Immune index data processing method and system based on graph neural network

Abstract

The invention relates to an immune index data processing method and system based on a graph neural network, and relates to the technical field of data processing. The method comprises the following steps: receiving an immunodetection target and screening related immune indexes; performing pairwise comparison analysis on the immune indexes to obtain an index relationship, wherein the index relationship comprises relevance and relevance intensity; constructing an immune index relation graph based on the indexes and the relation; acquiring historical data and constructing a processing data set in combination with the relation graph; training the graph neural network model to obtain an analysis model; performing graph embedding representation on the input data; and inputting the graph embedded vector into the model to obtain an analysis result. The technical problems of low efficiency and insufficient accuracy when the immune index data is analyzed and processed in the prior art are solved, and the technical effect of efficiently and accurately analyzing and processing the immune index data is achieved by constructing the immune index relation graph and combining the graph neural network to analyze and process the immune index data.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to an immune indicator data processing method and system based on a graph neural network. Background Art

[0002] With the development of medical testing technology, immune index test data plays an increasingly important role in assisting medical decision-making. The analysis and processing of immune index data affects the accuracy and efficiency of related decisions. At present, the commonly used immune index data analysis and processing methods mainly include statistical analysis methods and traditional machine learning methods. Statistical analysis methods usually use a single indicator or a simple combination of indicators for analysis, which cannot fully consider the complex correlation between immune indicators, resulting in limited accuracy of data analysis results. Although traditional machine learning methods can process multidimensional features, because immune indicators are modeled as independent features, the interaction and influence between indicators are ignored, resulting in low analysis efficiency and difficulty in capturing potential correlation patterns between indicators. Therefore, how to effectively utilize the correlation between immune indicators to achieve efficient and accurate analysis and processing of immune indicator data has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] The present invention aims to solve the technical problems of low efficiency and insufficient accuracy in analyzing and processing immune indicator data in the prior art, and provides an immune indicator data processing method and system based on a graph neural network.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] In a first aspect, the present invention provides an immune indicator data processing method based on a graph neural network, comprising: receiving an immune detection target, and screening out multiple immune indicators from an immune indicator library according to the immune detection target; performing pairwise comparison analysis on the multiple immune indicators to obtain an indicator relationship between any two immune indicators among the multiple immune indicators, wherein the indicator relationship includes correlation and correlation strength, wherein the correlation includes positive correlation, negative correlation, and no direct correlation; constructing an immune indicator relationship graph based on multiple immune indicators and indicator relationships, wherein the nodes in the immune indicator relationship graph correspond to multiple immune indicators, positive correlation corresponds to positive edges, negative correlation corresponds to negative edges, and no direct correlation corresponds to no edges, and the correlation strength is used as the weight of the corresponding edge; obtaining immune indicator historical data, processing the immune indicator historical data in combination with the immune indicator relationship graph, and constructing an immune indicator processing data set; using the immune indicator processing data set to train a graph neural network model to obtain an immune indicator analysis model; receiving immune indicator input data to be analyzed, and performing graph embedding representation on the immune indicator input data according to the immune indicator relationship graph to obtain an immune indicator graph embedding vector; inputting the immune indicator graph embedding vector into the immune indicator analysis model to obtain an immune analysis result for the immune indicator input data.

[0006] In the second aspect, the present invention provides an immune indicator data processing system based on a graph neural network, including: an indicator screening module, which is used to receive an immune detection target and screen out multiple immune indicators from an immune indicator library according to the immune detection target; a relationship analysis module, which is used to perform pairwise comparison analysis on multiple immune indicators to obtain the indicator relationship between any two immune indicators among the multiple immune indicators, wherein the indicator relationship includes correlation and correlation strength, wherein the correlation includes positive correlation, negative correlation and no direct correlation; a graph construction module, which is used to construct an immune indicator relationship graph based on multiple immune indicators and indicator relationships, wherein the nodes in the immune indicator relationship graph correspond to multiple immune indicators, positive correlation corresponds to positive edges, negative correlation corresponds to negative edges, and No direct association corresponds to no edge, and the association strength is used as the weight of the corresponding edge; the data processing module is used to obtain the historical data of immune indicators, process the historical data of immune indicators in combination with the immune indicator relationship diagram, and construct an immune indicator processing data set; the model training module is used to train the graph neural network model using the immune indicator processing data set to obtain an immune indicator analysis model; the graph embedding module is used to receive the immune indicator input data to be analyzed, and perform graph embedding representation on the immune indicator input data according to the immune indicator relationship diagram to obtain an immune indicator graph embedding vector; the result analysis module is used to input the immune indicator graph embedding vector into the immune indicator analysis model to obtain the immune analysis results for the immune indicator input data.

[0007] The beneficial effects of the present invention are:

[0008] First, the immune detection target is received, and multiple relevant immune indicators are screened from the immune indicator library according to the target to ensure the pertinence of subsequent analysis and the relevance of data. Secondly, the multiple immune indicators screened are compared and analyzed in pairs to obtain the indicator relationship between any two immune indicators, including correlation (positive correlation, negative correlation and no direct correlation) and correlation strength, so as to achieve a comprehensive quantification of the relationship between immune indicators. Then, based on the above immune indicators and indicator relationships, an immune indicator relationship graph is constructed. In this relationship graph, nodes correspond to immune indicators, positive correlation corresponds to positive edges, negative correlation corresponds to negative edges, no direct correlation corresponds to no edges, and the correlation strength is used as the weight of the corresponding edge, thereby converting the complex relationship between immune indicators into a computable graph structure representation. Next, the immune indicator historical data is obtained, and these historical data are processed in combination with the constructed immune indicator relationship graph to construct an immune indicator processing data set to prepare structured training data for subsequent model training. Afterwards, the constructed immune indicator processing data set is used to train the graph neural network model to obtain the immune indicator analysis model, which realizes the deep learning of the immune indicator data features. In practical applications, we first receive the input data of the immune indicators to be analyzed, and then embed the input data into a graph according to the immune indicator relationship graph to obtain the immune indicator graph embedding vector, which converts the complex relationship data into a computable vector representation. After that, we input the immune indicator graph embedding vector into the trained immune indicator analysis model to obtain the immune analysis results for the input data, thereby achieving efficient and accurate analysis and processing of the immune indicator data. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of the flow chart of the immune index data processing method based on graph neural network provided by the present invention;

[0010] Figure 2 A schematic diagram of the structure of the immune index data processing system based on graph neural network provided by the present invention.

