Disease classification model training method and device and disease classification system

Abstract

The invention discloses a disease classification model training method and device and a disease classification system, and is applied to the technical field of data processing, and the method comprises the steps: obtaining target disease gene data and a transcriptome data set associated with a target disease; classifying the transcriptome data set to obtain a discovery set, a training set and a test set; according to the discovery set and the target disease gene data, obtaining a plurality of target gene pair features associated with the target disease; processing the training set and the test set according to the multiple target gene pair features to obtain a processed training set and a processed test set; and according to the processed training set and the processed test set, carrying out model training and testing by adopting a machine learning method to obtain a disease classification model. And in the classification system, performing classification processing on the transcriptome data of the to-be-tested sample associated with the target disease according to the disease classification model to obtain the category of the to-be-tested sample. The training precision and training efficiency of the disease classification model are improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a training method and device for a disease classification model and a disease classification system. Background Art

[0002] Transcriptome data refers to the collection of all RNA molecules in an organism and the results of the measurement of their expression levels. In the related art, the transcriptome data of a specific disease is used for model training to obtain a classification model for a specific disease, and then the classification model of a specific disease is used to implement the classification processing of the specific disease. Before model training, the related art usually performs operations such as data standardization and batch effect removal on the transcriptome data in the test set and the training set, aiming to reduce the differences between different batches and ensure the alignment of the data of each batch. However, operations such as data standardization and batch effect removal often require that both the training set and the test set are standardized to the same data distribution space, which means that when a new test set arrives, the current training set and the current test set need to be used to integrate the new test set and re-standardize, or when a new training set arrives, the current test set and the current training set need to be used to integrate the new training set and re-standardize, resulting in a decrease in the training efficiency of the disease classification model. In addition, the related art still uses traditional data processing methods to process different types of transcriptome data, which is not effective in processing across data sets, resulting in a decrease in the accuracy of the disease classification model. Summary of the invention

[0003] The embodiments of the present application provide a disease classification model training method, device and disease classification system for improving the training accuracy and training efficiency of the disease classification model, thereby improving the classification accuracy of the disease classification model.

[0004] On the one hand, the present application embodiment provides a method for training a disease classification model, comprising the following steps: Obtaining target disease gene data and a transcriptome data set associated with the target disease; wherein the transcriptome data set includes transcriptome data and categories of several initial samples; Classifying the transcriptome data set to obtain a discovery set, a training set, and a test set; According to the discovery set and the target disease gene data, a plurality of target gene pair features associated with the target disease are obtained; Processing the training set and the test set according to the plurality of target gene pair features to obtain the processed training set and the processed test set; Model training and testing are performed using a machine learning method based on the processed training set and the processed test set to obtain a disease classification model, where the disease classification model is used to identify categories of samples associated with the target disease.

[0005] On the other hand, an embodiment of the present application provides a training device for a disease classification model, comprising: A first acquisition module is used to acquire target disease gene data and a transcriptome data set associated with the target disease; wherein the transcriptome data set includes transcriptome data and categories of several initial samples; A first processing module is used to classify the transcriptome data set to obtain a discovery set, a training set and a test set; A second processing module, configured to obtain a plurality of target gene pair features associated with the target disease based on the discovery set and the target disease gene data; A third processing module is used to process the training set and the test set according to the characteristics of the plurality of target gene pairs to obtain the processed training set and the processed test set; The fourth processing module is used to perform model training and testing using a machine learning method based on the processed training set and the processed test set to obtain a disease classification model, wherein the disease classification model is used to identify categories of samples associated with the target disease.

[0006] On the other hand, an embodiment of the present application provides a disease classification system, including: A second acquisition module is used to acquire transcriptome data of a sample to be tested associated with a target disease; The fifth processing module is used to classify the transcriptome data of the sample to be tested according to the disease classification model to obtain the category of the sample to be tested; wherein the disease classification model is trained by a disease classification model training method as described in any one of claims 1-8.

[0007] The beneficial effects of the present application are: providing a training method, device and disease classification system for a disease classification model, in which the target disease gene data and the transcriptome data set associated with the target disease are obtained in the training method and device; the transcriptome data set is classified and processed to obtain a discovery set, a training set and a test set; based on the discovery set and the target disease gene data, multiple target gene pair features associated with the target disease are obtained; the training set and the test set are processed according to the multiple target gene pair features to obtain a processed training set and a processed test set; a machine learning method is used to train and test the model according to the processed training set and the processed test set to obtain a disease classification model. In the classification system, the transcriptome data of the sample to be tested associated with the target disease is classified and processed according to the disease classification model to obtain the category of the sample to be tested. The present application can improve the training accuracy and training efficiency of the disease classification model, thereby improving the classification accuracy of the disease classification model. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a structural diagram of the training method of the disease classification model provided in this application; Figure 2 It is a schematic diagram of the training method of the disease classification model provided in this application; Figure 3 This is a comparison chart of the effects of the skin cancer classification task provided by this application; Figure 4 This is a comparison chart of the effects of the skin cancer immunotherapy response prediction task provided in this application. DETAILED DESCRIPTION

[0009] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0010] At present, transcriptome data can be obtained through deoxyribonucleic acid (DNA) microarray, transcriptome sequencing (RNA sequencing, RNA-seq) technology or single-cell RNA sequencing (scRNA-seq) technology. However, no matter which sequencing technology is used, the final result is often presented in the form of a gene expression matrix, which provides a basis for various downstream analyses, such as differential expression analysis, cluster analysis, principal component analysis, functional enrichment analysis and other tasks. In related technologies, the transcriptome data of a specific disease is used for model training to obtain a classification model for a specific disease, and then the classification model of a specific disease is used to implement the classification of the specific disease. Due to the differences between different sequencing technologies, the differences between different data normalization methods, and the experimental errors caused by different experimental conditions, there are large data differences between the test set and the training set. Therefore, before model training, related technologies usually perform operations such as data standardization and batch effect removal on the transcriptome data in the test set and the training set, aiming to reduce the differences between different batches and ensure the alignment of the data of each batch. However, operations such as unified data standardization and batch effect removal often require that both the training set and the test set be standardized to the same data distribution space, which means that when a new test set arrives, the current training set and the current test set need to be used to integrate the new test set and re-standardize, or when a new training set arrives, the current test set and the current training set need to be used to integrate the new training set and re-standardize, resulting in a decrease in the training efficiency of the disease classification model. In addition, the related art still uses traditional data processing methods to process different types of transcriptome data, which is not effective in processing across data sets, resulting in a decrease in the accuracy of the disease classification model. It is understandable that, since the sample type of the test set is different from the sample type of the training set, cross-datasets can refer to test sets and training sets. In addition, since the types of samples contained in the test set or training set are also different, cross-datasets can also refer to data sets composed of different types of transcriptome data.

[0011] In view of this, the embodiments of the present application provide a training method, device and disease classification system for a disease classification model, aiming to improve the training accuracy and training efficiency of the disease classification model, thereby improving the classification accuracy of the disease classification model.

[0012] It should be emphasized that the input of the disease classification model in the embodiment of the present application is a data sample pre-stored in a database, not data sampled from a living body, and the output of the disease classification model is the category of samples associated with the target disease, and is not intended to obtain disease diagnosis results or health status. The output of the disease classification model is only used to assist doctors in diagnosis, treatment or medical learning, and the final diagnosis result or treatment result still needs to be decided by the doctor.

[0013] First, the implementation steps of the disease classification model training method provided in this application will be described in detail below with reference to the accompanying drawings.

[0014] The training method of the disease classification model provided in this application can be applied to a terminal, a server, or software running in a terminal or a server. The terminal can be a tablet computer, a laptop computer, a desktop computer, etc., but is not limited to this. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms. In addition, the server can also be a node server in a blockchain network, but is not limited to this. Among them, blockchain is a new application model of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.

[0015] Reference Figure 1 , Figure 1 It is a flowchart of the training method of the disease classification model provided in the present application. The training method of the disease classification model may include the following steps S100-S500.

[0016] S100, obtaining target disease gene data and a transcriptome dataset associated with the target disease.

[0017] It should be noted that the target disease refers to the classification object of the disease classification model in the embodiment of the present application, and its type can be set according to the actual situation. For example, the target disease can be skin cancer, liver cancer, etc., but it is not limited to this. In addition, the target disease gene data refers to the relevant gene type or highly variable gene type of the target disease, which can be data pre-stored in a specific database, wherein the highly variable gene type refers to the type of gene whose expression amount in different cells or individuals is significantly different.

[0018] It is understandable that the transcriptome data set includes transcriptome data and categories of several initial samples. Among them, the transcriptome data of the initial sample may include at least one of the microarray data of the initial sample, the transcriptome sequencing data of the initial sample, or the single-cell transcriptome sequencing data of the initial sample. Among them, the microarray data refers to the initial microarray data after pretreatment, and the initial microarray data refers to the data obtained by measuring the initial sample by DNA microarray; the transcriptome sequencing data refers to the transcriptome sequencing expression matrix after pretreatment, and the transcriptome sequencing expression matrix refers to the data obtained by measuring the initial sample by RNA-seq technology; the single-cell transcriptome sequencing data refers to the single-cell transcriptome sequencing expression matrix after pretreatment, and the single-cell transcriptome sequencing expression matrix refers to the data obtained by measuring the initial sample by scRNA-seq technology. In addition, the category of the initial sample refers to the target disease category to which the initial sample belongs. For example, in the skin cancer classification task, the category of the initial sample may include melanoma, basal cell carcinoma, Merkel cell carcinoma, squamous cell carcinoma and non-skin cancer, but is not limited to this. For another example, in a skin cancer immunotherapy response prediction task, the categories of the initial samples may include two categories: abnormal skin cancer immunotherapy response and normal skin cancer immunotherapy response, but are not limited thereto.

