome-based bronchial asthma phenotype identification index screening method. BACKGROUND

[0002] Bronchial asthma is a complex heterogeneous disease with significant heterogeneity in etiology, clinical characteristics and response to treatment. Several large multi-center studies have used different clustering techniques to identify bronchial asthma phenotypes, although the disease characteristics used in each cluster, the calculation method and the number of clusters studied differ between these studies, but these results consistently show that there are about 4 main phenotypes of bronchial asthma: early onset mild allergic bronchial asthma, early onset moderate to severe allergic remodeling bronchial asthma, late onset non-eosinophilic non-allergic bronchial asthma and late onset non-allergic eosinophilic bronchial asthma. Phenotype is an observable feature based on the relationship between genes and environment, which depends on clinical, biochemical, functional and pharmacological characteristics, and the phenotype may change over time, however, it represents the first standard for distinguishing patients and a potential predictor of treatment response.

[0003] Exosomes are nanoscale vesicles shed from the cell membrane or secreted by cells, have a phospholipid bilayer structure, can mediate cell-to-cell communication, have various biological functions, and are involved in inflammation, immune response, angiogenesis, cell death, antigen presentation, neurodegenerative diseases, tumor proliferation and other processes. Studies have found that exosomes are involved in the development of house dust mite-induced asthma, and the levels of eosinophils, IgE and inflammatory cytokines are reduced after the exosome inhibitor acts on the asthma mouse, indicating that inhibiting exosome secretion can alleviate asthma characteristics. The etiology and pathogenesis of asthma have not been fully elucidated, and its clinical treatment has certain limitations, so it is particularly important to study the pathogenesis of asthma and establish an effective phenotype identification method. Therefore, how to screen exosome-based bronchial asthma phenotype identification index based on feature selection is a problem to be solved. SUMMARY

[0004] In order to solve the above technical problems, the present application provides an exosome-based bronchial asthma phenotype identification index screening method to solve the problems of single bronchial asthma phenotype identification index in the prior art, inability to timely capture changes in airway inflammation phenotype and the like, and provides a new idea for bronchial asthma phenotype identification.

[0005] The present application provides an exosome-based bronchial asthma phenotype identification index screening method, comprising:

[0006] Obtaining a plurality of bronchial asthma medical record data, constructing medical record data subsets based on different diagnostic phenotype categories, extracting multi-dimensional information in different medical record data subsets, and obtaining diagnostic characteristics of different bronchial asthma phenotypes;

[0007] Obtaining common diagnostic characteristics and common diagnostic characteristic distribution of different bronchial asthma phenotypes, and generating an original feature set based on the common diagnostic characteristics;

[0008] Retrieving exosome omics related knowledge according to a big data method, constructing a knowledge graph on the mechanism of bronchial asthma by exosomes from different cells, screening exosome features with importance meeting preset requirements by using the knowledge graph, and expanding the original feature set;

[0009] Constructing a feature selection task based on multi-objective optimization in the expanded original feature set, obtaining exosome feature combinations of different bronchial asthma phenotypes based on common diagnostic characteristics, and comparing and verifying identification accuracy according to the common diagnostic characteristic distribution, and determining exosome identification indicators of bronchial asthma phenotypes through the verification result.

[0010] In this scheme, medical record data subsets are constructed based on different diagnostic phenotype categories, and multi-dimensional information is extracted in different medical record data subsets, specifically:

[0011] According to bronchial asthma as a keyword, medical record data of bronchial asthma meeting the search label constraint is obtained, and diagnostic information is extracted, and bronchial asthma phenotypes are obtained according to the diagnostic information;

[0012] The corresponding bronchial asthma medical record data is aggregated by using the bronchial asthma phenotype, and medical record data subsets of different bronchial asthma phenotypes are constructed, detection indicators in different medical record data subsets are extracted, and abnormal indicators are screened;

[0013] The diagnostic basis is determined according to the abnormal indicators combined with the clinical symptoms in the medical record data, the low-dimensional features are obtained by using single-channel convolution on the diagnostic basis, and the high-dimensional features are obtained by using convolution calculation, normalization processing and activation operation based on the main branch of the low-dimensional features;

[0014] Meanwhile, the low-dimensional features are locally integrated by using average pooling processing based on the side branch, the outputs of the two branches are fused, multi-dimensional information is obtained, and diagnostic characteristics of different bronchial asthma phenotypes are generated according to the multi-dimensional information.

