Drug aging relationship modeling method based on dynamic network pharmacology
By constructing a dynamic network of patients and normal samples, simulating network changes after medication, and calculating difference values to evaluate drug efficacy, the problem of insufficient accuracy in drug efficacy evaluation in existing technologies is solved, and accurate guidance of drugs in clinical treatment is achieved.
Patent Information
- Application Number
- CN202510960964.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-12
- Publication Date
- 2025-10-17
AI Technical Summary
Existing drug efficacy evaluation methods rely on a single indicator, making it difficult to quantify the dynamic process of the disease network after drug intervention. They ignore the efficiency of repairing the overall network disorder and lack verification of the spatiotemporal correlation of the drug's action mechanism, resulting in one-sided or delayed evaluation and inability to accurately guide the optimization of clinical medication plans.
By constructing a disease dynamic network of patients and normal samples, simulating the changes in the network after medication, calculating the difference value to evaluate the drug-time effect relationship, and introducing dynamic network pharmacology to integrate multi-omics data, the effect of drugs in spatiotemporal changes is evaluated.
It improves the accuracy of drug efficacy assessment, can accurately guide the use of drugs in clinical treatment, and provide consistency verification between dynamic network difference changes and drug action mechanisms.
Smart Images
Figure CN120809276A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of drug evaluation, in particular to a drug time-effect relationship modeling method based on dynamic network pharmacology. BACKGROUND
[0002] Dynamic network pharmacology builds a dynamic biological network of drug-target-pathway-disease, integrates multi-omics dynamic data to capture the timing changes of biomolecular networks, and provides a systematic perspective for analyzing drug action mechanisms. Existing drug time-effect evaluation methods mostly rely on single indicators, which have significant limitations: first, it is difficult to quantify the dynamic process of disease network repair to normal state after drug intervention, and it is difficult to distinguish the correlation between local molecular interaction repair and global network topology reconstruction; second, the spatiotemporal correlation between evaluation indicators and drug action mechanism is insufficient, for example, only the target binding rate is used to judge drug efficacy, ignoring the repair efficiency of network overall disorder; third, there is a lack of intuitive verification of the dynamic consistency of drug action mechanism, local repair effect and global network improvement, leading to one-sidedness or lag of drug efficacy evaluation. These defects make it difficult for existing methods to accurately guide the optimization of clinical drug regimens, and there is an urgent need for a time-effect evaluation method that can integrate dynamic network difference changes and drug action mechanism. SUMMARY
[0003] In view of this, the present application provides a drug time-effect relationship modeling method based on dynamic network pharmacology, which can improve the accuracy of drug time-effect evaluation by introducing dynamic networks.
[0004] The technical scheme of the present application is as follows:
[0005] A drug time-effect relationship modeling method based on dynamic network pharmacology, comprising the following steps:
[0006] Step S1, obtaining disease information of a patient sample, and constructing a disease dynamic network of the patient sample based on the disease information;
[0007] Step S2, obtaining basic information of the patient sample, querying a normal sample that meets the requirements according to the basic information of the patient sample, and constructing a normal dynamic network of the normal sample according to the disease dynamic network;
[0008] Step S3, simulating the change data of the disease dynamic network of the patient sample at different time points after drug administration, and calculating the difference value between the disease dynamic network and the normal dynamic network based on the change data;
[0009] Step S4, evaluating the time-effect relationship of the drug according to the difference value.
[0010] Preferably, the specific steps of step S1 include:
[0011] Step S11, acquire the disease category to which the drug to be tested is directed, and query the patient samples with the same disease from the database based on the disease category;
[0012] Step S12, input the disease category into the trained neural network, and obtain the pathological tissue corresponding to the disease by processing the neural network;
[0013] Step S13, collect the disease information at the pathological tissue of the patient sample, and construct the disease dynamic network of the patient sample based on the disease information.
[0014] Preferably, the training step of the neural network in step S12 is: acquiring a large number of disease categories and corresponding pathological tissues as training data, dividing the training data into a training set and a test set according to a proportion, training the neural network through the training set, and testing the accuracy of the neural network through the test set.
