A method and apparatus for analyzing key characteristic proteins of lipoprotein disorders based on deep learning-based multi-target protein expression analysis.

Abstract

This invention relates to the field of bioinformatics and discloses a method and apparatus for analyzing important characteristic proteins of lipoprotein disorders based on deep learning and multi-target protein expression. The method involves obtaining the multi-round training weights of the first hidden layer in a lipoprotein disorder analysis model corresponding to each protein feature during multiple training rounds; determining the importance feature score for each protein feature based on the multi-round training weights; and identifying at least one important characteristic protein corresponding to lipoprotein disorders based on the importance feature score. By dynamically tracking the weight evolution of the first hidden layer in the lipoprotein disorder analysis model during multiple training rounds, the importance of protein features is precisely quantified. Important characteristic proteins highly correlated with lipoprotein disorders are screened based on the importance feature score, providing important characteristic proteins with strong biological interpretability for lipoprotein disorders.

Description

Technical Field

[0001] This invention relates to the field of bioinformatics technology, and in particular to a method and apparatus for analyzing the expression of multi-target proteins based on deep learning to identify important characteristic proteins of lipoprotein disorders. Background Technology

[0002] In the field of bioinformatics, protein characterization data, due to its comprehensive ability to reflect the expression levels and functional states of thousands of proteins in an organism, has become an important data source for disease mechanism research and biomarker discovery. Compared to genomic or transcriptomic data, proteins are the direct executors of cellular functions, and their expression profile changes are closer to the physiological and pathological processes of diseases. Especially in lipoprotein disorders (such as hyperlipidemia and atherosclerosis), the abnormal expression of specific proteins often directly participates in key pathological processes such as dyslipidemia, chronic inflammation, endothelial dysfunction, and oxidative stress. Therefore, correlation analysis based on protein expression profiles and disease phenotypes not only helps to reveal the molecular mechanisms of diseases but also provides potential protein biomarkers with high biological interpretability for disease analysis. However, despite the rich biological information contained in protein characterization data, accurately and robustly identifying protein biomarkers that are truly relevant to disease specificity remains a core challenge. Summary of the Invention

[0003] The purpose of this invention is to provide at least one method for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning. This method can at least solve the technical problem of how to accurately and robustly identify protein biomarkers that are truly related to disease specificity, and can at least achieve accurate and robust identification of protein biomarkers related to lipoprotein disorders.

[0004] To address the aforementioned technical problems, at least one embodiment of this application provides a method for analyzing important feature proteins of lipoprotein disorders based on deep learning multi-target protein expression, comprising: obtaining the multi-round training weights of a first hidden layer in a lipoprotein disorder analysis model corresponding to each protein feature during multi-round training; determining an important feature score corresponding to each protein feature based on the multi-round training weights; and determining at least one important feature protein corresponding to lipoprotein disorder based on the important feature score.

[0005] This approach achieves refined quantification of protein feature importance by dynamically tracking the weight evolution of the first hidden layer in a lipoprotein disorder analysis model during multiple training rounds. Based on the acquired training weights, an importance score for each protein feature is calculated, which not only enhances the model's interpretability but also effectively distinguishes noisy features from truly disease-related biological signals. Finally, important feature proteins highly correlated with lipoprotein disorders are selected based on the importance score, providing potential biomarkers (i.e., important feature proteins) with strong biological interpretability for lipoprotein disorders.

[0006] In some examples, determining at least one important feature protein corresponding to a lipoprotein disorder based on the important feature score includes: obtaining at least one reference target protein corresponding to the lipoprotein disorder; obtaining the important feature score corresponding to each reference target protein respectively, and using the reference target protein that meets the preset conditions as the important feature protein.

[0007] In some examples, the step of using the reference target protein that meets the preset conditions as the important feature protein includes: using at least one reference target protein whose important feature score is greater than or equal to the important feature score threshold as the important feature protein, or using a preset number of reference target proteins with the largest important feature score as the important feature protein.

[0008] In some examples, determining the important feature score corresponding to each protein feature based on the multi-round training weights includes: for each protein feature, obtaining the average weight corresponding to the protein feature based on the multi-round training weights; and using the average weight as the important feature score corresponding to the protein feature.

[0009] In some examples, obtaining the average weight corresponding to the protein feature based on the multi-round training weights includes: obtaining the round weight value corresponding to the multi-round training weights; and determining the average weight corresponding to the protein feature based on the round weight value, the multi-round training weights, and the total number of training rounds.

[0010] In some cases, the round weight value is positively correlated with the number of rounds.