[0011] In the accompanying drawings, the components represented by the reference numerals are as follows:

[0012] Indicator screening module 11, relationship analysis module 12, graph construction module 13, data processing module 14, model training module 15, graph embedding module 16, result analysis module 17. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0014] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0015] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0016] Embodiment 1:

[0017] like Figure 1 As shown, the embodiment of the present invention provides an immune index data processing method based on a graph neural network, including:

[0018] S100: receiving an immune detection target, and screening out a plurality of immune indicators from an immune indicator library according to the immune detection target.

[0019] Specifically, first, the immune detection target input by the user is received through a human-computer interaction interface, a data interface or an API. Among them, the immune detection target is expressed in the form of text or preset options. The immune index library is a pre-established database that stores a large amount of immune index data, such as antibody levels, cytokine concentrations and other immune indicators. Each immune indicator in the immune index library contains basic information such as the indicator name, indicator type, indicator description, reference range, and the association information between the indicator and different immune states. Then, according to the immune detection target, search and filter in the immune index library to screen out a group of indicators that are significantly correlated with the immune detection target, and obtain multiple immune indicators. These indicators will serve as the basic data for subsequent analysis. Each screened immune indicator has its unique numerical characteristics and biological significance, and can reflect the characteristics of the immune detection target from different angles.

[0020] By establishing the association between immune detection targets and specific immune indicators, the purpose of extracting relevant immune indicators from massive indicators can be achieved. Through targeted indicator screening, the efficiency of subsequent analysis is not only improved, but also the foundation for building an accurate immune indicator relationship diagram is laid.

[0021] S200: performing pairwise comparison analysis on the plurality of immune indicators to obtain an indicator relationship between any two immune indicators among the plurality of immune indicators, wherein the indicator relationship includes correlation and correlation strength, wherein the correlation includes positive correlation, negative correlation and no direct correlation.

[0022] Specifically, after the immune index screening is completed, these screened immune indexes are compared and analyzed in pairs to obtain the indicator relationship between each pair of immune indexes. First, the historical test data of the screened multiple immune indexes are obtained, including the test value records in different samples. Then, the historical test data of any two immune indexes are correlated, the correlation coefficient is calculated, and the absolute value of the correlation coefficient is used as the correlation strength between the two indexes.

[0023] According to the calculated correlation coefficient, the correlation is judged. When the correlation coefficient is greater than the first preset threshold, it is determined that the two immune indicators are positively correlated; when the correlation coefficient is less than the second preset threshold, it is determined to be a negative correlation; when the correlation coefficient is between the second preset threshold and the first preset threshold, it is determined that there is no direct correlation between the two indicators. Among them, the first preset threshold and the second preset threshold are reference values ​​pre-set according to expert experience for judging the correlation between immune indicators, the first preset threshold is a positive value, the second preset threshold is a negative value, and the second preset threshold is less than the first preset threshold. In this way, an association network between immune indicators is established to prepare for the subsequent construction of an immune indicator relationship diagram.

[0024] Through pairwise comparison analysis, the relationships between different immune indicators can be fully revealed, including both the direction and degree of association, providing support for understanding and analyzing the relationship between multiple immune indicators.

[0025] S300: Based on the multiple immune indicators and the indicator relationships, construct an immune indicator relationship graph, where the nodes in the immune indicator relationship graph correspond to the multiple immune indicators, the positive correlation corresponds to a positive edge, the negative correlation corresponds to a negative edge, and the no direct correlation corresponds to no edge, and the correlation strength is used as the weight of the corresponding edge.

[0026] Specifically, after obtaining the correlation between multiple immune indicators, we start to build the immune indicator relationship graph. First, each immune indicator is taken as a node in the graph. Then, the connection relationship between the nodes is established based on the obtained indicator relationship: for positively correlated indicator pairs, positive edges are established between the corresponding nodes; for negatively correlated indicator pairs, negative edges are established between the corresponding nodes; for indicator pairs without direct correlation, no edge connection is established between the corresponding nodes. At the same time, the correlation strength between each pair of related indicators is used as the weight value of the corresponding edge.

[0027] Among them, the immune indicator relationship graph is a data structure used to represent the relationship between multiple immune indicators, which includes a node set and an edge set. A positive edge indicates that two indicators are positively correlated, and a negative edge indicates that two indicators are negatively correlated. The weight of the edge reflects the strength of this correlation. The complex correlation network between immune indicators can be intuitively displayed through the representation of the graph structure.

[0028] Through the representation of graph structure, the relationship between immune indicators is converted into a processable data structure, which provides a basis for the subsequent use of graph neural networks for data processing, and also facilitates the analysis and understanding of the mutual influence relationship between multiple immune indicators.

[0029] S400: Acquire historical data of immune indicators, process the historical data of immune indicators in combination with the immune indicator relationship diagram, and construct an immune indicator processing data set.

[0030] Specifically, after obtaining the immune indicator relationship graph, the immune indicator historical data is obtained, wherein the immune indicator historical data includes multiple groups of historical test samples, and each group of historical test samples includes sample immune indicator test data and corresponding sample immune analysis results. Then, according to the constructed immune indicator relationship graph, the sample immune indicator test data in each group of historical test samples is represented by graph embedding to obtain the sample immune indicator graph embedding vector. Among them, the graph embedding representation is a data processing method that converts graph structure data into vector form. Afterwards, the sample immune indicator graph embedding vector is used as the input feature, and the corresponding sample immune analysis result is used as the output label to construct the immune indicator processing data set.