[0019] In this step, target disease gene data is obtained, and the target disease gene data is used to determine the target gene pair characteristics associated with the target disease, and transcriptome data and categories of several initial samples associated with the target disease are obtained and a transcriptome dataset is constructed, and the transcriptome dataset is used to determine the target gene pair characteristics associated with the target disease and perform model training and testing.

[0020] S200, classify the transcriptome data set to obtain a discovery set, a training set, and a test set.

[0021] It should be noted that the discovery set may include transcriptome data and categories of multiple discovery samples. Among them, the discovery sample refers to the initial sample classified into the discovery set, and the transcriptome data of the discovery sample may include any one of the single-cell transcriptome sequencing data of the discovery sample or the bulk transcriptome data of the discovery sample, and the bulk transcriptome data of the discovery sample may include the microarray data of the discovery sample and the transcriptome sequencing data of the discovery sample.

[0022] It is understandable that the training set may include transcriptome data and categories of multiple training samples. Wherein, the training sample refers to the initial sample divided into the training set, and the transcriptome data of the training sample may include microarray data of the training sample and transcriptome sequencing data of the training sample. In addition, the test set may include transcriptome data and categories of multiple test samples. Wherein, the test sample refers to the initial sample divided into the test set, and the transcriptome data of the test sample may include at least one of microarray data of the test sample, transcriptome sequencing data of the test sample, or single-cell transcriptome sequencing data of the test sample.

[0023] In this step, the transcriptome dataset is divided into three datasets: discovery set, training set and test set according to the preset division ratio. The discovery set is used to determine the target gene pair features associated with the target disease, while the training set and test set are used to participate in the training and testing of the model.

[0024] Optionally, the division ratio can be set according to actual conditions, and the present application embodiment does not specifically limit this. For example, the transcriptome data of several initial samples are divided into a discovery set, a training set, and a test set according to a division ratio of 2:5:3, but it is not limited thereto.

[0025] Optionally, the discovery set and the training set may be data sets of the same type or different types. For example, when the transcriptome data of each training sample in the training set is microarray data and transcriptome sequencing data, the transcriptome data of each discovery sample in the discovery set may be microarray data and transcriptome sequencing data, or may be single-cell transcriptome sequencing data.

[0026] S300, obtaining a plurality of target gene pair features associated with the target disease based on the discovery set and the target disease gene data.

[0027] It should be noted that the target gene pair feature refers to a feature obtained by combining two gene data that are most relevant to the target disease, and the gene data is the gene type.

[0028] In this step, after obtaining the discovery set and target disease gene data, the target disease gene data and the transcriptome data and categories of each discovery sample in the discovery set are processed to determine multiple gene data most relevant to the target disease, thereby constructing multiple target gene pair features associated with the target disease.

[0029] Optionally, the total number of target gene pair features can be set according to actual conditions, and this embodiment does not specifically limit this. For example, the total number of target gene pair features can be 50 or 100, or an integer value between 50 and 100, but is not limited thereto.

[0030] S400, processing the training set and the test set according to the characteristics of the plurality of target gene pairs to obtain a processed training set and a processed test set.

[0031] It should be noted that the processed training set may include multiple gene pair features and categories corresponding to each training sample, and the multiple gene pair features corresponding to the training sample are the features of the transcriptome data of each training sample, and the processed test set may include multiple gene pair features and categories corresponding to each test sample, and the multiple gene pair features corresponding to the test sample are the features of the transcriptome data of the test sample.

[0032] It can be understood that the processed training set and the processed test set are in the same data distribution space.

[0033] In this step, the training set and the test set are processed using multiple target gene pairs to extract features from the training set and the test set, and the training set and the test set are located in the same data distribution space, thereby obtaining the processed training set and the processed test set. It is understandable that in practical applications, when a new test set arrives, the new test set, the current test set, and the current training set can be standardized to the same data distribution space and feature extraction can be achieved through multiple target gene pairs, or when a new training set arrives, the new training set, the current test set, and the current training set can be standardized to the same data distribution space and feature extraction can be achieved through multiple target gene pairs. In this way, using multiple target gene pairs as the processing benchmark for the test set and the training set can not only effectively ensure the data uniformity between the test set and the training set, but also ensure the data uniformity within the test set and the data uniformity within the training set, thereby improving the training accuracy and training efficiency of the disease classification model.

[0034] S500, performing model training and testing using a machine learning method based on the processed training set and the processed test set to obtain a disease classification model. It should be noted that the disease classification model is used to identify the category of samples associated with the target disease. Exemplarily, in the skin cancer classification task, the disease classification model can be used to identify the specific type of skin cancer, such as melanoma, basal cell carcinoma, Merkel cell carcinoma, squamous cell carcinoma and non-skin cancer. Exemplarily, in the skin cancer immunotherapy response prediction task, the disease classification model can be used to identify whether the skin cancer immunotherapy response is normal, that is, the disease classification model outputs the result of normal skin cancer immunotherapy response or abnormal skin cancer immunotherapy response.

[0035] In this step, after obtaining the processed training set and the processed test set, a machine learning method is used to perform model training on the processed training set, and the model is tested on the processed test set, thereby obtaining a disease classification model.

[0036] Optionally, the machine learning method can be set according to actual conditions, and the embodiments of the present application do not specifically limit this. For example, the machine learning method may include logistic regression (LR), random forest (RF), K-nearest neighbor (KNN), support vector machine (SVM), etc., but is not limited thereto.

[0037] From the above, it can be seen that although the current gene pair analysis relies on the expression value of the gene, the expression values ​​of the two genes are often consistent in different data sets of the same tissue. Therefore, the embodiment of the present application believes that the characteristics of the gene pair can be applied to cross-dataset research and disease classification tasks, especially skin cancer classification tasks or skin cancer immunotherapy prediction tasks. Of course, the embodiment of the present application can also be applied to other disease classification tasks. Specifically, in the embodiment of the present application, the target disease gene data and the transcriptome data of several initial samples associated with the target disease are first obtained, and then the transcriptome data of several initial samples are classified and processed to obtain a discovery set, a training set and a test set. Then, the sample data of the discovery set is used to explore multiple target gene pair features associated with the target disease. These target gene pair features are less affected by the differences between different sequencing technologies, the differences between different data normalization methods, and the experimental errors caused by different experimental conditions. It has better generalization ability across data sets, and it can connect transcriptome data of different modalities in series. For example, Single-cell transcriptome sequencing data and transcriptome sequencing data can be connected in series, which is conducive to cross-dataset processing. These target gene pair features are then used to process the training set and the test set. This not only allows the gene pair features of the dataset to be accurately extracted as the sample features of the dataset, but also ensures that different types of transcriptome data are in the same data distribution space, effectively improving the processing efficiency and accuracy across datasets. Finally, the processed training set and the processed test set are used for model training and testing, thereby effectively improving the training accuracy and efficiency of the disease classification model, and further improving the interpretability and accuracy of the disease classification model.

[0038] The above steps will be further described below.

[0039] In some embodiments, reference Figure 2 , Figure 2It is a schematic diagram of the training method of the disease classification model provided in the present application. In the above step S100, the implementation process of obtaining the target disease gene data and the transcriptome data set associated with the target disease may include the following steps S110-S140.

[0040] S110, acquiring a gene type associated with a target disease as target disease gene data.

[0041] It should be noted that the gene type associated with the target disease refers to the relevant gene type or highly variable gene type of the target disease, which can be pre-stored in the corresponding database.

[0042] In this step, a keyword search is performed on a preset biological information database, aiming to retrieve and output the gene type associated with the target disease from the biological information database as the target disease gene data.

[0043] Optionally, the biological information database can be set according to actual conditions, and this embodiment does not specifically limit this. For example, the biological information database can be a database of a regional biotechnology information center, but is not limited thereto.

[0044] Optionally, the search keywords for the biological information database can be set according to actual conditions, and this embodiment does not specifically limit this. For example, when facing a skin cancer classification task or a skin cancer immunotherapy response prediction task, the search keyword can be skin cancer or Skin cancer, but is not limited thereto.

[0045] S120, obtaining initial transcriptome data and categories of several initial samples associated with the target disease.

[0046] It should be noted that the initial transcriptome data of the initial sample may include at least one of the transcriptome sequencing expression matrix of the initial sample, the single-cell transcriptome sequencing expression matrix of the initial sample, or the original microarray data of the initial sample. Among them, the initial microarray data refers to the data obtained by measuring the initial sample by DNA microarray; the transcriptome sequencing expression matrix refers to the data obtained by measuring the initial sample by RNA-seq technology, which is essentially a gene expression matrix; the single-cell transcriptome sequencing expression matrix refers to the data obtained by measuring the initial sample by scRNA-seq technology, which is essentially a gene expression matrix.

[0047] In this step, initial transcriptome data of several initial samples associated with the target disease are obtained, and at the same time, categories of several initial samples associated with the target disease are obtained, so as to construct a transcriptome dataset based on these data in subsequent steps.