[0015] In this scheme, common diagnostic characteristics and common diagnostic characteristic distribution of different bronchial asthma phenotypes are obtained, and an original feature set is generated based on the common diagnostic characteristics, specifically:

[0016] Obtaining a diagnostic feature set of different bronchial asthma phenotypes, taking the intersection of each diagnostic feature set, extracting common diagnostic features to generate an original feature set, and extracting corresponding examination data from each medical record data subset according to the common diagnostic features in the original feature set;

[0017] Mapping the examination data corresponding to the common diagnostic features to a low-dimensional space, obtaining an exponential distribution of the examination data in the low-dimensional space to generate a common diagnostic feature distribution, and obtaining an identification accuracy according to the common diagnostic feature distribution of different bronchial asthma phenotypes.

[0018] In this scheme, the knowledge graph is constructed by searching the knowledge graph based on the big data method and the action mechanism of the exosome of different cell sources on bronchial asthma.

[0019] The exosome proteomics professional corpus and exosome proteomics and bronchial asthma big data information are obtained based on the search engine through the big data method, and the feature vectors of the exosome proteomics related knowledge are obtained through structured processing and using an encoder model to capture deep feature information and obtain entities.

[0020] The semantic features are obtained by embedding based on the exosome proteomics and bronchial asthma big data information, the action mechanism of the exosome of different cell sources on bronchial asthma is read through the semantic features, the relationship between entities is extracted based on the action mechanism, and a relationship table is constructed.

[0021] The obtained entities and the relationship table are mapped into a triple form to construct a knowledge graph in a graphical manner.

[0022] In this scheme, the knowledge graph is used to screen exosome features that meet the preset requirements, and the original feature set is expanded, specifically as follows:

[0023] For each entity in the knowledge graph, the number of entities directly connected to the entity is obtained, the entity contribution degree is obtained according to the number of entities, and the adjacency matrix is constructed according to the entity contribution degree, and the knowledge graph is represented in the form of the adjacency matrix.

[0024] The knowledge graph is represented and learned using a graph neural network, the features of the entities in the knowledge graph are aggregated and propagated through neighbor entities, the entity embedding vector is constructed, the entity embedding vector is imported into a GRU unit to obtain the relationship between entities, and the centrality of the entity is obtained using a Softmax function.

[0025] The importance score of the entity is obtained according to the centrality and the initial score, the entities are sorted according to the importance score of the entity, the exosome features that meet the preset requirements are selected, the feature redundancy is reduced through the correlation calculation between the exosome features, and the original feature set is expanded.

[0026] In the scheme, a feature selection task based on multi-objective optimization is constructed in the expanded original feature set, specifically:

[0027] The expanded original feature set is obtained, and the feature importance and feature redundancy are used as evaluation indexes, and the maximum feature importance and minimum feature redundancy are used as optimization objectives;

[0028] A bronchial asthma phenotype classification model is constructed based on a convolutional autoencoder and a GUR unit, a dataset is constructed based on the expanded original feature set and the corresponding examination data, the dataset is divided into several subsets, each subset is used as a validation set, and the remaining subsets are used as training sample sets to train the bronchial asthma phenotype classification model, and several models are obtained;

[0029] The convolutional autoencoder part is composed of stacked convolutional layers, pooling layers and batch normalization layers, the convolutional autoencoder parameters are initialized, and the encoder part is trained in an unsupervised training manner to obtain the recognition classification features of the output of the encoder part;

[0030] The recognition classification features are imported into the GRU unit to obtain the time correlation of the features, the bronchial asthma phenotype classification is realized by deep learning of the time sequence features, and the average classification accuracy of the several bronchial asthma phenotype classification models is obtained as an importance evaluation index;

[0031] The mRMR algorithm is reconstructed using the Pearson correlation coefficient, the correlation between the selected features and the correlation between the features and the bronchial asthma phenotype are calculated, and the redundancy evaluation index of the feature combination is obtained.