[0031] By converting the original historical data of immune indicators into a data format suitable for machine learning, a high-quality training data set is provided for subsequent model training.

[0032] S500: Using the immune index processing data set to train a graph neural network model to obtain an immune index analysis model.

[0033] Specifically, after obtaining the immune indicator processing data set, the graph neural network model is trained based on the constructed immune indicator processing data set. The graph neural network model is a deep learning model specifically used to process graph structure data, which uses both node feature information and relationship information between nodes for learning.

[0034] During the training process, the sample immune indicator graph embedding vectors in the immune indicator processing data set are input into the graph neural network model. The model learns the inherent laws of the immune indicator data by extracting the immune indicator node features and the correlation features between indicators. According to the difference between the model output results and the actual sample immune analysis results, the model parameters are continuously adjusted to optimize the model performance, and finally an immune indicator analysis model with good analysis capabilities is obtained.

[0035] Through the model training process, the complex interactive relationships between immune indicators are learned, thus providing reliable model support for subsequent immune indicator analysis tasks.

[0036] S600: receiving immune indicator input data to be analyzed, and performing graph embedding representation on the immune indicator input data according to the immune indicator relationship graph to obtain an immune indicator graph embedding vector.

[0037] Specifically, first, the immune index input data to be analyzed is received. These input data include the detection values ​​of various immune indexes of the sample to be analyzed. Then, the constructed immune index relationship graph is used to perform graph embedding representation processing on the received immune index input data.

[0038] When performing graph embedding representation, the immune indicator input data is matched with the nodes in the immune indicator relationship graph, and the input data is converted into an immune indicator graph embedding vector containing relationship features based on the edge relationship and weight information in the graph. This graph embedding representation method can retain the association information between immune indicators, so that the subsequent analysis process can make full use of the relationship between indicators.

[0039] By converting the input data of the immune indicators to be analyzed into a data format suitable for model processing, preparation is made for subsequent analysis and processing.

[0040] S700: Input the immune index graph embedding vector into the immune index analysis model to obtain the immune analysis result for the immune index input data.

[0041] Specifically, the obtained immune indicator graph embedding vector is input into the trained immune indicator analysis model. The model analyzes and processes the input data based on the input graph embedding vector, combined with the immune indicator features and relationship patterns learned during the training process, and outputs the immune analysis results for the group of immune indicator input data. By using the trained model to analyze new immune indicator data, efficient and accurate analysis and processing of immune indicator data is achieved.

[0042] Furthermore, the embodiment of the present application also includes:

[0043] S110: Obtaining the target type of the immune detection target;

[0044] S120: searching, according to the target type, in the immune indicator library for a candidate immune indicator set corresponding to the target type;

[0045] S130: Calculating the correlation of the immune indicators in the candidate immune indicator set to obtain the correlation between each immune indicator and the immune detection target;

[0046] S140: Screening the candidate immune indicator set according to a correlation threshold, and taking immune indicators with correlation greater than the correlation threshold as the multiple immune indicators.

[0047] In a feasible implementation, the target type of the immune detection target is first obtained. The target type is used to characterize the detection category to which the immune detection target belongs, including but not limited to different categories such as immune function assessment and immune system status judgment. Then, according to the obtained target type, a search is performed in the immune index library to obtain a set of candidate immune indexes corresponding to the target type. Through the correspondence between the target type and the immune index, the immune indexes that may be related to the detection target are preliminarily screened out.

[0048] Next, the correlation is calculated for each immune indicator in the candidate immune indicator set to obtain the correlation between each immune indicator and the immune detection target. The correlation calculation is based on the statistical relationship between the immune indicator and the detection target in the historical data, reflecting the degree of association between the indicator and the target. Afterwards, the candidate immune indicator set is further screened according to the preset correlation threshold. The immune indicators with a correlation greater than the correlation threshold are retained as the final multiple immune indicators screened out. The correlation threshold is used to control the strictness of the screening, which is determined by the immune expert group to ensure that the screened indicators have a sufficiently strong correlation with the detection target.

[0049] By screening layer by layer, the indicator set most relevant to the immune detection target is selected from the immune indicator library to provide a reliable data basis for subsequent analysis and processing.

[0050] Furthermore, the correlation of the immune indicators in the candidate immune indicator set is calculated to obtain the correlation between each immune indicator and the immune detection target, including:

[0051] Construct a correlation calculation function:

[0052] R(i)=w 1 S(i)+w 2 T(i)+w 3 H(i);

[0053]

[0054]

[0055] H(i)=log(1+D i )·M i ;

[0056] Among them, w 1 、w 2 、w 3 is the weight coefficient and satisfies w 1 +w 2 +w 3 =1; S(i) represents semantic relevance, V i is the semantic vector representation of candidate immune index i, V t is the semantic vector representation of the immune detection target, σ is the smoothing factor; T(i) represents the clinical relevance, P i,j represents the frequency of occurrence of candidate immune indicator i in disease type j, C j represents the correlation coefficient between disease type j and immune detection target, N is the total number of disease types; H(i) represents the historical correlation, D i is the number of times candidate immune indicator i and immune detection target co-occur in historical data, M i is the degree of matching in the historical detection;

[0057] Traverse the candidate immune indicator set and obtain the first candidate immune indicator;

[0058] For the first candidate immune indicator, calculating the first candidate correlation between the first candidate immune indicator and the immune detection target according to the correlation calculation function;

[0059] The first candidate relevance is associated with the first candidate immunity indicator and stored in the candidate immunity indicator set.

[0060] In a preferred embodiment, first, a correlation calculation function is constructed. The correlation calculation function is used to comprehensively evaluate the degree of association between the candidate immune index and the immune detection target, specifically expressed as: R(i)=w1 S(i)+w 2 T(i)+w 3 H(i). Among them, w 1 、w 2 、w 3 is the weight coefficient and satisfies w 1 +w 2 +w 3 =1, used to adjust the importance of correlation in different dimensions.