[0048] S130, performing standardization processing on the initial transcriptome data of each initial sample to obtain the transcriptome data of each initial sample.

[0049] It should be noted that the transcriptome data of the initial sample may include at least one of the microarray data of the initial sample, the transcriptome sequencing data of the initial sample, or the single-cell transcriptome sequencing data of the initial sample.

[0050] In this step, for the initial transcriptome data of each initial sample, the initial transcriptome data of the current initial sample is preprocessed by a preset standardization method to obtain the transcriptome data of the current initial sample. By traversing the initial transcriptome data of several initial samples, the transcriptome data of several initial samples can be obtained. Specifically, for the initial transcriptome data of each initial sample, the initial microarray data of the current initial sample is standardized by the robust multi-array average method (RMA) to obtain the microarray data of the current initial sample, which can reduce technical variation and improve the comparability of data; and the transcriptome sequencing expression matrix of the current initial sample is standardized by the million transcripts method (TPM) to obtain the transcriptome sequencing data of the current initial sample, and the single-cell transcriptome sequencing expression matrix of the current initial sample is standardized by the million transcripts method to obtain the single-cell transcriptome sequencing data of the current initial sample, so that the gene expression between different samples can be comparable.

[0051] S140, obtaining a transcriptome data set associated with the target disease according to the transcriptome data and category of each initial sample.

[0052] In this step, after completing the standardized preprocessing, the transcriptome data set associated with the target disease is constructed through the transcriptome data and categories of each initial sample, thereby realizing the construction and processing of the transcriptome data set, effectively improving the accuracy of the transcriptome data set, thereby ensuring the accuracy of the subsequent target gene pair features and the training accuracy of the disease classification model.

[0053] In some embodiments, reference Figure 2 In the above step S300, the implementation process of obtaining multiple target gene pair features associated with the target disease based on the discovery set and the target disease gene data may include the following steps S310-S340.

[0054] S310, obtaining a plurality of target gene data associated with the target disease according to the transcriptome data of each discovery sample in the discovery set, the transcriptome data of each training sample in the training set, and the target disease gene data.

[0055] It should be noted that target gene data refers to gene types that exist in the discovery set, training set, and target disease gene data at the same time.

[0056] In this step, the genes in the intersection of the discovery set, the training set and the target disease gene data are taken as the target gene data. Specifically, first, based on the transcriptome data of each discovery sample in the discovery set, at least one discovery set gene data is obtained, and the discovery set gene data refers to the gene data existing in the discovery set, and the gene data is the gene type. At the same time, based on the transcriptome data of each training sample in the training set, at least one training set gene data is obtained, and the training set gene data refers to the gene data existing in the training set. Then, the intersection result of at least one discovery set gene data, at least one training set gene data and the target disease gene data is determined as a plurality of target gene data associated with the target disease, that is, the gene type existing in the discovery set, the training set and the target disease gene data at the same time is determined as the target gene data associated with the target disease and output.

[0057] S320, obtaining a gene expression matrix of the discovery set according to the multiple target gene data and the transcriptome data of each discovery sample in the discovery set.

[0058] It should be noted that the gene expression matrix of the discovery set is a second-order matrix. In the gene expression matrix of the discovery set, the rows represent the target gene data, the columns represent the discovery samples, the value of the element is the gene expression level of the target gene data in the discovery sample, and the gene expression level is a specific, measured value. For example, in a certain cell sample, the gene expression level of gene A is 3, and the gene expression level of gene B is 5, but it is not limited to this.

[0059] In this step, first, for each target gene data, the gene expression amount of the current target gene data in each discovery sample is determined based on the current target gene data and the transcriptome data of each discovery sample in the discovery set. It can be understood that the gene expression amount is measured by a preset sequencing technology and stored in the transcriptome data in the form of a specific numerical value. The gene expression amount of each target gene data in each discovery sample can be obtained by traversing multiple target gene data. Then, the gene expression matrix of the discovery set is constructed by using the target gene data as the row of the gene expression matrix of the discovery set, the discovery samples as the columns of the gene expression matrix of the discovery set, and the gene expression amount of each target gene data in each discovery sample as the value of each element in the gene expression matrix of the discovery set. The gene expression matrix of the discovery set contains each discovery sample and its corresponding multiple target gene data, such as Figure 2 As shown in part (b) of .

[0060] Alternatively, the gene expression matrix is ​​represented as , as shown in the following formula (1): (1); In formula (1), Represents the total number of genes, that is, the number of rows in the gene expression matrix; Represents the number of samples, that is, the number of columns in the gene expression matrix; element refers to the gene expression matrix Line Elements of columns, elements To indicate the The gene in The gene expression level of each sample, , It will be appreciated that the representation of the gene expression matrix is ​​applicable to the discovery set, the training set, and the test set.

[0061] S330, obtaining a gene pair matrix of the discovery set according to the gene expression matrix of the discovery set.

[0062] It should be noted that the gene pair matrix of the discovery set is a second-order matrix, in which the rows represent the discovery set gene pairs, which are composed of two target gene data, the columns represent the discovery samples, and the values ​​of the elements are the values ​​of the discovery set gene pairs in the discovery samples. The values ​​of the discovery set gene pairs in the discovery samples include any one of the first preset value, the second preset value, or the third preset value.

[0063] In this step, in order to construct an adaptive personalized gene pair matrix, exhaustively compare every two target gene data in each discovery sample according to the gene expression matrix of the discovery set, introduce the adaptive difference of the Welch test to capture the difference changes between different genes, and adaptively quantify the difference changes between different genes, so as to obtain multiple discovery set gene pairs and the value of each discovery set gene pair in each discovery sample. Then, the discovery set gene pairs are used as the rows of the discovery set gene pair matrix, the discovery samples are used as the columns of the discovery set gene pair matrix, and the values ​​of each discovery set gene pair in each discovery sample are used as the values ​​of each element in the discovery set gene pair matrix, so as to construct the discovery set gene pair matrix. The discovery set gene pair matrix includes each discovery sample and its corresponding multiple discovery set gene pairs, such as Figure 2 As shown in part (c) of , the gene expression matrix of the discovery set is converted into the gene pair matrix of the discovery set.

[0064] Alternatively, the gene pair matrix is ​​represented as , as shown in the following formula (2): (2); In formula (2), Represents the total number of gene pairs, that is, the number of rows in the gene pair matrix, where the total number of gene pairs is The number of combinations of data from two target genes from each gene; Represents the number of samples, that is, the number of columns in the gene pair matrix; element It refers to the gene pair matrix Line Elements of columns, elements To indicate the The gene pair The value of the samples, , It can be understood that the representation of the gene pair matrix is ​​applicable to the discovery set, training set and test set.

[0065] S340, screening and processing each discovery set gene pair in the discovery set gene pair matrix to obtain a plurality of target gene pair features associated with the target disease.

[0066] It should be noted that the target gene pair feature refers to a feature obtained by combining two gene data most relevant to the target disease. In other words, the target gene pair feature may include two gene data most relevant to the target disease, and both of the two gene data most relevant to the target disease are gene types that are significantly reversed in the categories of different discovery samples, and significant reversal means significantly improving the effect of immunotherapy response of the target disease or completely reversing the defects caused by the target disease.

[0067] In this step, after obtaining the gene pair matrix of the discovery set, cross-category analysis is performed. Specifically, Fisher's exact test or Pearson's chi-square test is applied to calculate the significance level value of each discovery set gene pair in the discovery set gene pair matrix, and the significance level value of each discovery set gene pair is used to screen each discovery set gene pair, aiming to screen out the discovery set gene pairs that are significantly reversed in different categories as the target gene pair feature output associated with the target disease, thereby obtaining multiple target gene pair features associated with the target disease.

[0068] In some embodiments, reference Figure 2 In the above step S330, the process of obtaining the gene pair matrix of the discovery set according to the gene expression matrix of the discovery set may include the following steps S331-S335: S331, taking out the number of combinations of two target gene data from the plurality of target gene data and determining it as the total number of gene pairs in the discovery set; S332, determining a plurality of discovery set gene pairs according to the total number of discovery set gene pairs and the gene expression matrix of the discovery set; S333, according to the gene expression matrix of the discovery set, obtaining the variance of the gene expression amount of each target gene data in all discovery samples as the expression amount variance of each target gene data; S334, obtaining the value of each discovery set gene pair in each discovery sample according to the gene expression matrix of the discovery set, the expression variance of each target gene data and the preset adaptive coefficient; S335, constructing the gene pair matrix of the discovery set with the discovery set gene pairs as the rows of the discovery set gene pair matrix, with the discovery samples as the columns of the discovery set gene pair matrix, and with the numerical values ​​of each discovery set gene pair in each discovery sample as the values ​​of each element in the discovery set gene pair matrix.

[0069] In this embodiment, first, The number of combinations of 2 target gene data from the target gene data is determined as the total number of gene pairs in the discovery set, that is, the total number of gene pairs in the discovery set is , and the gene expression matrix of the discovery set is sorted according to the total number of gene pairs in the discovery set The target gene data are combined and calculated to obtain There are discovery set gene pairs, and each discovery set gene pair is composed of two target gene data.

[0070] Then, according to the value of each element in the gene expression matrix of the discovery set, the variance of the gene expression of each target gene data in all discovery samples is calculated as the expression variance of each target gene data. Among them, the expression variance of the target gene data can describe the dispersion and volatility of the target gene data in all discovery samples. Since the calculation of variance belongs to the prior art, this embodiment will not be repeated here.