[0011] In some examples, obtaining the round weight values ​​corresponding to the multi-round training weights includes: obtaining the total number of training rounds of the lipoprotein barrier analysis model; obtaining the training progress corresponding to each training round; and determining the round weight value corresponding to each training round based on the training progress and the initial round weight value.

[0012] At least one embodiment of this application also provides an apparatus for analyzing important feature proteins of lipoprotein disorders based on deep learning multi-target protein expression, comprising: an acquisition unit for acquiring multi-round training weights of a first hidden layer in a lipoprotein disorder analysis model corresponding to each protein feature during multi-round training; a determination unit for determining an important feature score corresponding to each protein feature based on the multi-round training weights; the determination unit is further configured to determine at least one important feature protein corresponding to lipoprotein disorder based on the important feature score.

[0013] At least one embodiment of this application also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the above-described method for analyzing important characteristic proteins of lipoprotein barriers based on deep learning multi-target protein expression.

[0014] At least one embodiment of this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for analyzing important characteristic proteins of lipoprotein disorders based on deep learning multi-target protein expression. Attached Figure Description

[0015] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.

[0016] Figure 1 This is a flowchart illustrating a method for analyzing key characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning, according to one embodiment of this application. Figure 2 This is a schematic diagram illustrating the relationship between epoch and loss value, provided in one embodiment of this application; Figure 3 This is a schematic diagram of the structure of a device for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression analysis using deep learning, provided in another embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in another embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.

[0018] It should be noted that the acquisition or use of data in the embodiments of this application requires the user's consent. The relevant data can only be obtained after the user's authorization, and the acquisition or use of the data complies with the provisions of relevant laws and regulations.

[0019] To facilitate understanding of the embodiments of this application, the relevant content of the method for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning will be introduced first.

[0020] In the field of bioinformatics, protein characterization data, due to its comprehensive ability to reflect the expression levels and functional states of thousands of proteins in an organism, has become an important data source for disease mechanism research and biomarker discovery. Compared to genomic or transcriptomic data, proteins are the direct executors of cellular functions, and their expression profile changes are closer to the physiological and pathological processes of diseases. Especially in lipoprotein disorders (such as hyperlipidemia and atherosclerosis), the abnormal expression of specific proteins often directly participates in key pathological processes such as dyslipidemia, chronic inflammation, endothelial dysfunction, and oxidative stress. Therefore, correlation analysis based on protein expression profiles and disease phenotypes not only helps to reveal the molecular mechanisms of diseases but also provides potential protein biomarkers with high biological interpretability for disease analysis. However, despite the rich biological information contained in protein characterization data, accurately and robustly identifying protein biomarkers that are truly relevant to disease specificity remains a core challenge.

[0021] To address the aforementioned technical challenge of accurately and robustly identifying protein biomarkers truly relevant to disease specificity, this invention proposes a method for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning. The implementation details of this method are described below. These details are provided for ease of understanding and are not essential for implementing this solution.

[0022] Example 1: The method for analyzing key characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning in this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. Its specific process can be as follows: Figure 1 As shown, it includes: Step 110: Obtain the multi-round training weights of the first hidden layer in the lipoprotein barrier analysis model corresponding to each protein feature during the multi-round training process.

[0023] The lipoprotein disorder analysis model is used to identify key characteristic proteins from a large number of protein features, or in other words, to identify proteins that are related to or unrelated to lipoprotein disorders. The lipoprotein disorder analysis model uses a multi-layer neural network architecture called ProteinExpressionNet. ProteinExpressionNet contains three hidden layers: the first hidden layer has 128 nodes, the second has 64 nodes, and the third has 32 nodes. The outputs of each hidden layer are standardized, ReLU activated, and processed with a dropout rate of 0.3. Furthermore, during multiple rounds of training of the lipoprotein disorder analysis model (i.e., multiple training cycles), protein feature data from 53,013 subjects to be analyzed are used as sample data. These subjects can be humans, excised tissues, or deceased human or animal bodies. Each set of protein feature data includes 2,923 protein features. The 53,013 sample data points included some data related to lipoprotein disorders and some unrelated data. To balance the number of data points related to and unrelated to lipoprotein disorders, an oversampling algorithm was used to balance the data, ensuring a balanced distribution. Finally, 80% of the balanced data was allocated to the training set, and 20% to the test set. Oversampling involves augmenting minority class proteome data to a number comparable to non-minority class proteome data. This mitigates class imbalance and prevents the trained lipoprotein disorder analysis model from exhibiting prediction bias and low accuracy due to excessively large sample size discrepancies.