[0015] Preferably, the specific steps of step S13 are:
[0016] Step S131, perform genome sequencing, transcriptome sequencing, proteomics, and metabolomics on the pathological tissue to obtain multi-dimensional molecular characteristics;
[0017] Step S132, screen the molecules related to the disease from the multi-dimensional molecular characteristics as network nodes;
[0018] Step S133, calculate the Pearson correlation coefficient between the network nodes, connect the network nodes with a Pearson correlation coefficient greater than a correlation threshold as network edges, and construct a normal dynamic network.
[0019] Preferably, the specific steps of step S2 include:
[0020] Step S21, extract the demographic characteristics, clinical diagnosis, pathological characteristics, and treatment history of the patient sample from the hospital database as basic information;
[0021] Step S22, match the samples similar to the basic information of the patient from the volunteer database according to the basic information;
[0022] Step S23, determine whether the sample has the same disease as the patient sample, and if not, output as a normal sample;
[0023] Step S24, construct a normal dynamic network of the normal sample according to the disease dynamic network.
[0024] Preferably, the specific step of the step S24 is: extracting network nodes in the disease dynamic network as network nodes in the normal dynamic network, calculating the Pearson correlation coefficient between the network nodes based on the body data of the normal samples, connecting the network nodes with the Pearson correlation coefficient greater than a correlation threshold as network edges, and constructing the normal dynamic network.
[0025] Preferably, the specific step of the step S3 comprises:
[0026] The step S31 comprises: constructing a basic evolution model based on the disease dynamic network, and embedding the drug to be tested into the basic evolution model.
[0027] The step S32 comprises: initializing the basic evolution model and the parameters of the drug to be tested, and generating state data of the disease dynamic network at different time points after the drug is taken through numerical simulation.
[0028] The step S33 comprises: extracting reference feature data of the normal dynamic network, and calculating difference values of the state data at different time points and the reference feature data.
[0029] Preferably, the specific step of calculating the difference values in the step S33 is: for each time point of simulation, calculating a network node activity error and a network edge weight distance between the disease dynamic network after the drug is taken and the normal dynamic network as a local difference value, calculating an Euclidean distance of a network topology structure between the disease dynamic network after the drug is taken and the normal dynamic network, and obtaining a global difference value based on the Euclidean distance.
[0030] Preferably, when the Euclidean distance of the network topology structure between the disease dynamic network after the drug is taken and the normal dynamic network is calculated, core topology parameters are extracted according to the network topology structures of the disease dynamic network and the normal dynamic network, the core topology parameters form respective topology feature vectors, and the Euclidean distance of the topology feature vectors is calculated as the global difference value.
[0031] Preferably, the specific step of the step S4 comprises:
[0032] The step S41 comprises: obtaining a key index of an action mechanism of the drug to be tested, and drawing an action curve diagram of the key index changing over time.
[0033] The step S42 comprises: superimposing the local difference value and the global difference value changing over time on the action curve diagram respectively.
[0034] The step S43 comprises: when the local difference value curve and the global difference value curve change synchronously with the action curve diagram, judging that the drug is effective.
[0035] Compared with the prior art, the present application has the following beneficial effects:
[0036] The application discloses a drug-time effect relationship modeling method based on dynamic network pharmacology. BRIEF DESCRIPTION OF DRAWINGS
[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only preferred embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.
[0038] Figure 1 The flow chart of the drug-time effect relationship modeling method based on dynamic network pharmacology.
[0039] Figure 2 The flow chart of step S1 of the drug-time effect relationship modeling method based on dynamic network pharmacology.
[0040] Figure 3 The flow chart of step S13 of the drug-time effect relationship modeling method based on dynamic network pharmacology.
[0041] Figure 4 The flow chart of step S2 of the drug-time effect relationship modeling method based on dynamic network pharmacology.
[0042] Figure 5 The flow chart of step S3 of the drug-time effect relationship modeling method based on dynamic network pharmacology.
[0043] Figure 6 The flow chart of step S4 of the drug-time effect relationship modeling method based on dynamic network pharmacology. DETAILED DESCRIPTION
[0044] In order to better understand the technical content of the present application, a specific embodiment is provided below, and the present application is further described in combination with the drawings.
[0045] REFERENCE Figures 1 to 6The application provides a drug-time relationship modeling method based on dynamic network pharmacology, which comprises the following steps:
[0046] In step S1, disease information of a patient sample is acquired, and a disease dynamic network of the patient sample is constructed based on the disease information.