[0071] Afterwards, for each gene pair in the discovery set, the value of each element in the gene expression matrix of the discovery set, the expression variance of the target gene data and the preset adaptive coefficient are used to obtain the gene pair in the current discovery set. The value of the sample found. By traversing The discovery set gene pairs can be obtained The discovery set gene pairs The value of the discovered samples.

[0072] Specifically, for two target gene data contained in a single discovery sample and a single discovery set gene pair, the following steps are performed: first, the difference between the gene expression level of one of the target gene data in the discovery sample and the gene expression level of the other target gene data in the discovery sample is obtained as the expression level difference of the discovery set gene pair; then, the sum of the expression level variances of the two target gene data is obtained as the variance sum value, and the square root of the variance sum is taken to obtain the square root of the variance; thereafter, the product of the square root of the variance and the adaptive coefficient is obtained as the quantitative change value of the discovery set gene pair, and the quantitative change value can reflect the difference in expression levels between the two target gene data in the discovery set gene pair; finally, the size relationship between the gene expression levels of the two target gene data in the discovery sample is finally determined by comparing the expression level difference of the discovery set gene pair and the quantitative change value of the discovery set gene pair, and the size relationship is used to determine the numerical value of the discovery set gene pair in the discovery sample. More specifically, if the difference in expression of the discovery set gene pair is greater than the quantitative change value of the discovery set gene pair, it means that the gene expression of one of the target gene data in the discovery sample is greater than the gene expression of the other target gene data in the discovery sample, and the value of the discovery set gene pair in the discovery sample is determined as the first preset value. If the difference in expression of the discovery set gene pair is less than the quantitative change value of the discovery set gene pair, it means that the gene expression of one of the target gene data in the discovery sample is less than the gene expression of the other target gene data in the discovery sample, and the value of the discovery set gene pair in the discovery sample is determined as the second preset value. If the difference in expression of the discovery set gene pair is equal to the quantitative change value of the discovery set gene pair, it means that the gene expression of the two target gene data in the discovery sample is the same, and the value of the discovery set gene pair in the discovery sample is determined as the third preset value.

[0073] Finally, the discovery set gene pairs are used as the rows of the discovery set gene pair matrix, the discovery samples are used as the columns of the discovery set gene pair matrix, and the values ​​of each discovery set gene pair in each discovery sample are used as the values ​​of each element in the discovery set gene pair matrix, so as to construct an adaptive individualized gene pair matrix, that is, the discovery set gene pair matrix. The gene pair matrix contains each discovery sample and its corresponding multiple discovery set gene pairs, such as Figure 2 As shown in part (c) of .

[0074] It can be seen that in the process of constructing the gene pair matrix, this implementation method first obtains multiple gene pairs from the gene expression matrix by combination, and then determines the expression difference between the two genes in each gene pair based on the rank method, and uses these expression differences to assign values ​​to each element of the gene pair matrix. Finally, the gene pair matrix is ​​constructed with the gene pairs as the rows of the gene pair matrix and the samples as the columns of the gene pair matrix.

[0075] The current rank-based method often measures the relative relationship between the expression levels of different genes by the difference between the gene expression levels of different genes. This method often ignores the true difference between the gene expression levels of different genes, and its accuracy needs to be improved. Therefore, this embodiment introduces adaptive quantification of the difference change of gene pairs in the rank-based method, which is specifically manifested in capturing the difference change of the gene expression level between the two genes in the gene pair through the Welch test, and quantifying the difference change of these gene expression levels, so as to better reflect the expression difference between the two genes in the gene pair and assign values ​​to each element of the gene pair matrix, thereby retaining the quantitative information between the gene expression levels of the two genes in the gene pair, so that more quantitative information can be retained for the rank-based method, and the relative expression of the gene pair is reduced by the difference between different sequencing technologies, the difference between different data normalization methods, and the experimental error caused by different experimental conditions. Since the relative expression of the gene pair is reduced by the influence of technical variation, the gene pair can connect the transcriptome data of different modes in series, thereby effectively improving the generalization ability of the gene pair in cross-dataset processing. In addition, this embodiment can accurately determine the value of each element in the gene pair matrix, improve the accuracy of the gene pair matrix, and ensure the accuracy of subsequent target gene pair features, thereby improving the processing accuracy and efficiency across data sets.

[0076] Optionally, the adaptive coefficient can be set according to actual conditions, and the present embodiment does not specifically limit this. For example, the adaptive coefficient can be 0, 1 or 2, but is not limited thereto. According to the actual experimental results of the present application embodiment, when the transcriptome data of each discovery sample in the discovery concentration is single-cell transcriptome sequencing data, the adaptive coefficient can take a value of 2, and when the transcriptome data of each discovery sample in the discovery concentration is batch transcriptome data, the adaptive coefficient can take a value of 0, so that the best overall performance can be ensured.

[0077] Optionally, for discovery set gene pairs and The samples found, The discovery set gene pairs are target gene data and target gene data, if the first preset value is 1, the second preset value is -1, and the third preset value is 0, then The discovery set gene pairs are in The value of the discovered samples can be expressed as follows: , (3); In formula (3), the element It refers to the gene pair matrix Line The elements of the column, , ,element To indicate the The discovery set gene pairs are in The number of samples found; Indicates The variance of gene expression of target gene data in all discovered samples; Indicates The variance of gene expression of target gene data in all discovered samples; represents the adaptive coefficient; Can describe the target gene data and The quantitative change of gene expression level between target gene data. In this formula, when When The target gene data is in The gene expression level of the sample is greater than that of The target gene data is in The gene expression level of the sample The discovery set gene pairs are in The value of the discovered sample is 1. When The target gene data is in The gene expression level of the sample is less than that of The target gene data is in The gene expression level of the sample The discovery set gene pairs are in The value of the discovered sample is -1. When The target gene data is in The gene expression level of the sample is equal to that of the The target gene data is in The gene expression level of the sample The discovery set gene pairs are in The value of the sample found is 0. For example, for a gene pair (A, B) consisting of gene A and gene B, the gene expression level of gene pair A in a certain sample is 3, and the gene expression level of gene B in a certain sample is 5, then the difference between the gene expression level of gene pair A and the gene expression level of gene pair B is calculated as the expression level difference of gene pair (A, B), that is, 3-5=-2; represents the variance of gene A in all samples, Represents the variance of gene B in all samples, adaptive coefficient If the value is 1, the quantitative change value of the gene pair (A, B) is 1; since the difference in the expression levels of the gene pair (A, B) is less than the quantitative change value of the gene pair (A, B), the actual expression level of gene B is greater than the actual expression level of gene A, so the value of the gene pair (A, B) in this sample is -1.

[0078] In some embodiments, reference Figure 2 In the above step S340, the process of screening and processing each discovery set gene pair in the discovery set gene pair matrix to obtain multiple target gene pair features associated with the target disease may include the following steps S341-S343.

[0079] S341, for each discovery set gene pair, according to the value of the discovery set gene pair in each discovery sample and the category of all discovery samples, obtain the first target quantity and the second target quantity of each category relative to the discovery set gene pair as the category quantity feature of the discovery set gene.

[0080] It should be noted that the category refers to the category of the discovery sample, such as melanoma, basal cell carcinoma, Merkel cell carcinoma, squamous cell carcinoma and non-skin cancer, etc. Among them, for each category, the first target number of the current category relative to the discovery set gene pair refers to the total number of discovery set gene pairs with a value of the first preset value in the discovery samples belonging to the current category, and the second target number of the current category relative to the discovery set gene pair refers to the total number of discovery set gene pairs with a value of the second preset value in the discovery samples belonging to the current category.

[0081] In this step, first, for each discovery set gene pair, there are: for each category, according to the value of each element in the discovery set gene pair matrix, the total number of discovery set gene pairs with a first preset value in the discovery samples belonging to the current category is obtained as the first target number of the current category relative to the discovery set gene pairs, and the total number of discovery set gene pairs with a second preset value in the discovery samples belonging to the current category is obtained as the second target number of the current category relative to the discovery set gene pairs; by traversing the categories of all discovery samples, the first target number and the second target number of all categories relative to the discovery set gene pairs can be obtained as the category quantity characteristics of the discovery set gene pairs. By traversing all discovery set gene pairs, the category quantity characteristics of all discovery set gene pairs can be obtained, and the category quantity characteristics of each discovery set gene pair are used to calculate the significance level value of the discovery set gene pair.

[0082] Optionally, the number of categories for a single discovery set gene pair can be characterized by the matrix To express, if Figure 2As shown by the dotted line in part (d) of . In this matrix, the first row represents the first preset value, the second row represents the second preset value, and the columns represent the categories of the discovered samples, where for the first row and all columns of the matrix, the value of the element is the first target number of each category relative to the discovery set gene pair, and for the second row and all columns of the matrix, the value of the element is the second target number of each category relative to the discovery set gene pair. The category quantity feature of a single discovery set gene pair satisfies the following formula (4): (4); In formula (4), Indicates category Relative to the first target number of gene pairs in the discovery set, Indicates category Number of secondary targets relative to the discovery set gene pair. Indicates the total number of categories of samples found. It can be understood that if the first preset value is 1 and the second preset value is -1, then It represents the first target number of the first category relative to the discovery set gene pairs, that is, the total number of discovery set gene pairs with a value of 1 in the discovery samples belonging to the first category. Indicates the category number, Indicates category Relative to the first target number of gene pairs in the discovery set, i.e., in the category The total number of discovery set gene pairs with a value of 1 in the discovery sample, represents the second target number of the first category relative to the discovery set gene pairs, that is, the total number of discovery set gene pairs with a value of -1 in the discovery samples belonging to the first category. Indicates category The number of second targets relative to the discovery set gene pairs, i.e., in the category The total number of discovery set gene pairs with a value of -1 in the discovery sample. S342, obtaining the significance level value of each discovery set gene pair according to the category quantity characteristics of each discovery set gene pair and the total category quantity of the discovery samples.