[0024] In some examples, the sample data can be protein expression data loaded from a CSV (Comma-Separated Values) file. For this sample data, missing value imputation, data cleaning, and data standardization are performed to ensure data quality, avoid interference from noise, improve model robustness, and ensure data consistency. Specifically, missing values ​​in the sample data can be imputed using the median.

[0025] In some examples, when training the lipoprotein barrier analysis model, the training patience value is set to 5, the Adam optimizer is used for optimization, and the learning rate of the Adam optimizer is set to 0.001. Specifically, the learning rate of the Adam optimizer can also be dynamically adjusted based on the ReduceLROnPlateau scheduler.

[0026] Among them, the lipoprotein barrier analysis model is used for binary classification to identify proteins that are related to or not related to lipoprotein barriers. Its goal is to optimize the class boundary, so the model usually converges quickly. Setting a small patience value of 5 is suitable for tasks with fast convergence, preventing overfitting to synthetic samples and reducing unnecessary training time.

[0027] Therefore, based on the balanced sample data, when training the lipoprotein barrier analysis model constructed by deep learning, a patience value of 5 is set for the early stopping mechanism. That is, if the validation loss does not improve within 5 consecutive training cycles, training is automatically terminated, and the model weights with the best validation performance are retained. This strategy effectively prevents model overfitting and improves its stability and generalization ability in the task of predicting important characteristic proteins of lipoprotein barriers.

[0028] Specifically, when training the lipoprotein barrier analysis model, the training batch size was set to 16 and the maximum epoch (training rounds) was set to 200.

[0029] For example, see Figure 2 When the training epoch is between 15 and 25, the training loss and validation loss of the lipoprotein barrier analysis model change relatively slowly. At this point, the analysis model has basically converged. Therefore, setting the maximum number of training epochs to 200 provides enough training epochs for the analysis model to converge.

[0030] Therefore, the batch size is set to 16 to reduce memory usage while ensuring gradient estimation stability and adapting to high-dimensional proteomics data; the maximum number of training rounds is set to 200 to provide ample learning opportunities for the lipoprotein barrier analysis model, and dynamic termination is achieved by combining an early stopping mechanism.

[0031] Step 120: Based on the weights from multiple rounds of training, determine the important feature score corresponding to each protein feature.

[0032] The multi-round training weights are used to evaluate the importance of each protein feature in training the lipoprotein barrier analysis model. A larger multi-round training weight indicates greater importance of the corresponding protein feature in training the lipoprotein barrier analysis model. The importance feature score determines the degree of importance of a protein feature in training the lipoprotein barrier analysis model; a larger importance feature score indicates greater importance of the corresponding protein feature in training the lipoprotein barrier analysis model.

[0033] Specifically, based on the weights from multiple rounds of training, the important feature score corresponding to each protein feature is determined, including: for each protein feature, obtaining the average weight corresponding to the protein feature based on the weights from multiple rounds of training; and using the average weight as the important feature score corresponding to the protein feature.

[0034] Furthermore, the average weight of the protein feature is obtained based on the weights of multiple rounds of training, including: obtaining the round weight values ​​corresponding to the weights of multiple rounds of training; and determining the average weight of the protein feature based on the round weight values, the weights of multiple rounds of training, and the total number of training rounds.

[0035] Among them, the round weight value refers to the round importance of the protein feature in each round of training.

[0036] Specifically, the weight value of each round is positively correlated with the number of rounds. The more rounds there are, the larger the corresponding weight value of each round, in order to reduce the influence of non-important feature proteins in the early training process and improve the accuracy of the model.

[0037] For example, the weight values ​​of each round are added together with the weight values ​​of multiple rounds of training to obtain the weight values ​​of each round. The absolute values ​​of the weight values ​​of each round are added together to obtain the total weight value. The difference between the total weight value and the total number of training rounds is used as the average weight value.

[0038] Furthermore, the round weight values ​​corresponding to the training weights in multiple rounds are obtained, including: obtaining the total number of training rounds of the lipoprotein barrier analysis model; obtaining the training progress corresponding to each training round; and determining the round weight value corresponding to each training round based on the training progress and the initial round weight value.

[0039] The initial round weight value refers to the weight value corresponding to the protein feature during the first round of training.

[0040] For example, for each training round, the product of the training progress and the initial round weight value can be used as the round weight value of the protein feature in that training round, or the sum of the training progress and the initial round weight value can be used as the round weight value of the protein feature in that training round.

[0041] Step 130: Based on the important feature score, identify at least one important feature protein corresponding to the lipoprotein disorder.