[0061] S(i) represents semantic relevance, and the specific calculation formula is: Among them, V i is the semantic vector representation of candidate immune index i, V t is the semantic vector representation of the immune detection target, and σ is the smoothing factor. The normalized semantic relevance value is obtained by calculating the distance between the vector representations and mapping them through the exponential function.

[0062] T(i) represents clinical relevance, and the specific calculation formula is: Among them, P i,j represents the frequency of occurrence of candidate immune indicator i in disease type j, C j It represents the correlation coefficient between disease type j and immune detection target, and N is the total number of disease types. By comprehensively considering the frequency of occurrence of immune indicators in various diseases and their correlation with the target, the correlation evaluation at the clinical application level is obtained.

[0063] H(i) represents the historical correlation, and the specific calculation formula is H(i) = log(1+D i )·M i , where D i is the number of times candidate immune indicator i and immune detection target co-occur in historical data, M i is the matching degree in historical detection. By introducing a logarithmic function to process the number of co-occurrences and combining it with the matching degree, a quantitative evaluation of the correlation of indicators in historical data can be achieved.

[0064] Subsequently, by traversing the candidate immune indicator set, any one of the candidate immune indicators is picked up each time as the first candidate immune indicator. Then, for the first candidate immune indicator obtained, the first candidate correlation with the immune detection target is calculated according to the constructed correlation calculation function. After that, the calculated first candidate correlation is associated with the corresponding first candidate immune indicator and stored in the candidate immune indicator set to provide a basis for subsequent indicator screening.

[0065] Through the correlation calculation function, a multi-dimensional quantitative evaluation of the degree of association between candidate immune indicators and detection targets is achieved, providing reliable data support for subsequent indicator screening. The correlation calculation function integrates information from three dimensions: semantic relevance, clinical relevance, and historical relevance. It not only takes into account the characteristics of the indicator itself, but also combines clinical experience in practical applications, while using the statistical information provided by historical data to ensure the comprehensiveness and reliability of the correlation evaluation results. Furthermore, the embodiments of the present application also include:

[0066] S210: Acquire historical detection data of the plurality of immune indicators;

[0067] S220: performing correlation analysis on historical detection data of any two immune indicators to obtain a correlation coefficient, and the absolute value of the correlation coefficient is used as the correlation strength;

[0068] S230: When the correlation coefficient is greater than a first preset threshold, determining that the two immune indicators are positively correlated;

[0069] S240: When the correlation coefficient is less than a second preset threshold, determining that the two immune indicators are negatively correlated;

[0070] S250: When the correlation coefficient is between the second preset threshold and the first preset threshold, it is determined that there is no direct correlation between the two immune indicators.

[0071] In a feasible implementation, first, the historical detection data of multiple immune indicators are obtained. These historical detection data contain the detection value records of each immune indicator in different samples, providing a data basis for subsequent association analysis. Then, the historical detection data of any two immune indicators are subjected to correlation analysis, the correlation coefficient is calculated, and the absolute value of the correlation coefficient is used as the strength of association between the two indicators. Among them, the correlation coefficient reflects the degree of consistency of the change trend between the two immune indicators, and the greater the absolute value, the stronger the degree of association between the two indicators. Then, the correlation is judged based on the calculated correlation coefficient. When the correlation coefficient is greater than the first preset threshold, it is determined that the two immune indicators are positively correlated; when the correlation coefficient is less than the second preset threshold, it is determined to be a negative correlation; when the correlation coefficient is between the second preset threshold and the first preset threshold, it is determined that there is no direct association between the two indicators. Among them, the first preset threshold and the second preset threshold are judgment reference values ​​pre-set according to expert experience, the first preset threshold is a positive value, the second preset threshold is a negative value, and the second preset threshold is less than the first preset threshold.

[0072] Through pairwise comparison analysis, we can comprehensively obtain the correlation between immune indicators, laying the foundation for constructing an immune indicator relationship diagram.

[0073] Furthermore, the embodiment of the present application also includes:

[0074] S410: Acquire historical data of immune indicators, wherein the historical data of immune indicators includes multiple groups of historical test samples, and each group of historical test samples includes sample immune indicator test data and corresponding sample immune analysis results;

[0075] S420: According to the immune index relationship graph, the sample immune index detection data in each group of historical detection samples are represented by graph embedding to obtain a sample immune index graph embedding vector;

[0076] S430: Using the sample immune index graph embedding vector as input feature and the corresponding sample immune analysis result as output label to construct the immune index processing data set.

[0077] In a feasible implementation, first, the historical data of immune indicators is obtained. The historical data contains multiple groups of historical test samples, each of which consists of two parts: one part is the sample immune indicator test data, which records the specific test values ​​of each immune indicator; the other part is the sample immune analysis results, which contains the medical experts' diagnostic conclusions or analysis and evaluation of the group of test data. These historical data come from real test records in clinical practice. Then, using the immune indicator relationship graph constructed above, the sample immune indicator test data in each group of historical test samples is processed by graph embedding representation to obtain the sample immune indicator graph embedding vector. In the graph embedding representation process, the test data is corresponded to the nodes in the relationship graph, and the positive and negative correlations and correlation strengths between the nodes are considered, so that this information is encoded into a vector of fixed dimension. This representation method not only retains the original test value information, but also incorporates the relationship information between indicators. Afterwards, the processed sample immune indicator graph embedding vector is used as the input feature of the model, and the corresponding sample immune analysis result is used as the output label to construct a standard machine learning data set format to obtain an immune indicator processing data set. This dataset structure enables the model to learn the characteristics of a single indicator and grasp the correlation patterns between indicators during training, thereby improving the accuracy of the analysis.

[0078] Through the above processing, the original immune indicator historical data is converted into a structured training data set, providing high-quality data support for subsequent model training.