[0083] In this step, different significance level calculation methods are used according to the total number of categories of the discovery samples. Specifically, for each discovery set gene pair, when the total number of categories is 2, that is, , Fisher's exact test was used to calculate the significance level of the gene pairs in the discovery set, as shown in the following formula (5):

[0084] (5); In formula (5), Indicates the significance level value of the gene pair in the discovery set; represents the first target number of the second category relative to the discovery set gene pairs, that is, the total number of discovery set gene pairs with a value of 1 in the discovery samples belonging to the second category; It represents the second target number of the second category relative to the discovery set gene pairs, that is, the total number of discovery set gene pairs with a value of -1 in the discovery samples belonging to the second category.

[0085] When the total number of categories is greater than 2, that is, , the Pearson chi-square test is used to calculate the significance level of the discovery set gene pair. The significance level of the discovery set gene pair is the right-tailed probability of the test statistic of the Pearson chi-square test, as shown in the following formula (6): (6); In formula (3), represents the test statistic for the discovery set gene pair; Indicates that in all found samples, or The sum of the number of The degrees of freedom are The chi-square distribution of .

[0086] S343, screening and processing each discovery set gene pair in the discovery set gene pair matrix according to the significance level value of each discovery set gene pair, and obtaining multiple target gene pair features associated with the target disease.

[0087] It should be noted that the significance level value of the discovery set gene pair is negatively correlated with the significant reversal degree of the discovery set gene pair, that is, the smaller the significance level value of the discovery set gene pair, the higher the significant reversal degree of the discovery set gene pair, and the more likely the discovery set gene pair is to be significantly reversed; the larger the significance level value of the discovery set gene pair, the lower the significant reversal degree of the discovery set gene pair, and the less likely the discovery set gene pair is to be significantly reversed.

[0088] It can be understood that the target gene pair feature refers to the discovery set gene pair whose significance level value is greater than the preset level threshold, or the target gene pair feature refers to the discovery set gene pair whose arrangement number is less than or equal to the preset number threshold when all the discovery set gene pairs are arranged from small to large according to the size of the significance level value.

[0089] In this step, since the smaller the significance level value of the discovery set gene pair, the more likely the discovery set gene pair is to be significantly reversed, each discovery set gene pair in the discovery set gene pair matrix is ​​screened according to the significance level value of each discovery set gene pair, aiming to screen out the discovery set gene pairs with lower significance level values ​​as the target gene pair features. In this way, representative target gene pair features that are significantly reversed in each category can be accurately screened to improve the accuracy of the target gene pair features, thereby improving the processing accuracy and efficiency across data sets.

[0090] Exemplarily, in some embodiments, screening is performed by sorting. Specifically, first, all discovery set gene pairs are sorted using the significance level values ​​of each discovery set gene pair to obtain the discovery set gene pairs after sorting. Among them, the sorting number of the discovery set gene pair is positively correlated with the significance level value of the discovery set gene pair, that is, the smaller the significance level value of the discovery set gene pair, the smaller the sorting number of the discovery set gene pair, and the larger the significance level value of the discovery set gene pair, the larger the sorting number of the discovery set gene pair. Then, from all the sorted discovery set gene pairs, the first N discovery set gene pairs (i.e., discovery set gene pairs with arrangement numbers less than or equal to N, N is a preset numbering threshold) are selected as multiple target gene pair features associated with the target disease. Among them, the numbering threshold can be flexibly set according to the actual situation, and this embodiment is not specifically limited to this. For example, the numbering threshold can be 50 or 100, but is not limited to this.

[0091] For example, in other embodiments, a threshold comparison method is used for screening. Specifically, the discovery set gene pairs with significance level values ​​less than a preset level threshold are used as target gene pair features. The level threshold can be flexibly set according to actual conditions, and this embodiment does not specifically limit this.

[0092] In some embodiments, in the above step S400, the implementation process of processing the training set and the test set according to the characteristics of multiple target gene pairs to obtain the processed training set and the processed test set may include the following steps S01-S03.

[0093] S01, obtaining a gene pair matrix of the training set according to the transcriptome data of each training sample in the training set.

[0094] It should be noted that the gene pair matrix of the training set is a second-order matrix. In the gene pair matrix of the training set, rows represent training set gene pairs, columns represent training samples, and the values ​​of the elements are the numerical values ​​of the training set gene pairs in the training samples. The numerical values ​​of the training set gene pairs in the training samples include any one of the first preset value, the second preset value or the third preset value.

[0095] In this step, first, at least one training set gene data is obtained according to the transcriptome data of each training sample in the training set. The training set gene data refers to the gene data existing in the training set, and the gene data is the gene type. Then, according to the at least one training set gene data and the transcriptome data of each training sample in the training set, the gene expression matrix of the training set is obtained. The gene expression matrix of the training set includes each training sample and its corresponding multiple training set gene data. Among them, the gene expression matrix of the training set is a second-order matrix. In the gene expression matrix of the training set, the row represents the training set gene data, the column represents the training sample, and the value of the element is the gene expression amount of the training set gene data in the training sample. The representation form of the gene expression matrix of the training set is the same as the gene expression matrix shown in the above formula (1), and the specific implementation of constructing the gene expression matrix of the training set is the same as the implementation method of the aforementioned step S320, which will not be repeated in this embodiment. Afterwards, according to the gene expression matrix of the training set, the gene pair matrix of the training set is obtained. The gene pair matrix of the training set includes each training sample and its corresponding multiple training set gene pairs. The implementation method of converting the gene expression matrix of the training set into the corresponding gene pair matrix is ​​the same as the implementation method of the aforementioned step S330, and the representation form of the gene pair matrix of the training set is the same as the gene pair matrix shown in the above formula (2), which will not be described in detail in this embodiment.

[0096] S02, obtaining a plurality of gene pair features corresponding to each training sample according to the plurality of target gene pair features and the gene pair matrix of the training set.

[0097] It should be noted that the multiple gene pair features corresponding to the training samples are used as features of the transcriptome data of the training samples.

[0098] In this step, first, for each training set gene pair, if the current training set gene pair is present in a set consisting of multiple target genes to features, it is indicated that the current training set gene is identical to a certain target gene to feature, and the current training set gene has model training significance, and now the current training set gene pair is retained, otherwise the current training set gene is different from any target gene to feature, and the current training set gene does not have model training significance, and now the current training set gene pair is screened out. By traversing multiple target genes to features and multiple training set gene pairs in the gene pair matrix of the training set, the gene pair matrix after screening of the training set can be obtained. Then, from the gene pair matrix after screening of the training set, extract multiple training set gene pairs corresponding to each training sample as multiple gene pair features corresponding to each training sample.

[0099] S03, obtaining a processed training set according to the multiple gene pair features and categories corresponding to each training sample.

[0100] In this step, for each training sample, multiple gene pair features of the training sample are used as sample features, and the category of the training sample is used as the sample label. Through the multiple gene pair features and categories corresponding to each training sample, a processed training set can be constructed, and the processed training set will be used to train the model.

[0101] It can be seen that this embodiment can accurately extract the gene pair features of the training set as the sample features of the training set, ensuring the data uniformity of different types of transcriptome data in the training set. In addition, multiple target gene pair features are used as the common processing benchmark for the training set and the test set. Since the processing benchmarks of the training set and the test set are the same, during the processing of the training set and the test set, it can be effectively ensured that the training set and the test set are in the same data distribution space, thereby effectively improving the processing efficiency across data sets and improving the sample accuracy across data sets.

[0102] In some embodiments, in the above step S400, the test set and the test set are processed according to the features of multiple target gene pairs to obtain the processed test set and the specific implementation of the processed test set may also include the following steps S11-S13.

[0103] S11, obtaining a gene pair matrix of the test set according to the transcriptome data of each test sample in the test set.

[0104] It should be noted that the gene pair matrix of the test set is a second-order matrix. In the gene pair matrix of the test set, the rows represent the test set gene pairs, the columns represent the test samples, the values ​​of the elements are the numerical values ​​of the test set gene pairs in the test samples, and the numerical values ​​of the test set gene pairs in the test samples include any one of the first preset value, the second preset value or the third preset value.