[0032] In the scheme, the exosome feature combination of different bronchial asthma phenotypes is obtained based on the common diagnostic features, specifically:

[0033] The optimal exosome feature combination is obtained according to the optimization target using the optimized eagle algorithm, the eagle algorithm parameters are initialized, the expanded original feature set is encoded, and the common diagnostic features are used as the feature reference to randomly generate an initial feature combination as an eagle individual on the feature reference;

[0034] The fitness of the current position of the initialized eagle population is calculated, the global optimal fitness value and the optimal position are updated, a preset iteration number threshold is set, when the iteration number does not exceed the preset iteration number threshold, the search is expanded for global optimization, otherwise, the search is narrowed for local search;

[0035] An adaptive control factor is introduced to adjust the flight speed of the eagle individual in global optimization and local search, the fitness value of the new position of the eagle individual is calculated, the feature combination with the highest fitness value is obtained, the recombination of the feature combination is strengthened by differential calculation, and the global optimal fitness value and the optimal position are recorded;

[0036] When the objective function of the feature combination converges, an optimal feature combination is obtained, and the common diagnostic features are removed in the optimal feature combination to obtain an exosome feature combination.

[0037] In the scheme, the identification accuracy is compared and verified according to the common diagnostic feature distribution, and the exosome identification index of the bronchial asthma phenotype is determined according to the verification result, specifically:

[0038] The exosome feature distribution is obtained by mapping the exosome feature combination data to a low-dimensional space based on the exosome feature combination data obtained based on the bronchial asthma medical record data, and the corresponding identification accuracy is obtained according to the exosome feature distribution of different bronchial asthma phenotypes.

[0039] The identification accuracy corresponding to the common diagnostic feature distribution of different bronchial asthma phenotypes is obtained, and the deviation of the identification accuracy corresponding to the exosome feature distribution and the identification accuracy corresponding to the common diagnostic feature distribution is compared.

[0040] If the deviation is greater than a preset deviation threshold, the exosome features of the current bronchial asthma phenotype are reobtained, and if the deviation is not greater than the preset deviation threshold, the exosome feature combination is output.

[0041] Compared with the prior art, the beneficial effects of the present application are:

[0042] The present application can screen exosome omics indexes highly related to bronchial asthma diagnosis, obtain biomarkers for predicting allergic and asthma conditions, and provide reliable auxiliary references for bronchial asthma related diagnosis and treatment.

[0043] The present application improves the accuracy of bronchial asthma phenotype identification by screening exosome identification indexes, improves the efficiency of diagnosis and identification, provides a simple and easy means to judge airway inflammation phenotype, and timely captures the changes of airway inflammation phenotype, which has important significance for the treatment and prognosis of asthma. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments or examples of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or examples. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0045] Figure 1 The flowchart of the bronchial asthma phenotype identification index screening method based on exosome omics of the present application is shown;

[0046] Figure 2A flow chart showing the present application expanding the original feature set with exosome features;

[0047] Figure 3 A flow chart showing the present application obtaining exosome feature combinations of different bronchial asthma phenotypes;

[0048] Figure 4 A block diagram showing the present application exosome-based bronchial asthma phenotype identification index screening system.

[0049] The purposes, functional characteristics and advantages of the present application will be further described with reference to the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0050] In order to more clearly understand the above-mentioned purposes, features and advantages of the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0051] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, however, the present application can also be implemented in other ways different from those described herein, therefore, the protection scope of the present application is not limited by the specific embodiments disclosed below.

[0052] Figure 1 A flow chart showing the present application exosome-based bronchial asthma phenotype identification index screening method;

[0053] As shown in Figure 1 The first embodiment of the present application provides an exosome-based bronchial asthma phenotype identification index screening method, which comprises:

[0054] S102, obtaining a plurality of bronchial asthma medical record data, constructing a medical record data subset based on different diagnostic phenotype categories, extracting multi-dimensional information in different medical record data subsets, and obtaining diagnostic features of different bronchial asthma phenotypes;

[0055] S104, obtaining common diagnostic features and common diagnostic feature distributions of different bronchial asthma phenotypes, and generating an original feature set based on the common diagnostic features;

[0056] S106, retrieving exosome-related knowledge according to a big data method, constructing a knowledge graph through the mechanism of exosomes of different cell sources on bronchial asthma, and using the knowledge graph to screen exosome features with importance meeting preset requirements, and expanding the original feature set;

[0057] S108, construct a feature selection task based on multi-objective optimization in the expanded original feature set, obtain an exosome feature combination of different bronchial asthma phenotypes based on the common diagnostic features, and perform identification accuracy comparison and verification according to the common diagnostic feature distribution, and determine the exosome identification index of the bronchial asthma phenotype through the verification result.