[0047] In step S2, basic information of the patient sample is acquired, a normal sample meeting the requirement is queried according to the basic information of the patient sample, and a normal dynamic network of the normal sample is constructed according to the disease dynamic network.
[0048] In step S3, change data of the disease dynamic network at different time points after the patient sample is simulated to take medicine is simulated, and a difference value calculation of the disease dynamic network and the normal dynamic network is performed based on the change data.
[0049] In step S4, the time-effect relationship of the drug is evaluated according to the difference value.
[0050] The application discloses a drug time-effect relationship modeling method based on dynamic network pharmacology, which is applied to drug efficacy evaluation of new drugs.
[0051] The application introduces dynamic network, simulates the space-time change in the disease dynamic network when the test drug acts on the patient sample, and includes the flow direction, action position and action effect of the test drug in the disease dynamic network, instead of traditional static comparison, thereby improving the accuracy of drug time-effect evaluation and providing guidance for clinical use of the test drug.
[0052] Preferably, the specific steps of the step S1 include:
[0053] The step S11 comprises the following steps:
[0054] Step S12, input the disease category into the trained neural network, and obtain the pathological tissue corresponding to the disease by the neural network processing;
[0055] Step S13, collect the disease information of the pathological tissue of the patient sample, and construct the disease dynamic network of the patient sample based on the disease information.
[0056] After the production of the test drug is completed, there will be several disease categories for intervention treatment. After the disease category is determined, the patient with the same disease can be queried from the hospital database, and the patient is taken as the patient sample. In order to construct the disease dynamic network, the corresponding part of the human body acted on by the test drug, that is, the pathological tissue, needs to be determined. The present application introduces a deep learning network, takes the disease category as the input of the neural network, and automatically identifies the pathological tissue by the neural network. Then the disease information of the pathological tissue of the patient sample can be collected, and the disease dynamic network about the patient sample can be constructed based on the disease information.
[0057] Preferably, the training step of the neural network in step S12 is: obtaining a large number of disease categories and corresponding pathological tissues as training data, dividing the training data into a training set and a test set according to a proportion, training the neural network through the training set, and testing the accuracy of the neural network through the test set.
[0058] The neural network is trained by a large amount of existing data, wherein the disease category and the corresponding pathological tissue are taken as a group of data. A large amount of data is divided into a training set and a test set according to a proportion of 7:3. A neural network containing multiple layers is constructed. The neural network is trained through the training set. After the training is completed, the accuracy is tested through the test set. When the accuracy is greater than a preset threshold, the construction of the neural network is stopped. At this time, the neural network can be used for identifying the pathological tissue.
[0059] Preferably, the specific steps of step S13 are:
[0060] Step S131, genome sequencing, transcriptome sequencing, proteomics and metabolomics are performed on the pathological tissue to obtain multi-dimensional molecular characteristics;
[0061] Step S132, screening the molecules related to the disease from the multi-dimensional molecular characteristics as network nodes;
[0062] Step S133, calculating the Pearson correlation coefficient between the network nodes, connecting the network nodes with the Pearson correlation coefficient greater than a correlation threshold as network edges, and constructing a normal dynamic network.
[0063] After pathological tissues of patient samples are obtained, disease data needs to be extracted, wherein a dynamic network is mainly composed of nodes and edges, after feature extraction of pathological tissues, multi-dimensional molecular features can be obtained, and molecules closely related to diseases are screened from the multi-dimensional molecular features as network nodes, such as mutant genes, differentially expressed proteins, abnormal metabolites and the like, and then the network nodes need to be connected to form edges, and the connection is not random connection, but the correlation between different network nodes needs to be calculated, and the Pearson correlation coefficient calculation method is adopted in the application to calculate the Pearson correlation coefficient of the network nodes, if the Pearson correlation coefficient is greater than a preset correlation threshold, it indicates that the correlation between the two network nodes is large, at this time, the corresponding two network nodes can be connected to form a network edge, after calculation and connection of all network nodes, a disease dynamic network containing a plurality of network nodes and network edges can be formed.