[0105] In this step, first, at least one test set gene data is obtained based on the transcriptome data of each test sample in the test set. The test set gene data refers to the gene data existing in the test set, and the gene data is the gene type. Then, based on the at least one test set gene data and the transcriptome data of each test sample in the test set, the gene expression matrix of the test set is obtained. The gene expression matrix of the test set includes each test sample and its corresponding multiple test set gene data. Among them, the gene expression matrix of the test set is a second-order matrix. In the gene expression matrix of the test set, the row represents the test set gene data, the column represents the test sample, and the value of the element is the gene expression amount of the test set gene data in the test sample. The representation form of the gene expression matrix of the test set is the same as the gene expression matrix shown in the above formula (1), and the specific implementation of constructing the gene expression matrix of the test set is the same as the implementation method of the aforementioned step S320, which will not be repeated in this embodiment. Afterwards, according to the gene expression matrix of the test set, the gene pair matrix of the test set is obtained. The gene pair matrix of the test set includes each test sample and its corresponding multiple test set gene pairs. The implementation method of converting the gene expression matrix of the test set into the corresponding gene pair matrix is ​​the same as the implementation method of the aforementioned step S330, and the representation form of the gene pair matrix of the test set is the same as the gene pair matrix shown in the above formula (2), which will not be described in detail in this embodiment.

[0106] S12, obtaining multiple gene pair features corresponding to each test sample according to the multiple target gene pair features and the gene pair matrix of the test set.

[0107] It should be noted that the multiple gene pair features corresponding to the test sample are used as features of the transcriptome data of the test sample.

[0108] In this step, first, for each test set gene pair, if the current test set gene pair exists in a set consisting of multiple target gene pair features, it means that the current test set gene is identical to a certain target gene pair feature, and the current test set gene has model testing significance, and the current test set gene pair is retained at this time, otherwise it means that the current test set gene is different from any target gene pair feature, and the current test set gene does not have model testing significance, and the current test set gene pair is screened out at this time. By traversing multiple target gene pair features and multiple test set gene pairs in the gene pair matrix of the test set, the gene pair matrix after screening of the test set can be obtained. Then, multiple test set gene pairs corresponding to each test sample are extracted from the gene pair matrix after screening of the test set as multiple gene pair features corresponding to each test sample.

[0109] S13, obtaining a processed test set according to the multiple gene pair features and categories corresponding to each test sample.

[0110] In this step, for each test sample, multiple gene pair features of the test sample are used as sample features, and the category of the test sample is used as the sample label. Through the multiple gene pair features and categories corresponding to each test sample, a processed test set can be constructed, and the processed test set will be used to test the model.

[0111] It can be seen that this embodiment can accurately extract the gene pair features of the test set as the sample features of the test set, ensuring the data uniformity of different types of transcriptome data in the test set. In addition, multiple target gene pair features are used as the common processing benchmark for the training set and the test set. Since the processing benchmarks of the training set and the test set are the same, during the processing of the training set and the test set, it can be effectively ensured that the training set and the test set are in the same data distribution space, thereby effectively improving the processing efficiency across data sets and improving the sample accuracy across data sets.

[0112] In some embodiments, reference Figure 2 In the above step S500, the model training and testing are performed using a machine learning method according to the processed training set and the processed test set, and the implementation process of obtaining the disease classification model may include the following steps S510-S530: S510, using a machine learning method to perform model training on the processed training set to obtain various machine learning models; S520, testing each machine learning model using the processed test set to obtain a test result of each machine learning model; S530, obtaining a disease classification model according to the test results of each machine learning model.

[0113] In this embodiment, first, a corresponding initial machine learning model is constructed according to each machine learning method. Then, for each initial machine learning model, the training set is input into the current initial machine learning model, and the gene pair features of the training sample are used as features, and the category of the training sample is used as the target to train the current initial machine learning model so that the current initial machine learning model can learn the mapping relationship between the gene pair features and the category, and obtain the machine learning model. Afterwards, each machine learning model is tested using a test set, so that the test results of each machine learning model can reflect the performance of the machine learning model. Finally, according to the test results of each machine learning model, the machine learning model with the best test results is selected as the disease classification model output. According to the actual experimental results of the embodiment of the present application, the performance of the random forest model is the best, so the disease classification model is a trained random forest model.

[0114] Optionally, during the model training process, in order to improve the accuracy of the model training set, after using a machine learning method to perform model training on the processed training set to obtain each machine learning model, it is determined whether the type of the discovery set is the same as the type of the training set; if so, it means that there are no sample features with a feature importance of zero in the processed training set, and then directly enter step S520 to perform model testing; if not, it means that there are sample features with a feature importance of zero in the processed training set, and at this time, for each machine learning model, the feature importance of each sample feature in the processed training set is obtained through the machine learning model, and the sample features with a feature importance of zero in the processed training set are screened out to obtain a new training set and re-perform model training on the new training set.

[0115] Optionally, during the model testing process, the indicators used to evaluate the performance of each machine learning model can be flexibly set according to the actual situation, for example, the indicator can be F1 score, accuracy, etc., but is not limited thereto. According to the actual experimental results of the embodiments of the present application, in the skin cancer classification task, the indicator is F1 score; in the skin cancer immunotherapy response prediction task, the indicator is accuracy.

[0116] Optionally, in the overall process of model training and testing, a cross-validation method such as a three-fold cross-validation method can be introduced to improve the training efficiency and training accuracy of the machine learning model. For example, the training set is divided into a training subset and a test subset for each training round according to the three-fold cross-validation method, and the test subset of each training round will be merged with the processed test set for model testing, but is not limited to this.

[0117] Secondly, the embodiment of the present application also provides a training device for a disease classification model, which may include: a first acquisition module, a first processing module, a second processing module, a third processing module and a fourth processing module. Among them, the first acquisition module is used to acquire target disease gene data and a transcriptome data set associated with the target disease; wherein the transcriptome data set includes transcriptome data and categories of several initial samples; the first processing module is used to classify and process the transcriptome data set to obtain a discovery set, a training set and a test set; the second processing module is used to obtain multiple target gene pair features associated with the target disease based on the discovery set and the target disease gene data; the third processing module is used to process the training set and the test set based on the multiple target gene pair features to obtain a processed training set and a processed test set; the fourth processing module is used to use a machine learning method to perform model training and testing based on the processed training set and the processed test set to obtain a disease classification model, and the disease classification model is used to identify the category of samples associated with the target disease.

[0118] The contents of the above method embodiments are all applicable to the present device embodiments. The functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0119] Finally, an embodiment of the present application provides a disease prediction system, which includes: a second acquisition module and a fifth processing module. The second acquisition module is used to obtain the transcriptome data of the sample to be tested associated with the target disease. It should be stated that the sample to be tested is a sample pre-stored in a database, rather than a sample directly obtained from a living body. The fifth processing module is used to classify the transcriptome data of the sample to be tested according to the disease classification model to obtain the category of the sample to be tested; wherein the disease classification model is trained by the above-mentioned training method of a disease classification model.

[0120] In some embodiments, the fifth processing module may include a preprocessing unit and a prediction unit, the prediction unit is configured with a disease classification model, the preprocessing unit is used to obtain at least one gene pair feature of the sample to be tested according to the transcriptome data of the sample to be tested, and the prediction unit is used to classify the at least one gene pair feature of the sample to be tested according to the disease classification model to obtain the category of the sample to be tested. When the preprocessing unit is used to obtain at least one gene pair feature of the sample to be tested according to the transcriptome data of the sample to be tested, it is specifically used to perform the following operations: first, according to the transcriptome data of the sample to be tested, at least one gene data to be tested is obtained, and the gene data to be tested is the gene type, and according to the at least one gene data to be tested and the transcriptome data of the sample to be tested, the gene expression amount of each gene data to be tested in the sample to be tested is determined, and then the gene expression matrix of the sample to be tested is constructed with the gene data to be tested as the row, the sample to be tested as the column, and the gene expression amount of each gene data to be tested in the sample to be tested as the value of each element. Then, according to the gene expression matrix of the sample to be tested, the gene pair matrix of the sample to be tested is obtained. The specific implementation of converting the gene expression matrix into the gene pair matrix is ​​the same as the aforementioned step S330, which will not be described in detail. In the gene pair matrix of the sample to be tested, the rows represent the gene pair features to be tested, that is, the gene pair features of the sample to be tested, the columns represent the sample to be tested, and the value of each element is the numerical value of each gene pair feature to be tested in the sample to be tested. The representation form of the gene pair matrix of the sample to be tested is the same as the above formula (2), which will not be described in detail. Finally, at least one gene pair feature to be tested of the sample to be tested can be extracted through the gene pair matrix of the sample to be tested.

[0121] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0122] The following is a principle description of the training method, device and disease classification system of the disease classification model in the embodiment of the present application using a specific application scenario. In this application scenario, the disease classification model is used to identify the category of the sample to be tested for skin cancer.

[0123] T01, obtain the highly variable genes or related genes of skin cancer as skin cancer gene data through the preset database, and construct the transcriptome data set of skin cancer at the same time. Specifically, first, obtain the initial transcriptome data and categories of several skin cancer samples. Among them, the initial transcriptome data of skin cancer samples include the transcriptome sequencing expression matrix, single-cell transcriptome sequencing expression matrix and original microarray data of skin cancer samples, and the categories of skin cancer samples include melanoma, basal cell carcinoma, Merkel cell carcinoma, squamous cell carcinoma and non-skin cancer (ie, healthy). Then, for the initial transcriptome data of each skin cancer sample, the robust multi-array average method is used to standardize the initial microarray data to obtain microarray data, and the million transcripts method is used to standardize the transcriptome sequencing expression matrix to obtain transcriptome sequencing data, and the million transcripts method is used to standardize the single-cell transcriptome sequencing expression matrix to obtain single-cell transcriptome sequencing data. Finally, the transcriptome data set of skin cancer is constructed through the transcriptome data and categories of each skin cancer sample. Among them, the transcriptome data of skin cancer samples include microarray data, transcriptome sequencing data and single-cell transcriptome sequencing data of skin cancer samples.