[0058] It should be noted that, access to different sources of medical record database as a retrieval space, according to bronchial asthma as a keyword to search medical record data, get bronchial asthma medical record data that meet the search label constraints, and extract diagnostic information, obtain bronchial asthma phenotypes according to the diagnostic information; use the bronchial asthma phenotype to aggregate the corresponding bronchial asthma medical record data, construct medical record data subsets of different bronchial asthma phenotypes, extract the detection indexes involved in different medical record data subsets, select the indexes that do not meet the examination standard range as abnormal indexes according to the detection indexes; determine the diagnostic basis according to the abnormal indexes combined with the clinical symptoms such as asthma, cough and chest tightness in the medical record data, use single channel convolution to filter the abnormal index examination data and symptom information of different sources, obtain low-dimensional features, and ensure that there is no data omission when extracting features from all abnormal index examination data and symptom information. The main branch introduces CNN structure, obtains high-dimensional features based on the convolution calculation, normalization processing and activation operation of the low-dimensional features of the main branch; at the same time, the low-dimensional features are locally integrated based on the side branch using average pooling processing, the outputs of the two branches are fused to obtain multi-dimensional information, the multi-dimensional features retain more features in the data, and the structure of double branch feature extraction improves the network convergence speed, and the multi-dimensional information is used to generate diagnostic features of different bronchial asthma phenotypes, which represent the detection data features relied on in the diagnosis of bronchial asthma phenotypes.

[0059] It should be noted that, obtain the diagnostic feature set of different bronchial asthma phenotypes, take intersection of each diagnostic feature set to extract common diagnostic features to generate an original feature set, and extract corresponding examination data in each medical record data subset according to the common diagnostic features in the original feature set; map the examination data corresponding to the common diagnostic features to a low-dimensional space to obtain the index distribution of the examination data in the low-dimensional space to generate the common diagnostic feature distribution, and the corresponding probability density function f(x) is: x represents the diagnostic feature, θ represents the mean of identification accuracy, and the corresponding likelihood function L is obtained: y i represents the bronchial asthma phenotype diagnosis result sample, r represents the number of errors of the bronchial asthma phenotype diagnosis result sample, n represents the total number of samples, and the identification accuracy of different bronchial asthma phenotypes is obtained according to the common diagnostic feature distribution by using the maximum likelihood method

[0060] It should be noted that, using big data methods, we obtained exosome-related corpora and big data information on exosomes and bronchial asthma through search engines. After structured processing, we used an encoder model to obtain feature vectors of exosome-related knowledge, capturing deep-level feature information to obtain entities. Since the same exosome entity names may have different languages ​​or expressions in different literature data or research materials, we used big data methods to clarify the different expressions in the exosome-related corpora. Based on the different expressions of the same related terms, we determined equivalent concepts, unified the entities of related terms in the exosome-related corpora and big data information on exosomes and bronchial asthma based on equivalent concepts, and achieved entity resolution through retrieval and matching. After unification, we performed data conflict detection. Based on exomics and big data information on bronchial asthma, semantic features are obtained through methods such as attention embedding. These semantic features are then used to understand the mechanisms of action of exosomes from different cell sources on bronchial asthma. The cell sources are broadly categorized as exosomes from immune effector cells and exosomes from lung structural cells. Relationships between entities are extracted based on these mechanisms, and a relation table is constructed. For example, 12 miRNAs are upregulated in mice with allergic airway inflammation, with miR-233-3p and miR-142a being the most abundant. These miRNAs are also upregulated in the sputum of patients with severe asthma, indicating that their expression levels are correlated with disease severity. That is, exosomes miR-233-3p and miR-142a are associated with the severity of bronchial asthma. The obtained entities and relation tables are mapped to triples to construct a knowledge graph graphically.

[0061] Figure 2 The flowchart of the present invention for expanding the original feature set by screening exosome features is shown.