[0042] Among them, important characteristic proteins refer to protein features that are related to lipoprotein disorders and have a high correlation. Based on important characteristic proteins, it is possible to better reflect whether the subject under analysis has lipoprotein disorders.

[0043] Specifically, in step 130 above, determining at least one important feature protein corresponding to lipoprotein disorder based on the important feature score includes: obtaining at least one reference target protein corresponding to lipoprotein disorder; obtaining the important feature score corresponding to each reference target protein respectively, and taking the reference target protein that meets the preset conditions as the important feature protein.

[0044] Reference target proteins are proteins associated with lipoprotein disorders. Reference target proteins can be important characteristic proteins.

[0045] Furthermore, the reference target proteins that meet the preset conditions are designated as important feature proteins, including: designating at least one reference target protein with an important feature score greater than or equal to an important feature score threshold as an important feature protein, or designating a preset number of reference target proteins with the largest important feature scores as important feature proteins.

[0046] Specifically, selecting the reference target proteins with the highest important feature scores as important feature proteins means selecting the reference target proteins corresponding to the highest-ranked, predetermined number of important feature scores as important feature proteins, in descending order of importance feature scores.

[0047] For example, as shown in Table 1, ten key characteristic proteins were identified based on the analytical model.

[0048] Table 1. Ten Key Characteristic Proteins

[0049]

[0050] In some examples, this also includes: reliability validation of trained lipoprotein barrier analysis models.

[0051] Specifically, the reliability of the trained lipoprotein barrier analysis model is verified, including: calculating at least one evaluation index of the lipoprotein barrier analysis model; if at least one evaluation index meets the reliability verification conditions, the analysis model is determined to meet the reliability verification.

[0052] Specifically, the evaluation metrics can be computational accuracy and AUC (Area Under the ROC Curve). If the computational accuracy exceeds the preset computational accuracy and the AUC exceeds the preset AUC, then the reliability verification is considered to be satisfied.

[0053] AUC reflects the classification ability of the analysis model. Calculation accuracy reflects the classification accuracy of the analysis model. The preset calculation accuracy can be set to any value higher than 50%, and the preset AUC can be set to any value higher than 0.5. The higher the preset calculation accuracy and preset AUC, the higher the requirements for the analysis model, and the better the model performance.

[0054] For example, in the test set, the lipoprotein barrier analysis model achieved a computational accuracy of 73.59% and an AUC of 0.7798. These experimental data demonstrate that the lipoprotein barrier analysis model exhibits high accuracy and classification ability.

[0055] In summary, this approach obtains the multi-round training weights of the first hidden layer in the lipoprotein disorder analysis model corresponding to each protein feature during multiple training rounds. Based on these weights, it determines the importance feature score for each protein feature. Then, based on the importance feature score, it identifies at least one important feature protein corresponding to lipoprotein disorder. By dynamically tracking the weight evolution of the first hidden layer during multi-round training, it achieves refined quantification of the importance of protein features. Calculating the importance feature score for each protein feature based on the obtained multi-round training weights not only enhances the model's interpretability but also effectively distinguishes noise features from truly disease-related biological signals. Finally, based on the importance feature score, it screens out important feature proteins highly correlated with lipoprotein disorder, providing potential biomarkers (i.e., important feature proteins) with strong biological interpretability for lipoprotein disorder.

[0056] Example 2: Another embodiment of this application relates to a device for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning. The implementation details of this device are described below. The following details are for ease of understanding and are not essential for implementing this solution. A schematic diagram of the device 340 for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning in this embodiment can be seen as follows: Figure 3 As shown, it includes an acquisition unit 301 and a determination unit 302.

[0057] The acquisition unit 301 is used to acquire the multi-round training weights of the first hidden layer in the lipoprotein barrier analysis model and the corresponding protein features during the multi-round training process.

[0058] The determining unit 302 is used to determine the important feature score corresponding to each protein feature based on the multi-round training weights.

[0059] The determining unit 302 is further configured to determine at least one important feature protein corresponding to lipoprotein disorder based on the important feature score.

[0060] In some examples, when determining at least one important feature protein corresponding to a lipoprotein barrier based on the important feature score, the determining unit 302 is specifically used to: obtain at least one reference target protein corresponding to the lipoprotein barrier; obtain the important feature score corresponding to each reference target protein respectively, and take the reference target protein that meets the preset conditions as the important feature protein.

[0061] In some examples, when determining the reference target protein that meets the preset conditions as the important feature protein, the determining unit 302 is specifically used to: determine at least one reference target protein whose important feature score is greater than or equal to the important feature score threshold as the important feature protein, or determine a preset number of reference target proteins with the largest important feature score as the important feature protein.