[0079] Furthermore, the embodiment of the present application also includes:

[0080] S441: Counting the sample sizes of the multiple groups of historical detection samples to obtain historical sample sizes;

[0081] S442: When the historical sample size is less than a preset sample size threshold, data perturbation processing is performed on the sample immune index detection data in the multiple groups of historical detection samples to obtain multiple groups of sample immune index perturbation data;

[0082] S443: Submitting the plurality of groups of sample immune index disturbance data to an expert system, and generating a plurality of corresponding disturbance sample immune analysis results according to the plurality of groups of sample immune index disturbance data;

[0083] S444: mapping and associating multiple groups of sample immune index disturbance data with multiple disturbance sample immune analysis results to form multiple groups of disturbance detection samples;

[0084] S445: Combining multiple groups of historical detection samples and multiple groups of disturbance detection samples, constructing the immune index processing data set.

[0085] In a preferred embodiment, first, the sample size of multiple groups of historical test samples is counted to obtain the historical sample size. When it is found that the historical sample size is less than a preset sample size threshold, data expansion processing is required. The expansion processing first performs data perturbation processing on the sample immune index detection data in multiple groups of historical test samples to obtain multiple groups of sample immune index perturbation data. Among them, the data perturbation processing can use methods such as adding Gaussian noise, random scaling or translation to generate new sample data while maintaining the basic characteristics of the data. Then, the generated multiple groups of sample immune index perturbation data are submitted to the expert system. Based on medical knowledge and rules, the expert system analyzes these perturbation data and generates corresponding perturbation sample immune analysis results. The introduction of the expert system ensures that the analysis results of the perturbation data are professionally reliable.

[0086] Next, multiple groups of sample immune indicator perturbation data are mapped one by one with the immune analysis results of the perturbed samples generated by the expert system to form multiple groups of perturbation detection samples. Each group of perturbation detection samples contains the perturbed indicator data and the corresponding analysis results. After that, the original multiple groups of historical detection samples and the newly generated multiple groups of perturbation detection samples are combined to construct a complete immune indicator processing data set. This data expansion scheme not only increases the sample size, but also ensures the reliability of the data, providing more sufficient data support for model training.

[0087] Through data expansion processing, the problem of insufficient sample size in practical applications is effectively solved, and the effect of subsequent model training is improved.

[0088] Furthermore, the embodiment of the present application also includes:

[0089] S510: Constructing a graph neural network structure, wherein the graph neural network structure includes a node feature extraction channel and an association feature extraction channel;

[0090] S520: extracting feature information of each immune indicator node in the sample immune indicator graph embedding vector based on the node feature extraction channel to obtain sample node features;

[0091] S530: extracting feature information of nodes directly associated with each immune indicator node in the sample immune indicator graph embedding vector based on the associated feature extraction channel to obtain sample associated features;

[0092] S540: Fusing the sample node feature and the sample association feature to obtain a fused feature;

[0093] S550: Based on the immune index processing data set, iteratively train the immune index analysis model according to the fusion features and the corresponding sample immune analysis results.

[0094] In a feasible implementation, a graph neural network structure is first constructed, which includes two feature extraction channels: a node feature extraction channel and an associated feature extraction channel. Among them, the node feature extraction channel is used to extract the feature information of a single immune indicator node, and the associated feature extraction channel is used to extract the associated feature information between immune indicators. Then, through the node feature extraction channel, feature extraction is performed on each immune indicator node in the sample immune indicator graph embedding vector to obtain sample node features. These node features reflect the independent characteristics of each immune indicator. At the same time, through the associated feature extraction channel, the feature information of the nodes directly associated with each immune indicator node is extracted to obtain sample associated features. These associated features reflect the interaction relationship between different immune indicators.

[0095] Next, the extracted sample node features and sample association features are fused to obtain fused features. The feature fusion process can organically combine the information of the two features by means of feature splicing, weighted summation, etc. After that, based on the constructed immune index processing data set, the model is trained using the fused features and the corresponding sample immune analysis results. During the training process, the model parameters are continuously iterated and optimized so that the model can accurately capture the characteristic patterns of the immune index data, and finally an immune index analysis model with good analysis capabilities is obtained.

[0096] Through the dual-channel feature extraction structure, the model can simultaneously learn the independent features and associated features of immune indicators, improving the accuracy and reliability of the analysis.

[0097] Furthermore, the embodiment of the present application also includes:

[0098] S561: The graph neural network structure further includes a second-order correlation feature extraction channel, based on which feature information of associated nodes corresponding to directly associated nodes of each immune indicator node in the sample immune indicator graph embedding vector is extracted to obtain the sample second-order correlation feature;

[0099] S562: Fusing the sample node feature, the sample association feature and the sample second-order association feature to obtain an enhanced fusion feature;

[0100] S563: Based on the immune index processing data set, iteratively train the immune index analysis model according to the enhanced fusion features and the corresponding sample immune analysis results.

[0101] In a preferred embodiment, in order to further enhance the analytical capability of the model, a second-order correlation feature extraction channel is added to the graph neural network structure. Based on the second-order correlation feature extraction channel, the second-order correlation features of each immune indicator node in the sample immune indicator graph embedding vector are extracted. Among them, the second-order correlation refers to a node that is indirectly associated with the target node through an intermediate node, that is, the associated node corresponding to the directly associated node of the target node. By extracting the feature information of the second-order associated nodes, the sample second-order correlation features are obtained, thereby capturing the indirect influence relationship of more remote indicators.

[0102] Then, the sample node features, sample association features and newly acquired sample second-order association features are fused to obtain enhanced fusion features. This enhanced fusion feature not only includes the independent features and direct association features of the immune indicators, but also includes the deep features generated by indirect associations, making the feature expression richer and more complete. After that, based on the immune indicator processing data set, the model is trained using the enhanced fusion features and the corresponding sample immune analysis results. Through the iterative optimization process, the model can learn more complex association patterns between immune indicators, thereby improving the analysis performance of the model.

[0103] By introducing the second-order correlation feature extraction channel, the model can capture deeper correlations between immune indicators, further improving the accuracy of the analysis results.