[0062] According to an embodiment of the present invention, the original feature set is expanded by using the knowledge graph to screen exosome features whose importance meets preset requirements, specifically as follows:

[0063] S202, for each entity in the knowledge graph, obtain the number of entities directly connected to the entity, obtain the entity contribution based on the number of entities, construct an adjacency matrix based on the entity contribution, and represent the knowledge graph in the form of an adjacency matrix;

[0064] S204, use a graph neural network to perform representation learning on the knowledge graph, construct entity embedding vectors by aggregating and propagating the features of entities in the knowledge graph through neighbor entity aggregation, import the entity embedding vectors into the GRU unit to obtain the relationship between entities, and use the Softmax function to obtain the centrality of entities;

[0065] S206, obtaining an importance score of the entity according to the centrality and the initial score, sorting the entities according to the importance scores, selecting exosome features meeting preset requirements, expanding the original feature set after reducing feature redundancy according to the correlation between the exosome features.

[0066] It should be noted that the graph neural network aggregates and propagates the features of the entities in the knowledge graph to generate entity embedding vectors, and learns the feature information of the triples in the knowledge graph through the GRU unit to obtain the centrality of the entity in the knowledge graph, which represents the importance of the entity. According to the characteristics of the knowledge graph, the more nodes connected to a node, the more information the node may imply. When calculating the importance score of the entity, the preference degree of the bronchial asthma phenotype classification to the entity is considered, the corresponding relationship of the entity is obtained according to the knowledge graph, and the relevance of the relationship and the bronchial asthma phenotype classification is calculated, and the initial score of the entity is obtained according to the relevance.

[0067] It should be noted that the expanded original feature set is obtained, the feature importance and feature redundancy are used as evaluation indexes, and the maximum feature importance and minimum feature redundancy are used as optimization objectives; a bronchial asthma phenotype classification model is constructed based on a convolutional autoencoder and a GUR unit, a dataset is constructed based on the expanded original feature set and corresponding examination data, the dataset is divided into a plurality of subsets, each subset is used as a validation set, and the remaining subsets are used as training sample sets to train the bronchial asthma phenotype classification model to obtain a plurality of models; the convolutional autoencoder part is composed of stacked convolutional layers, pooling layers and batch normalization layers, the convolutional autoencoder parameters are initialized, the training is performed in an unsupervised manner, and the recognition classification features of the output of the encoder part are obtained; the time correlation of the features is obtained by importing the recognition classification features into the GRU unit, the bronchial asthma phenotype classification is realized by deep learning of the time sequence features, the average classification accuracy of the plurality of bronchial asthma phenotype classification models is obtained as an importance evaluation index, and is expressed as: Wherein, n represents the number of classification models, F represents the classification model parameters, D represents the training sample set, x i represents a training sample.

[0068] The mRMR algorithm is reconstructed using the Pearson correlation coefficient, the correlation between the selected features and the correlation between the features and the bronchial asthma phenotype are calculated, the redundancy evaluation index of the feature combination is obtained, and is expressed as: Wherein, R(x i ; y) represents the Pearson correlation coefficient between the ith feature x i and the bronchial asthma phenotype classification y, R(x i ; x j ) represents the Pearson correlation coefficient between the ith feature x iPearson correlation coefficient between the jth feature x j S represents the selected feature combination, |S| represents the number of features of the feature combination, and Ω s represents the expanded original feature set.

[0069] Figure 3 A flowchart of the present application for obtaining an exosome feature combination of different bronchial asthma phenotypes is shown.

[0070] According to an embodiment of the present application, the exosome feature combination of different bronchial asthma phenotypes is obtained on the basis of common diagnostic features, specifically:

[0071] S302, obtaining the optimal exosome feature combination according to the optimization target by using the optimized eagle algorithm, initializing the eagle algorithm parameters, encoding the expanded original feature set, and randomly generating an initial feature combination on the feature benchmark as an eagle individual, wherein the common diagnostic features are used as the feature benchmark.

[0072] S304, calculating the fitness of the current position of the initialized eagle population, updating the global optimal fitness value and the optimal position, and presetting the iteration number threshold, when the iteration number does not exceed the preset iteration number threshold, expanding the search for global optimization, otherwise, narrowing the search for local search;

[0073] S306, introducing an adaptive control factor to adjust the flight speed of the eagle individual in global optimization and local search, calculating the fitness value of the new position of the eagle individual, obtaining the feature combination with the highest fitness value, using difference calculation to strengthen the recombination of the feature combination, and recording the global optimal fitness value and the optimal position;

[0074] S308, when the objective function of the feature combination converges, obtaining the optimal feature combination, and removing the common diagnostic features from the optimal feature combination to obtain the exosome feature combination.