[0062] In some examples, when determining the important feature score corresponding to each protein feature based on the multi-round training weights, the determining unit 302 is specifically used to: for each protein feature, obtain the average weight corresponding to the protein feature based on the multi-round training weights; and use the average weight as the important feature score corresponding to the protein feature.

[0063] In some examples, when determining the average weight corresponding to the protein feature based on the multi-round training weights, the determining unit 302 is specifically used to: obtain the round weight value corresponding to the multi-round training weights; and determine the average weight corresponding to the protein feature based on the round weight value, the multi-round training weights, and the total number of training rounds.

[0064] In some cases, the round weight value is positively correlated with the number of rounds.

[0065] In some examples, when determining the weight value of the training round corresponding to the weight of the multi-round training, the determining unit 302 is specifically used to: obtain the total number of training rounds of the lipoprotein barrier analysis model; for each training round, obtain the training progress corresponding to the training round; and determine the weight value of each training round based on the training progress and the initial weight value of the training round.

[0066] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed in this application; however, this does not mean that other units are absent in this embodiment.

[0067] Example 3: Another embodiment of this application relates to an electronic device, such as... Figure 4 As shown, it includes: at least one processor 901; and a memory 902 communicatively connected to the at least one processor 901; wherein the memory 902 stores instructions executable by the at least one processor 901, the instructions being executed by the at least one processor 901 to enable the at least one processor 901 to perform the method for analyzing important characteristic proteins of lipoprotein barriers based on deep learning in the above embodiments.

[0068] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0069] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0070] Example 4: Another embodiment of this application relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0071] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0072] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.

Claims

1. A method for analyzing the expression of key characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning, characterized in that, include: Obtain the multi-round training weights of the first hidden layer in the lipoprotein barrier analysis model corresponding to each protein feature during multiple rounds of training. Based on the multi-round training weights, the important feature score corresponding to each protein feature is determined; Based on the important feature score, at least one important feature protein corresponding to lipoprotein disorder is identified.

2. The method for analyzing key characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning according to claim 1, characterized in that, The step of determining at least one important feature protein corresponding to lipoprotein disorder based on the important feature score includes: Obtain at least one reference target protein corresponding to the lipoprotein barrier; The important feature scores corresponding to each reference target protein are obtained, and the reference target proteins that meet the preset conditions are taken as the important feature proteins.

3. The method for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning according to claim 2, characterized in that, The step of using the reference target protein that meets the preset conditions as the important feature protein includes: At least one reference target protein whose important feature score is greater than or equal to the important feature score threshold is designated as the important feature protein, or The reference target proteins with the highest number of important feature scores are selected as the important feature proteins.

4. The method for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning according to claim 1, characterized in that, The determination of the important feature score corresponding to each protein feature based on the multi-round training weights includes: For each protein feature, the average weight corresponding to the protein feature is obtained based on the multi-round training weights; The average weight is used as the important feature score corresponding to the protein feature.

5. The method for analyzing key characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning according to claim 4, characterized in that, The step of obtaining the average weight corresponding to the protein feature based on the multi-round training weights includes: Obtain the weight values ​​for each training round corresponding to the weights in multiple rounds; The average weight of the protein feature is determined based on the round weight value, the multi-round training weight, and the total number of training rounds.

6. The method for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning according to claim 5, characterized in that, The round weight value is positively correlated with the number of rounds.

7. The method for analyzing important characteristic proteins of lipoprotein disorders based on multi-target protein expression using deep learning according to claim 5, characterized in that, The step of obtaining the round weight values ​​corresponding to the multi-round training weights includes: Obtain the total number of training rounds for the lipoprotein barrier analysis model; For each training round, obtain the training progress corresponding to that training round; Based on the training progress and the initial round weight value, the round weight value corresponding to each training round is determined.

8. A device for analyzing the expression of multi-target proteins based on deep learning to identify key characteristic proteins of lipoprotein disorders, characterized in that, include: The acquisition unit is used to acquire the multi-round training weights of the first hidden layer in the lipoprotein barrier analysis model corresponding to each protein feature during multi-round training. A determining unit is used to determine the important feature score corresponding to each protein feature based on the multi-round training weights; The determining unit is further configured to determine at least one important feature protein corresponding to lipoprotein disorder based on the important feature score.

9. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method for analyzing important characteristic proteins of lipoprotein barriers based on deep learning, as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method for analyzing important characteristic proteins of lipoprotein disorders based on deep learning for multi-target protein expression analysis as described in any one of claims 1 to 7.

Images