[0064] Preferably, the specific steps of the step S2 include:
[0065] Step S21, extracting demographic characteristics, clinical diagnosis, pathological characteristics and treatment history of the patient sample from the hospital database as basic information;
[0066] Step S22, matching samples similar to the basic information of the patient from the volunteer database according to the basic information;
[0067] Step S23, judging whether the sample has the same disease as the patient sample, if not, outputting as a normal sample;
[0068] Step S24, constructing a normal dynamic network of the normal sample according to the disease dynamic network.
[0069] After the disease dynamic network is constructed, a normal dynamic network also needs to be constructed, therefore, a normal sample needs to be selected, the normal sample needs to be basically consistent with the patient sample, therefore, the basic information of the patient sample needs to be extracted first, including age, gender and other demographic characteristics, disease type, stage and other clinical diagnosis data, lesion tissue type, location and other pathological characteristics and treatment history, and then matching can be performed in the volunteer database, the matching rule is defined as: similar age, same sex, normal tissue at the same location as the pathological tissue of the patient sample, no treatment history and the like, after the normal sample is matched, the normal dynamic network can be constructed.
[0070] Preferably, the specific steps of the step S24 are: extracting the network nodes in the disease dynamic network as the network nodes of the normal dynamic network, calculating the Pearson correlation coefficient between the network nodes based on the body data of the normal sample, connecting the network nodes with the Pearson correlation coefficient greater than the correlation threshold as the network edges, and constructing the normal dynamic network.
[0071] In constructing the normal dynamic network, the selected network nodes are consistent, ensuring that the core research objects of both are completely the same, avoiding comparison deviation caused by network node difference, and after determining the network nodes, the network edges also need to be determined. The determination method of the network edges is consistent with the determination method of the disease dynamic network, and is also obtained by calculating the Pearson correlation coefficient. Because the data of the corresponding tissues of the normal samples and the patient samples, i.e. the body data, are different, the correlation coefficients obtained by calculation will also be different, and the network edges finally obtained will also be different, so that a normal dynamic network different from the disease dynamic network can be constructed.
[0072] Preferably, the specific steps of the step S3 include:
[0073] Step S31, constructing a basic evolution model based on the disease dynamic network, and embedding the drug to be tested into the basic evolution model;
[0074] Step S32, initializing the parameters of the basic evolution model and the drug to be tested, and generating state data of the disease dynamic network at different time points after taking the drug through numerical simulation;
[0075] Step S33, extracting the reference feature data of the normal dynamic network, and calculating the difference value between the state data at different time points and the reference feature data.
[0076] After obtaining the disease dynamic network and the normal dynamic network, the difference between the two in space-time evolution needs to be calculated. Therefore, a basic evolution model needs to be constructed first, and the drug to be tested is embedded into the basic evolution model. After initializing the parameters of the basic evolution model and the drug to be tested, the evolution process of the disease dynamic network after taking the drug can be simulated. Through numerical simulation, state data of the disease dynamic network at different time points can be generated. At the same time, the reference feature data of the normal dynamic network at the corresponding time point is collected, and the difference value is calculated.
[0077] Preferably, the specific steps of calculating the difference value in the step S33 are as follows: for each simulated time point, the network node activity error and the network edge weight distance between the disease dynamic network after taking the drug and the normal dynamic network are calculated as local difference values, the Euclidean distance of the network topology structure between the disease dynamic network after taking the drug and the normal dynamic network is calculated, and the global difference value is obtained based on the Euclidean distance.
[0078] The difference value includes local difference value and overall difference value. In the drug-taking process, the drug will affect different network nodes and network edges, so the network node activity error and the network edge weight distance at the same time point can be calculated as the local difference value. At the same time, the topology structure of the disease dynamic network will also change, and the Euclidean distance of the topology structure between the disease dynamic network at different time points and the normal dynamic network is calculated as the global difference value. The drug efficacy is evaluated by the local difference value and the global difference value.
[0079] Preferably, in the calculation of the Euclidean distance of the network topology of the disease dynamic network after drug administration and the normal dynamic network, the core topological parameters are extracted according to the network topology of the disease dynamic network and the normal dynamic network, the topological feature vectors are formed according to the core topological parameters, and the Euclidean distance of the topological feature vectors is calculated as the global difference value.