[0075] It should be noted that the eagle algorithm simulates the behavior of eagles capturing prey to achieve rapid positioning and solution, and has high search efficiency and fast convergence speed. Encoding the expanded original feature set can better express the feature combination, and grouping the exosome features of different cell sources when randomly generating the initial feature combination, and generating the initial feature combination based on grouping, so that the initial feature combination has local features of exosomes of different cell sources, and the diversity of the initial feature combination is increased.

[0076] By introducing an adaptive control factor to improve the eagle algorithm, in the early iteration, the eagle individual expands the search range with a high flight speed, improves the global search ability, and in the later iteration, the eagle individual locks the prey for capture with a low flight speed, improves the optimization performance in the later period. Its adaptive control factor λ(i) is represented as: λ(i) = λc -i(λ c -λ z ) / T max ,λ c 、λ z distribution represents the range node of the control factor, i represents the iteration number, T max represents the total number of iterations. The fitness is associated with the objective function to establish a mapping relationship, the mapping relationship is used to optimize the iteration convergence of the feature combination, the differential variation idea is introduced in the eagle algorithm, the feature combination with a high fitness value is selected, the difference between other feature combinations and the feature combination is obtained, the feature combination is evolved based on the obtained difference, the information exchange between different feature combinations is realized, and the feature combination is optimized in a predetermined direction.

[0077] It should be noted that the examination data of the exosome feature combination obtained based on the bronchial asthma medical record data is mapped to a low-dimensional space to obtain an exosome feature distribution, and the corresponding identification accuracy is obtained according to the exosome feature distribution of different bronchial asthma phenotypes; based on the probability density function of the exponential distribution, the maximum likelihood estimation is calculated according to the maximum likelihood idea, the identification accuracy corresponding to the common diagnostic feature distribution of different bronchial asthma phenotypes is obtained, and the deviation of the identification accuracy corresponding to the exosome feature distribution and the identification accuracy corresponding to the common diagnostic feature distribution is obtained. If the deviation is greater than a preset deviation threshold, the exosome features of the current bronchial asthma phenotype are reacquired, and if the deviation is not greater than the preset deviation threshold, the exosome feature combination is output.

[0078] Figure 4 A block diagram of the bronchial asthma phenotype identification index screening system based on exosome omics of the application is shown.

[0079] The second embodiment of the application also provides a bronchial asthma phenotype identification index screening system 4 based on exosome omics, which comprises a memory 41 and a processor 42, and the memory comprises a bronchial asthma phenotype identification index screening method program based on exosome omics.

[0080] The third embodiment of the application also provides a computer readable storage medium comprising a bronchial asthma phenotype identification index screening method program based on exosome omics, and when the bronchial asthma phenotype identification index screening method program based on exosome omics is executed by a processor, the steps of the bronchial asthma phenotype identification index screening method based on exosome omics are realized.

[0081] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other manners. The described device embodiments are merely illustrative. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling, or direct coupling or communication connection between the components can be indirect coupling or communication connection through some interfaces, devices, or units, and can be electrical, mechanical, or in other forms.

[0082] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units; they can be located in one place, or distributed on multiple network units; and some or all of the units can be selected as needed to achieve the purposes of the embodiments.

[0083] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; and the integrated unit can be implemented in the form of hardware or hardware plus software functional units.

[0084] Those of ordinary skill in the art can understand that all or part of the steps of the above-described method embodiments can be completed by a program instructing related hardware, and the aforementioned program can be stored in a computer-readable storage medium, and when the program is executed, the steps of the method embodiments are executed; and the aforementioned storage medium includes mobile storage devices, read-only memories (ROMs), random access memories (RAMs), magnetic disks or optical disks, and various media that can store program codes.

[0085] Alternatively, the integrated units of the present application, if implemented in the form of software functional modules and sold or used as independent products, can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present application. The aforementioned storage medium includes: mobile storage devices, ROMs, RAMs, magnetic disks or optical disks, and various media that can store program codes.