[0104] The immune index data processing method based on graph neural network provided by the embodiment of the present invention has at least the following technical effects:

[0105] Receive the immune detection target, and select multiple immune indicators from the immune indicator library according to the immune detection target to ensure that the data used for subsequent analysis is related to the detection target, so as to avoid introducing irrelevant indicators and causing analysis deviation. Perform a pairwise comparison analysis on multiple immune indicators to obtain the indicator relationship between any two immune indicators among the multiple immune indicators. The indicator relationship includes correlation and correlation strength, where correlation includes positive correlation, negative correlation and no direct correlation, providing basic data for building a relationship graph. Based on multiple immune indicators and indicator relationships, an immune indicator relationship graph is constructed. The nodes in the immune indicator relationship graph correspond to multiple immune indicators, positive correlation corresponds to positive edges, negative correlation corresponds to negative edges, and no direct correlation corresponds to no edges. The correlation strength is used as the weight of the corresponding edge, and the immune indicators and their relationships are converted into a graph structure representation, where the nodes represent indicators, the edges represent the relationship types, and the weights represent the correlation strength, so that complex indicator relationships can be efficiently processed by computers. Obtain the historical data of immune indicators, process the historical data of immune indicators in combination with the immune indicator relationship graph, and construct an immune indicator processing data set to provide structured data support for model training. Use the immune indicator processing data set to train the graph neural network model to obtain the immune indicator analysis model, so that the model can learn the complex relationship patterns between immune indicators. Receive the immune indicator input data to be analyzed, perform graph embedding representation on the immune indicator input data according to the immune indicator relationship graph, obtain the immune indicator graph embedding vector, convert the new input data into a graph embedding vector in the same form as the training data, so that it can be processed by the trained model. Input the immune indicator graph embedding vector into the immune indicator analysis model to obtain the immune analysis result for the immune indicator input data, use the trained model to analyze the graph embedding vector, output the final analysis result, and realize efficient and accurate processing of the immune indicator data.

[0106] Embodiment 2:

[0107] like Figure 2 As shown, based on the same inventive concept as the immune index data processing method based on the graph neural network provided in Example 1, the embodiment of the present invention also provides an immune index data processing system based on the graph neural network, including:

[0108] The index screening module 11 is used to receive an immune detection target and screen a plurality of immune indicators from an immune indicator library according to the immune detection target;

[0109] The relationship analysis module 12 is used to perform a pairwise comparison analysis on the multiple immune indicators to obtain an indicator relationship between any two immune indicators among the multiple immune indicators, wherein the indicator relationship includes correlation and correlation strength, wherein the correlation includes positive correlation, negative correlation and no direct correlation;

[0110] A graph construction module 13 is used to construct an immune indicator relationship graph based on the multiple immune indicators and the indicator relationships, wherein the nodes in the immune indicator relationship graph correspond to the multiple immune indicators, the positive correlation corresponds to a positive edge, the negative correlation corresponds to a negative edge, and the indirect correlation corresponds to no edge, and the correlation strength is used as the weight of the corresponding edge;

[0111] The data processing module 14 is used to obtain historical data of immune indicators, process the historical data of immune indicators in combination with the immune indicator relationship diagram, and construct an immune indicator processing data set;

[0112] A model training module 15 is used to train a graph neural network model using the immune index processing data set to obtain an immune index analysis model;

[0113] A graph embedding module 16, configured to receive immune index input data to be analyzed, and to perform graph embedding representation on the immune index input data according to the immune index relationship graph to obtain an immune index graph embedding vector;

[0114] The result analysis module 17 is used to input the immune index graph embedding vector into the immune index analysis model to obtain the immune analysis result for the immune index input data.

[0115] Furthermore, the indicator screening module 11 includes the following execution steps:

[0116] Obtaining a target type of the immune detection target;

[0117] Retrieving a candidate immune indicator set corresponding to the target type in the immune indicator library according to the target type;

[0118] Calculating the correlation of the immune indicators in the candidate immune indicator set to obtain the correlation between each immune indicator and the immune detection target;

[0119] The candidate immune indicator set is screened according to a correlation threshold, and immune indicators with correlations greater than the correlation threshold are used as the multiple immune indicators.

[0120] Furthermore, the indicator screening module 11 also includes the following execution steps:

[0121] Construct a correlation calculation function:

[0122] R(i)=w 1 S(i)+w 2 T(i)+w 3 H(i);

[0123]

[0124] H(i)=log(1+Di )·M i ;

[0125] Among them, w 1 、w 2 、w 3 is the weight coefficient and satisfies w 1 +w 2 +w 3 =1; S(i) represents semantic relevance, V i is the semantic vector representation of candidate immune index i, V t is the semantic vector representation of the immune detection target, σ is the smoothing factor; T(i) represents the clinical relevance, P i,j represents the frequency of occurrence of candidate immune indicator i in disease type j, C j represents the correlation coefficient between disease type j and immune detection target, N is the total number of disease types; H(i) represents the historical correlation, D i is the number of times candidate immune indicator i and immune detection target co-occur in historical data, M i is the degree of matching in the historical detection;

[0126] Traverse the candidate immune indicator set and obtain the first candidate immune indicator;

[0127] For the first candidate immune indicator, calculating the first candidate correlation between the first candidate immune indicator and the immune detection target according to the correlation calculation function;

[0128] The first candidate relevance is associated with the first candidate immunity indicator and stored in the candidate immunity indicator set.

[0129] Furthermore, the relationship analysis module 12 includes the following execution steps:

[0130] Obtaining historical test data of the multiple immune indicators;

[0131] Performing correlation analysis on the historical test data of any two immune indicators to obtain a correlation coefficient, and the absolute value of the correlation coefficient is used as the correlation strength;

[0132] When the correlation coefficient is greater than a first preset threshold, it is determined that the two immune indicators are positively correlated;

[0133] When the correlation coefficient is less than a second preset threshold, it is determined that the two immune indicators are negatively correlated;

[0134] When the correlation coefficient is between the second preset threshold and the first preset threshold, it is determined that there is no direct correlation between the two immune indicators.