[0124] T02, divide the transcriptome data set of skin cancer into three data sets: a discovery set, a training set and a test set according to a preset division ratio. Specifically, the discovery set may include transcriptome data and categories of multiple discovery samples. Among them, the discovery sample refers to the initial sample divided into the discovery set, and the transcriptome data of the discovery sample may include any one of the single-cell transcriptome sequencing data of the discovery sample or the batch transcriptome data of the discovery sample, and the batch transcriptome data of the discovery sample may include the microarray data of the discovery sample and the transcriptome sequencing data of the discovery sample. The training set may include transcriptome data and categories of multiple training samples. Among them, the training sample refers to the initial sample divided into the training set, and the transcriptome data of the training sample may include the microarray data of the training sample and the transcriptome sequencing data of the training sample. In addition, the test set may include transcriptome data and categories of multiple test samples. Among them, the test sample refers to the initial sample divided into the test set, and the transcriptome data of the test sample may include at least one of the microarray data of the test sample, the transcriptome sequencing data of the test sample or the single-cell transcriptome sequencing data of the test sample.

[0125] T03, construct the gene pair matrix of the discovery set. Specifically, first, the genes in the intersection of the discovery set, the training set and the skin cancer gene data are taken as the target gene data. Then, based on the multiple target gene data and the transcriptome data of each discovery sample in the discovery set, the gene expression matrix of the discovery set is obtained. The gene expression matrix of the discovery set contains each discovery sample and its corresponding multiple target gene data, such as Figure 2 As shown in part (b) of the figure. The gene expression matrix of the discovery set is a second-order matrix. In the gene expression matrix of the discovery set, the rows represent the target gene data, the columns represent the discovery samples, and the element values ​​are the gene expression values ​​of the target gene data in the discovery samples. Then, the gene expression matrix of the discovery set is converted into the gene pair matrix of the discovery set by the above formula (3). The gene pair matrix of the discovery set includes each discovery sample and its corresponding multiple discovery set gene pairs, as shown in Figure 2 As shown in part (c) of the figure. The gene pair matrix of the discovery set is a second-order matrix. In the gene pair matrix of the discovery set, the rows represent the gene pairs of the discovery set, which are composed of two target gene data. The columns represent the discovery samples, and the element values ​​are the values ​​of the gene pairs of the discovery set in the discovery samples. The values ​​of the gene pairs of the discovery set in the discovery samples are 1, -1, or 0.

[0126] T04, extracting target gene pair features of multiple skin cancers. Specifically, first, for each discovery set gene pair, there are: for each category, according to the value of each element in the discovery set gene pair matrix, the total number of discovery set gene pairs with a value of 1 in the discovery samples belonging to the current category is obtained as the first target number of the current category relative to the discovery set gene pair, and the total number of discovery set gene pairs with a value of -1 in the discovery samples belonging to the current category is obtained as the second target number of the current category relative to the discovery set gene pair; by traversing the categories of all discovery samples, the first target number and the second target number of all categories relative to the discovery set gene pairs can be obtained as the category quantity features of the discovery set gene pairs. And by traversing all discovery set gene pairs, the category quantity features of all discovery set gene pairs can be obtained. Then, according to the category quantity features of each discovery set gene pair, combined with the above formula (5) or the above formula (6), the significance level value of each discovery set gene pair is calculated. Afterwards, all discovery set gene pairs are sorted from small to large according to the size of the significance level value, and the top 50 or 100 discovery set gene pairs are selected as the target gene pair features.

[0127] T05, update the training set and the test set. Specifically, for the training set, first, according to the transcriptome data of each training sample in the training set, the gene expression matrix of the training set is constructed, and then the gene expression matrix of the training set is converted into the gene pair matrix of the training set by the above formula (3). Among them, the gene pair matrix of the training set is a second-order matrix. In the gene pair matrix of the training set, the rows represent the training set gene pairs, the columns represent the training samples, the values ​​of the elements are the values ​​of the training set gene pairs in the training samples, and the values ​​of the training set gene pairs in the training samples are 1, -1 or 0. Then, for each training set gene pair in the gene pair matrix of the training set, if the current training set gene pair exists in the set composed of multiple target gene pair features, the current training set gene pair is retained, otherwise the current training set gene pair is filtered out. By traversing multiple target gene pair features and multiple training set gene pairs in the gene pair matrix of the training set, the filtered gene pair matrix of the training set can be obtained. Finally, multiple training set gene pairs corresponding to each training sample are extracted from the filtered gene pair matrix of the training set as multiple gene pair features corresponding to each training sample. For the test set, the processing steps are the same as those for the training set, and will not be repeated here.

[0128] T06, training and testing of disease classification models. Specifically, first, a corresponding initial machine learning model is constructed according to each machine learning method. Then, for each initial machine learning model, the training set is input into the current initial machine learning model to train the current initial machine learning model and obtain a machine learning model. Afterwards, determine whether the type of the discovery set is the same as the type of the training set; if so, use the test set to test each machine learning model to obtain the test results of each machine learning model, and select the machine learning model with the best test results as the disease classification model output; if not, for each machine learning model, use the machine learning model to obtain the feature importance of each sample feature in the processed training set, screen out the sample features with zero feature importance in the processed training set, obtain a new training set and retrain the model on the new training set to improve the training accuracy. In addition, a three-fold cross-validation method can be introduced to implement model training and testing.

[0129] T07, real-time classification of skin cancer. Specifically, first, the transcriptome data of the sample to be tested of skin cancer is obtained through a preset database. Then, based on the transcriptome data of the sample to be tested, the gene pair features to be tested of the sample to be tested are obtained. Finally, the gene pair features to be tested are input into the disease classification model, and the category of the sample to be tested is obtained through the disease classification model.

[0130] The following two embodiments are used to verify the technical effects provided by the embodiments of the present application.

[0131] Example 1: Skin cancer classification task. In the skin cancer classification task, the categories of skin cancer are melanoma (MEL), basal cell carcinoma (BCC), Merkel cell carcinoma (MCC), squamous cell carcinoma (SCC) and healthy (Healthy). The transcriptome profiles of 330 samples of the melanoma category, 44 samples of the basal cell carcinoma category, 68 samples of the Merkel cell carcinoma category, 88 samples of the squamous cell carcinoma category and 320 samples of the healthy category are collected by DNA microarray technology, a total of 850 samples and their transcriptome profiles, to construct a skin cancer data set, as shown in Table 1 below. Among them, the samples of the healthy category include samples of the healthy skin category and samples of the healthy mole category. The samples of the melanoma category include samples of the primary melanoma category and samples of the metastatic melanoma category. The gene chips used for DNA microarrays include Affymetrix and Illumina.

[0132] Table 1: Skin cancer dataset

[0133]

[0134] Three types of classification tasks are defined, including two-classification tasks (i.e., 2class, identifying whether the sample is a healthy sample or a non-healthy sample), three-classification tasks (i.e., 3class, identifying whether the sample is a healthy sample, a melanoma sample (MEL), and a non-healthy sample and a non-melanoma sample (Non-MEL)), and five-classification tasks (i.e., 5class, identifying whether the sample specifically belongs to health, BCC, SCC, MCC, and MEL). For each classification task, the embodiment of the present application finds 30 pairs of target gene pair features (AIGPS-30) from the skin cancer dataset and trains and tests the model to obtain the skin cancer classification model for each classification task. The skin classification models corresponding to these three classification tasks are compared with the existing benchmark methods (galvez degs). The evaluation indicators of the models include weighted F1 score (F1-weighted), macro F1 score (F1-macro), weighted precision (precision-weighted), macro precision (precision-macro), weighted recall (Recall-weighted), macro recall (Recall-macro) and accuracy (acc). The comparison results are as follows Figure 3As shown, in the two-classification task and the three-classification task, all indicators of the skin cancer classification model of the present application are higher than the existing benchmark method, and in the five-classification task, the macro precision rate indicator of the skin cancer classification model of the present application is slightly lower than the existing benchmark method, and other indicators except the macro precision rate are higher than the existing benchmark method. It can be seen that compared with the existing benchmark method, the embodiment of the present application can significantly improve the performance of the skin cancer classification model.

[0135] Example 2: Melanoma Immunotherapy Response Prediction Task In the melanoma immunotherapy response prediction task, a total of 252 skin cancer samples were extracted from the single-cell transcriptome dataset GSE12057552 of melanoma (which includes 5928 pre-treated cells) and 6 RNA-seq datasets of melanoma, Riaz, Van, Hugo, Gide, Lee and MGH, to construct the melanoma dataset, as shown in Table 2 below. The melanoma single-cell cohort can be found in the Gene Expression Omnibus (GEO) database with a sequence ID of GSE120575. The Riaz dataset can be found in the GEO database with a sequence ID of GSE91061. The Hugo dataset can be found in the GEO database with a sequence ID of GSE78220. The MGH dataset can be found in the GEO database with sequence IDs of GSE115821 and GSE168204. The Gide cohort can be found in the BioProject database with an accession number of PRJEB23709. The Lee dataset is available in the European Genome-phenome Archive (EGA) with the dataset accession number EGAD00001005738. The Van dataset is available in the database of Genotypes and Phenotypes (dbGaP) with the accession number phs000452.v2.p1.