Claims

1. An exosome-based bronchial asthma phenotype identification index screening method, characterized in that, The method comprises the following steps: Obtaining a plurality of bronchial asthma medical record data, constructing medical record data subsets based on different diagnostic phenotype categories, extracting multi-dimensional information in different medical record data subsets, and obtaining diagnostic characteristics of different bronchial asthma phenotypes; Obtaining common diagnostic characteristics and common diagnostic characteristic distribution of different bronchial asthma phenotypes, and generating an original feature set based on the common diagnostic characteristics; Retrieving exosome-related knowledge according to a big data method, constructing a knowledge graph through the mechanism of different cell-derived exosomes on bronchial asthma, screening exosome features with importance meeting preset requirements by using the knowledge graph, and expanding the original feature set; Constructing a feature selection task based on multi-objective optimization in the expanded original feature set, obtaining exosome feature combinations of different bronchial asthma phenotypes based on the common diagnostic characteristics, and comparing and verifying the identification accuracy according to the common diagnostic characteristic distribution, and determining the exosome identification index of the bronchial asthma phenotype through the verification result.

2. The exosome-based bronchial asthma phenotype identification index screening method according to claim 1, characterized in that, The medical record data subsets are constructed based on different diagnostic phenotype categories, and multi-dimensional information is extracted in different medical record data subsets, specifically: According to the bronchial asthma disease as the keyword, the medical record data of the bronchial asthma disease meeting the search label constraint is obtained, and the diagnostic information is extracted, and the bronchial asthma phenotype is obtained according to the diagnostic information; The corresponding bronchial asthma medical record data is aggregated by using the bronchial asthma phenotype, and medical record data subsets of different bronchial asthma phenotypes are constructed, detection indexes in different medical record data subsets are extracted, and abnormal indexes are screened; According to the abnormal indexes and the clinical symptoms in the medical record data, the diagnostic basis is determined, the low-dimensional features are obtained by using single-channel convolution, the high-dimensional features are obtained by using convolution calculation, normalization processing and activation operation based on the main branch of the low-dimensional features; At the same time, the low-dimensional features are locally integrated by using average pooling processing based on the side branch, the outputs of the two branches are fused, multi-dimensional information is obtained, and diagnostic characteristics of different bronchial asthma phenotypes are generated according to the multi-dimensional information.

3. The exosome-based bronchial asthma phenotype identification index screening method according to claim 1, characterized in that, The common diagnostic characteristics and the common diagnostic characteristic distribution of different bronchial asthma phenotypes are obtained, and an original feature set is generated based on the common diagnostic characteristics, specifically: Obtaining diagnostic feature sets of different bronchial asthma phenotypes, performing intersection processing on each diagnostic feature set, extracting common diagnostic characteristics to generate an original feature set, and extracting corresponding examination data in each medical record data subset according to the common diagnostic characteristics in the original feature set; The examination data corresponding to the common diagnostic characteristics is mapped to a low-dimensional space, the index distribution of the examination data in the low-dimensional space is obtained to generate a common diagnostic characteristic distribution, and the identification accuracy is obtained according to the common diagnostic characteristic distribution of different bronchial asthma phenotypes.

4. The exosome-based bronchial asthma phenotype identification index screening method according to claim 1, characterized in that, Retrieving exosome-related knowledge according to a big data method, constructing a knowledge graph through the mechanism of different cell-derived exosomes on bronchial asthma, specifically: Exosome proteomics professional corpus and exosome proteomics and bronchial asthma big data information are obtained based on a search engine through a big data method, structured processing is performed, a feature vector of exosome proteomics related knowledge is obtained by using an encoder model, deep feature information is captured to obtain entities; Semantic features are obtained by embedding based on the exosome proteomics and bronchial asthma big data information, the mechanism of different cell-derived exosomes on bronchial asthma is read through the semantic features, the relationship between entities is extracted based on the mechanism, and a relationship table is constructed; The obtained entities and the relationship table are mapped into a triple form to construct a knowledge graph in a graphical manner.