[0135] Furthermore, the graph construction module 13 includes the following execution steps:

[0136] Acquire immune index historical data, wherein the immune index historical data includes multiple groups of historical test samples, each group of historical test samples includes sample immune index test data and corresponding sample immune analysis results;

[0137] According to the immune index relationship graph, the sample immune index detection data in each group of historical detection samples are represented by graph embedding to obtain a sample immune index graph embedding vector;

[0138] The sample immune index graph embedding vector is used as an input feature, and the corresponding sample immune analysis result is used as an output label to construct the immune index processing data set.

[0139] Furthermore, the graph construction module 13 also includes the following execution steps:

[0140] Counting the sample sizes of the multiple groups of historical detection samples to obtain historical sample sizes;

[0141] When the historical sample size is less than a preset sample size threshold, performing data perturbation processing on the sample immune index detection data in multiple groups of historical detection samples to obtain multiple groups of sample immune index perturbation data;

[0142] Submitting the plurality of groups of sample immune index perturbation data to an expert system, and generating a plurality of corresponding perturbed sample immune analysis results according to the plurality of groups of sample immune index perturbation data;

[0143] Mapping and associating multiple groups of sample immune index perturbation data with multiple perturbation sample immune analysis results to form multiple groups of perturbation detection samples;

[0144] The immune index processing data set is constructed by combining multiple groups of historical detection samples and multiple groups of disturbance detection samples.

[0145] Furthermore, the model training module 15 includes the following execution steps:

[0146] Constructing a graph neural network structure, wherein the graph neural network structure includes a node feature extraction channel and an association feature extraction channel;

[0147] Extracting feature information of each immune indicator node in the sample immune indicator graph embedding vector based on the node feature extraction channel to obtain sample node features;

[0148] Extracting feature information of nodes directly associated with each immune indicator node in the sample immune indicator graph embedding vector based on the associated feature extraction channel to obtain sample associated features;

[0149] Fusion of the sample node feature and the sample association feature to obtain a fusion feature;

[0150] Based on the immune index processing data set, the immune index analysis model is iteratively trained according to the fusion features and the corresponding sample immune analysis results.

[0151] Furthermore, the model training module 15 also includes the following execution steps:

[0152] The graph neural network structure also includes a second-order correlation feature extraction channel;

[0153] Extracting feature information of associated nodes corresponding to directly associated nodes of each immune indicator node in the sample immune indicator graph embedding vector based on the second-order association feature extraction channel to obtain the sample second-order association feature;

[0154] Fusion of the sample node feature, the sample association feature and the sample second-order association feature to obtain an enhanced fusion feature;

[0155] Based on the immune index processing data set, the immune index analysis model is iteratively trained according to the enhanced fusion features and the corresponding sample immune analysis results.

[0156] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0157] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0158] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0159] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0160] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0161] Although preferred embodiments of the present invention have been described, additional changes and modifications may occur to these embodiments once those skilled in the art understand the basic inventive concepts.

[0162] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technology, the present invention is also intended to include these changes and variations.

Claims

1. An immune index data processing method based on graph neural network, characterized in that: The method comprises: Receiving an immune detection target, and screening a plurality of immune indicators from an immune indicator library according to the immune detection target; Performing pairwise comparison analysis on the multiple immune indicators to obtain an indicator relationship between any two immune indicators among the multiple immune indicators, wherein the indicator relationship includes correlation and correlation strength, wherein the correlation includes positive correlation, negative correlation and no direct correlation; Based on the multiple immune indicators and the indicator relationships, an immune indicator relationship graph is constructed, wherein the nodes in the immune indicator relationship graph correspond to the multiple immune indicators, the positive correlation corresponds to a positive edge, the negative correlation corresponds to a negative edge, and the indirect correlation corresponds to no edge, and the correlation strength is used as the weight of the corresponding edge; Acquire historical data of immune indicators, process the historical data of immune indicators in combination with the immune indicator relationship diagram, and construct an immune indicator processing data set; Using the immune index processing data set to train a graph neural network model to obtain an immune index analysis model; Receiving immune indicator input data to be analyzed, and performing graph embedding representation on the immune indicator input data according to the immune indicator relationship graph to obtain an immune indicator graph embedding vector; The immune index graph embedding vector is input into the immune index analysis model to obtain the immune analysis result for the immune index input data.

2. The immune index data processing method based on graph neural network according to claim 1 is characterized in that: According to the immune detection target, multiple immune indicators are screened from the immune indicator library, including: Obtaining a target type of the immune detection target; Retrieving a candidate immune indicator set corresponding to the target type in the immune indicator library according to the target type; Calculating the correlation of the immune indicators in the candidate immune indicator set to obtain the correlation between each immune indicator and the immune detection target; The candidate immune indicator set is screened according to a correlation threshold, and immune indicators with correlations greater than the correlation threshold are used as the multiple immune indicators.

3. The immune index data processing method based on graph neural network according to claim 2 is characterized in that: The correlation of the immune indicators in the candidate immune indicator set is calculated to obtain the correlation between each immune indicator and the immune detection target, including: Construct a correlation calculation function: R(i)=w1S(i)+w2T(i)+w3H(i); H(i)=log(1+D i )·M i ; Among them, w1, w2, w3 are weight coefficients, and they satisfy w1+w2+w3=1; S(i) represents the semantic relevance, V i is the semantic vector representation of candidate immune index i, V t is the semantic vector representation of the immune detection target, σ is the smoothing factor; T(i) represents the clinical relevance, P i,j represents the frequency of occurrence of candidate immune indicator i in disease type j, C j represents the correlation coefficient between disease type j and immune detection target, N is the total number of disease types; H(i) represents the historical correlation, D i is the number of times candidate immune indicator i and immune detection target co-occur in historical data, M i is the degree of matching in the historical detection; Traverse the candidate immune indicator set and obtain the first candidate immune indicator; For the first candidate immune indicator, calculating the first candidate correlation between the first candidate immune indicator and the immune detection target according to the correlation calculation function; The first candidate relevance is associated with the first candidate immunity indicator and stored in the candidate immunity indicator set.