[0136] Table 2: Melanoma dataset

[0137] The melanoma immunotherapy response prediction task is mainly a binary classification task, that is, to identify whether a melanoma sample responds to immunotherapy. For this prediction task, the present embodiment found 27 pairs of target gene pair features (AIGPS-27) and compared AIGPS-27 with existing transcriptome-based prediction markers, including PASS_PRE, IMPRES, , T_cell_inflamed, MHC-I, MHC-II, Immune Score, NRS, Chemokine and CYT. The results of the comparison are as follows: Figure 4 As shown, in Hugo, Lee, Gide, MGH and all data sets (Overall), the average AUC values ​​of the embodiments of the present application are 0.59, 0.65, 0.70, 0.75 and 0.68, respectively, which shows that the overall performance of AIGPS-27 is better than other markers, indicating the versatility of AIGPS-27 across data sets. Among them, the AUC value refers to the area under the receiver operating characteristic curve (Receiver Operating Characteristic Curve, ROC).

[0138] In summary, it can be seen that the embodiments of the present application can improve the training accuracy and training efficiency of the disease classification model, thereby improving the interpretability and classification accuracy of the disease classification model.

[0139] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

[0140] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the described embodiments. Technical personnel familiar with the field may make various equivalent modifications or substitutions without violating the spirit of the present application. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A method for training a disease classification model, characterized in that: The steps include: Obtaining target disease gene data and a transcriptome data set associated with the target disease; wherein the transcriptome data set includes transcriptome data and categories of several initial samples; Classifying the transcriptome data set to obtain a discovery set, a training set, and a test set; According to the discovery set and the target disease gene data, a plurality of target gene pair features associated with the target disease are obtained; Processing the training set and the test set according to the plurality of target gene pair features to obtain the processed training set and the processed test set; Model training and testing are performed using a machine learning method based on the processed training set and the processed test set to obtain a disease classification model, where the disease classification model is used to identify categories of samples associated with the target disease.

2. The method for training a disease classification model according to claim 1, characterized in that: The step of obtaining target disease gene data and a transcriptome data set associated with the target disease includes: Acquiring a gene type associated with the target disease as the target disease gene data; Acquire initial transcriptome data and categories of several initial samples associated with the target disease, wherein the initial transcriptome data of the initial samples include at least one of a transcriptome sequencing expression matrix of the initial samples, a single-cell transcriptome sequencing expression matrix of the initial samples, or initial microarray data of the initial samples; Standardizing the initial transcriptome data of each of the initial samples to obtain the transcriptome data of each of the initial samples; wherein the transcriptome data of the initial samples includes at least one of the microarray data of the initial samples, the transcriptome sequencing data of the initial samples, or the single-cell transcriptome sequencing data of the initial samples; According to the transcriptome data and categories of each of the initial samples, a transcriptome data set associated with the target disease is obtained.

3. The method for training a disease classification model according to claim 1, characterized in that: The step of obtaining a plurality of target gene pair features associated with the target disease based on the discovery set and the target disease gene data includes: Obtaining a plurality of target gene data associated with the target disease according to the transcriptome data of each discovery sample in the discovery set, the transcriptome data of each training sample in the training set, and the target disease gene data; According to the plurality of target gene data and the transcriptome data of each discovery sample in the discovery set, a gene expression matrix of the discovery set is obtained; wherein the gene expression matrix of the discovery set is a second-order matrix, in which the rows represent the target gene data, the columns represent the discovery samples, and the values ​​of the elements are the gene expression amounts of the target gene data in the discovery samples; According to the gene expression matrix of the discovery set, a gene pair matrix of the discovery set is obtained; wherein the gene pair matrix of the discovery set is a second-order matrix, in which rows represent discovery set gene pairs, columns represent discovery samples, and element values ​​are numerical values ​​of the discovery set gene pairs in the discovery samples; the discovery set gene pairs include two target gene data, and the numerical values ​​of the discovery set gene pairs in the discovery samples include any one of a first preset value, a second preset value, or a third preset value; Screening is performed on each of the discovery set gene pairs in the discovery set gene pair matrix to obtain a plurality of target gene pair features associated with the target disease.

4. The method for training a disease classification model according to claim 3, characterized in that: The step of obtaining a gene pair matrix of the discovery set according to the gene expression matrix of the discovery set comprises: Determining the number of combinations of two target gene data taken from the plurality of target gene data as the total number of gene pairs in the discovery set; Determining a plurality of discovery set gene pairs according to the total number of the discovery set gene pairs and the gene expression matrix of the discovery set; According to the gene expression matrix of the discovery set, the variance of the gene expression amount of each target gene data in all the discovery samples is obtained as the expression amount variance of each target gene data; According to the gene expression matrix of the discovery set, the expression variance of each target gene data and a preset adaptive coefficient, the value of each discovery set gene pair in each discovery sample is obtained; The gene pair matrix of the discovery set is constructed by using the gene pairs of the discovery set as the rows of the gene pair matrix of the discovery set, using the discovery samples as the columns of the gene pair matrix of the discovery set, and using the numerical values ​​of each gene pair of the discovery set in each discovery sample as the values ​​of each element in the gene pair matrix of the discovery set.

5. The method for training a disease classification model according to claim 3, characterized in that: The screening process is performed on each of the discovery set gene pairs in the discovery set gene pair matrix to obtain a plurality of target gene pair features associated with the target disease, including: For each of the discovery set gene pairs, according to the values ​​of the discovery set gene pairs in each of the discovery samples and the categories of all the discovery samples, a first target quantity and a second target quantity of each category relative to the discovery set gene pair are obtained as category quantity characteristics of the discovery set genes; wherein the first target quantity of the category relative to the discovery set gene pair is the total number of the discovery set gene pairs whose values ​​are the first preset value in the discovery samples belonging to the category, and the second target quantity of the category relative to the discovery set gene pairs is the total number of the discovery set gene pairs whose values ​​are the second preset value in the discovery samples belonging to the category; Obtaining a significance level value of each discovery set gene pair according to the category quantity characteristics of each discovery set gene pair and the total category quantity of the discovery samples; Each of the discovery set gene pairs in the discovery set gene pair matrix is ​​screened according to the significance level value of each of the discovery set gene pairs to obtain a plurality of target gene pair features associated with the target disease.

6. The method for training a disease classification model according to claim 1, characterized in that: The step of processing the training set and the test set according to the plurality of target gene pair features to obtain the processed training set and the processed test set comprises: According to the transcriptome data of each training sample in the training set, a gene pair matrix of the training set is obtained; wherein the gene pair matrix of the training set is a second-order matrix, in which rows represent training set gene pairs, columns represent training samples, and element values ​​are values ​​of the training set gene pairs in the training samples, and the values ​​of the training set gene pairs in the training samples include any one of a first preset value, a second preset value, or a third preset value; According to the plurality of target gene pair features and the gene pair matrix of the training set, a plurality of gene pair features corresponding to each of the training samples are obtained; The processed training set is obtained according to the multiple gene pair features and categories corresponding to each of the training samples.

7. The method for training a disease classification model according to claim 1, characterized in that: The step of processing the training set and the test set according to the plurality of target gene pair features to obtain the processed training set and the processed test set comprises: According to the transcriptome data of each test sample in the test set, a gene pair matrix of the test set is obtained; wherein the gene pair matrix of the test set is a second-order matrix, in which rows represent test set gene pairs, columns represent test samples, and element values ​​are numerical values ​​of the test set gene pairs in the test samples, and the numerical values ​​of the test set gene pairs in the test samples include any one of a first preset value, a second preset value, or a third preset value; According to the plurality of target gene pair features and the gene pair matrix of the test set, a plurality of gene pair features corresponding to each of the test samples are obtained; The processed test set is obtained according to the multiple gene pair features and categories corresponding to each of the test samples.

8. The method for training a disease classification model according to claim 1, characterized in that: The method of using a machine learning method to perform model training and testing based on the processed training set and the processed test set to obtain a disease classification model includes: Using a machine learning method to perform model training on the processed training set to obtain various machine learning models; Using the processed test set to test each of the machine learning models, to obtain a test result of each of the machine learning models; According to the test results of each of the machine learning models, the disease classification model is obtained.

9. A training device for a disease classification model, characterized in that: include: A first acquisition module is used to acquire target disease gene data and a transcriptome data set associated with the target disease; wherein the transcriptome data set includes transcriptome data and categories of several initial samples; A first processing module is used to classify the transcriptome data set to obtain a discovery set, a training set and a test set; A second processing module, configured to obtain a plurality of target gene pair features associated with the target disease based on the discovery set and the target disease gene data; A third processing module is used to process the training set and the test set according to the characteristics of the plurality of target gene pairs to obtain the processed training set and the processed test set; The fourth processing module is used to perform model training and testing using a machine learning method based on the processed training set and the processed test set to obtain a disease classification model, wherein the disease classification model is used to identify categories of samples associated with the target disease.

10. A disease classification system, characterized in that: include: A second acquisition module is used to acquire transcriptome data of a sample to be tested associated with a target disease; The fifth processing module is used to classify the transcriptome data of the sample to be tested according to the disease classification model to obtain the category of the sample to be tested; wherein the disease classification model is trained by a disease classification model training method as described in any one of claims 1-8.

Images