5. The exosome-based bronchial asthma phenotype identification index screening method according to claim 1, characterized in that, The knowledge graph is used to screen exosome features that meet preset requirements in terms of importance, and the original feature set is expanded, specifically as follows: For each entity in the knowledge graph, the number of entities directly connected to the entity is obtained, the entity contribution degree is obtained according to the number of entities, and the knowledge graph is represented in the form of an adjacency matrix according to the entity contribution degree; A graph neural network is used to perform representation learning on the knowledge graph, the features of entities in the knowledge graph are aggregated and propagated through neighbor entities, and an entity embedding vector is constructed; The entity embedding vector is input into a GRU unit to obtain the relationship between entities, and the centrality of the entity is obtained by using a Softmax function; 6. The exosome-based bronchial asthma phenotype identification index screening method according to claim 1, characterized in that, The importance score of the entity is obtained according to the centrality and the initial score, the entities are sorted according to the importance score, the exosome features that meet the preset requirements are selected, and the original feature set is expanded after reducing feature redundancy through correlation calculation between exosome features. A feature selection task based on multi-objective optimization is constructed in the expanded original feature set, specifically as follows: The expanded original feature set is obtained, and the feature importance and feature redundancy are used as evaluation indexes, and the maximum feature importance and the minimum feature redundancy are used as optimization objectives; A bronchial asthma phenotype classification model is constructed based on a convolutional autoencoder and a GUR unit, a dataset is constructed based on the expanded original feature set and corresponding examination data, the dataset is divided into several subsets, each subset is used as a validation set, the remaining subsets are used as training sample sets to train the bronchial asthma phenotype classification model, and several models are obtained; The convolutional autoencoder part is composed of stacked convolutional layers, pooling layers and batch normalization layers, the parameters of the convolutional autoencoder are initialized, and the convolutional autoencoder is trained in an unsupervised training manner to obtain recognition classification features of the output of the encoder part; The recognition classification features are input into a GRU unit to obtain the time correlation of the features, bronchial asthma phenotype classification is realized through deep learning of time sequence features by using a fully connected layer, and the average classification accuracy of the several bronchial asthma phenotype classification models is used as an importance evaluation index; 7. The exosome-based bronchial asthma phenotype identification index screening method according to claim 1, characterized in that, The mRMR algorithm is reconstructed by using a Pearson correlation coefficient, the correlation between the selected features and the correlation between the features and the bronchial asthma phenotype are calculated, and a redundancy evaluation index of the feature combination is obtained. Exosome feature combinations of different bronchial asthma phenotypes are obtained based on common diagnostic features, specifically as follows: The optimized eagle algorithm is used to obtain an optimal exosome feature combination according to an optimization target, parameters of the eagle algorithm are initialized, the expanded original feature set is encoded, and a common diagnostic feature is taken as a feature benchmark, and an initial feature combination is randomly generated on the feature benchmark as an eagle individual; The fitness of the current position of the initialized eagle population is calculated, the global optimal fitness value and the optimal position are updated, a preset iteration number threshold is set, when the iteration number is not more than the preset iteration number threshold, the search is expanded for global optimization, otherwise, the search is reduced for local search; An adaptive control factor is introduced to adjust the flight speed of the eagle individual in global optimization and local search, the fitness value of the new position of the eagle individual is calculated, the feature combination with the highest fitness value is obtained, differential calculation is used to strengthen the recombination of the feature combination, and the global optimal fitness value and the optimal position are recorded; When the target function of the feature combination converges, the optimal feature combination is obtained, and the exosome feature combination is obtained by removing the common diagnostic feature in the optimal feature combination.

8. The exosome-based bronchial asthma phenotype identification index screening method according to claim 1, characterized in that, According to the distribution of the common diagnostic features, the accuracy of identification is compared and verified, and the exosome identification index of the bronchial asthma phenotype is determined according to the verification result, specifically: The exosome feature combination is obtained based on the bronchial asthma medical record data, the inspection data of the exosome feature combination is mapped to a low-dimensional space to obtain an exosome feature distribution, and the corresponding identification accuracy is obtained according to the exosome feature distribution of different bronchial asthma phenotypes; The identification accuracy corresponding to the common diagnostic feature distribution of different bronchial asthma phenotypes is obtained, and the deviation of the identification accuracy corresponding to the exosome feature distribution and the identification accuracy corresponding to the common diagnostic feature distribution is compared and obtained; If the deviation is greater than a preset deviation threshold, the exosome features of the current bronchial asthma phenotype are reacquired, and if the deviation is not greater than the preset deviation threshold, the exosome feature combination is output.

Images