4. The immune index data processing method based on graph neural network according to claim 1 is characterized in that: Performing pairwise comparison analysis on the multiple immune indicators to obtain an indicator relationship between any two immune indicators among the multiple immune indicators includes: Obtaining historical test data of the multiple immune indicators; Performing correlation analysis on the historical test data of any two immune indicators to obtain a correlation coefficient, and the absolute value of the correlation coefficient is used as the correlation strength; When the correlation coefficient is greater than a first preset threshold, it is determined that the two immune indicators are positively correlated; When the correlation coefficient is less than a second preset threshold, it is determined that the two immune indicators are negatively correlated; When the correlation coefficient is between the second preset threshold and the first preset threshold, it is determined that there is no direct correlation between the two immune indicators.

5. The immune index data processing method based on graph neural network according to claim 1 is characterized in that: Acquiring immune index historical data, processing the immune index historical data in combination with the immune index relationship diagram, and constructing an immune index processing data set, including: Acquire immune index historical data, wherein the immune index historical data includes multiple groups of historical test samples, each group of historical test samples includes sample immune index test data and corresponding sample immune analysis results; According to the immune index relationship graph, the sample immune index detection data in each group of historical detection samples are represented by graph embedding to obtain a sample immune index graph embedding vector; The sample immune index graph embedding vector is used as an input feature, and the corresponding sample immune analysis result is used as an output label to construct the immune index processing data set.

6. The immune index data processing method based on graph neural network according to claim 5 is characterized in that: The method further comprises: Counting the sample sizes of the multiple groups of historical detection samples to obtain historical sample sizes; When the historical sample size is less than a preset sample size threshold, performing data perturbation processing on the sample immune index detection data in multiple groups of historical detection samples to obtain multiple groups of sample immune index perturbation data; Submitting the plurality of groups of sample immune index perturbation data to an expert system, and generating a plurality of corresponding perturbed sample immune analysis results according to the plurality of groups of sample immune index perturbation data; Mapping and associating multiple groups of sample immune index perturbation data with multiple perturbation sample immune analysis results to form multiple groups of perturbation detection samples; The immune index processing data set is constructed by combining multiple groups of historical detection samples and multiple groups of disturbance detection samples.

7. The immune index data processing method based on graph neural network according to claim 5 is characterized in that: The immune index processing data set is used to train a graph neural network model to obtain an immune index analysis model, including: Constructing a graph neural network structure, wherein the graph neural network structure includes a node feature extraction channel and an association feature extraction channel; Extracting feature information of each immune indicator node in the sample immune indicator graph embedding vector based on the node feature extraction channel to obtain sample node features; Extracting feature information of nodes directly associated with each immune indicator node in the sample immune indicator graph embedding vector based on the associated feature extraction channel to obtain sample associated features; Fusion of the sample node feature and the sample association feature to obtain a fusion feature; Based on the immune index processing data set, the immune index analysis model is iteratively trained according to the fusion features and the corresponding sample immune analysis results.

8. The immune index data processing method based on graph neural network according to claim 7 is characterized in that: The graph neural network structure also includes a second-order correlation feature extraction channel; Extracting feature information of associated nodes corresponding to directly associated nodes of each immune indicator node in the sample immune indicator graph embedding vector based on the second-order association feature extraction channel to obtain the sample second-order association feature; Fusion of the sample node feature, the sample association feature and the sample second-order association feature to obtain an enhanced fusion feature; Based on the immune index processing data set, the immune index analysis model is iteratively trained according to the enhanced fusion features and the corresponding sample immune analysis results.

9. The immune index data processing system based on graph neural network is characterized by: The system for implementing the immune index data processing method based on graph neural network according to any one of claims 1 to 8 comprises: An indicator screening module, the indicator screening module is used to receive an immune detection target, and screen a plurality of immune indicators from an immune indicator library according to the immune detection target; A relationship analysis module, wherein the relationship analysis module is used to perform a pairwise comparison analysis on the multiple immune indicators to obtain an indicator relationship between any two immune indicators among the multiple immune indicators, wherein the indicator relationship includes correlation and correlation strength, wherein the correlation includes positive correlation, negative correlation and no direct correlation; A graph construction module, wherein the graph construction module is used to construct an immune indicator relationship graph based on the multiple immune indicators and the indicator relationship, wherein the nodes in the immune indicator relationship graph correspond to the multiple immune indicators, the positive correlation corresponds to a positive edge, the negative correlation corresponds to a negative edge, and the indirect correlation corresponds to no edge, and the correlation strength is used as the weight of the corresponding edge; A data processing module, the data processing module is used to obtain historical data of immune indicators, process the historical data of immune indicators in combination with the immune indicator relationship diagram, and construct an immune indicator processing data set; A model training module, wherein the model training module is used to train a graph neural network model using the immune index processing data set to obtain an immune index analysis model; A graph embedding module, the graph embedding module is used to receive immune index input data to be analyzed, and perform graph embedding representation on the immune index input data according to the immune index relationship graph to obtain an immune index graph embedding vector; A result analysis module is used to input the immune index graph embedding vector into the immune index analysis model to obtain the immune analysis result for the immune index input data.

10. The immune index data processing system based on graph neural network according to claim 9 is characterized in that: The relationship analysis module includes: A target type acquisition submodule, wherein the target type acquisition submodule is used to acquire the target type of the immune detection target; An indicator retrieval submodule, the indicator retrieval submodule is used to retrieve a candidate immune indicator set corresponding to the target type in the immune indicator library according to the target type; A correlation calculation submodule, wherein the correlation calculation submodule is used to perform correlation calculation on the immune indicators in the candidate immune indicator set to obtain the correlation between each immune indicator and the immune detection target; A threshold screening submodule is used to screen the candidate immune indicator set according to a correlation threshold, and to use the immune indicators with a correlation greater than the correlation threshold as the multiple immune indicators.

Images