[0080] And in the calculation of the Euclidean distance, the core topological parameters of the topological structure need to be extracted first, including average degree, clustering coefficient, shortest path length, hub node betweenness centrality, etc. Then the topological feature vectors are formed, and the Euclidean distance of the topological feature vectors of the disease dynamic network and the normal dynamic network is calculated. If the drug is effective, the topological structure will be more similar, and the Euclidean distance will be smaller.
[0081] Preferably, the specific steps of the step S4 include:
[0082] Step S41, obtaining the key indicators of the action mechanism of the drug to be tested, and drawing an action curve graph changing with time;
[0083] Step S42, superimposing the curves of the local difference value and the global difference value changing with time respectively on the action curve graph;
[0084] Step S43, when the changes of the local difference value curve and the global difference value curve are synchronized with the action curve graph, judging that the drug is effective.
[0085] When evaluating the time-effect relationship of the drug, the key indicators of the action mechanism of the drug to be tested, such as target binding rate and pathway activation degree, are needed to draw the action curve graph changing with time, and the local difference value and the global difference value changing with time will also draw the corresponding curves. The curves are superimposed on the action curve graph, and whether the drug is effective is determined according to whether the curves are synchronized. Only when the local difference value curve and the global difference value curve are synchronized with the action curve graph, it can be judged that the drug is effective. When one of the curves is not synchronized, it means that the drug has not effectively transmitted to the network level, and related tests need to be performed again.
[0086] The above only describes the preferred embodiments of the present application and should not be used to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.
Claims
1. A drug-time-effect relationship modeling method based on dynamic network pharmacology, characterized in that: The following steps are involved: Step S1: Obtain disease information of patient samples and construct a disease dynamic network of patient samples based on the disease information; Step S2: Obtain basic information of the patient sample, search for normal samples that meet the requirements based on the basic information of the patient sample, and construct a normal dynamic network of the normal sample based on the disease dynamic network; Step S3, simulating the change data of the disease dynamic network at different times after the patient sample takes the medicine, and calculating the difference value between the disease dynamic network and the normal dynamic network based on the change data; Step S4: Evaluate the time-effect relationship of the drug based on the difference value.
2. A drug-time-effect relationship modeling method based on dynamic network pharmacology according to claim 1, characterized in that: The specific steps of step S1 include: Step S11: Obtain the disease type targeted by the drug to be tested, and query the database for patient samples with the same disease based on the disease type; Step S12: input the disease type into the trained neural network, and the neural network processes the disease to obtain the pathological tissue corresponding to the disease; Step S13: collecting disease information from the pathological tissue of the patient sample, and constructing a disease dynamic network of the patient sample based on the disease information.
3. The method for modeling drug-effect relationship based on dynamic network pharmacology according to claim 2, characterized in that: The training steps of the neural network in step S12 are: obtaining a large number of disease types and corresponding pathological tissues as training data, dividing the training data into a training set and a test set in proportion, training the neural network through the training set, and testing the accuracy of the neural network through the test set.
4. The method for modeling drug-effect relationship based on dynamic network pharmacology according to claim 2, characterized in that: The specific steps of step S13 are: Step S131: performing genome sequencing, transcriptome sequencing, proteomics, and metabolomics on the pathological tissue to obtain multi-dimensional molecular features; Step S132: selecting molecules related to the disease from the multi-dimensional molecular features as network nodes; Step S133: Calculate the Pearson correlation coefficient between network nodes, connect network nodes whose Pearson correlation coefficient is greater than the correlation threshold as network edges, and construct a normal dynamic network.
5. The method for modeling drug-effect relationship based on dynamic network pharmacology according to claim 4, characterized in that: The specific steps of step S2 include: Step S21: extracting the demographic characteristics, clinical diagnosis, pathological characteristics, and treatment history of the patient sample from the hospital database as basic information; Step S22: matching samples with similar basic information to the patient from the volunteer database based on the basic information; Step S23: determine whether the sample suffers from the same disease as the patient sample. If not, output it as a normal sample; Step S24: constructing a normal dynamic network of normal samples based on the disease dynamic network.
6. The method for modeling drug-time-effect relationship based on dynamic network pharmacology according to claim 5, characterized in that: The specific steps of step S24 are: extracting network nodes in the disease dynamic network as network nodes of the normal dynamic network, calculating the Pearson correlation coefficient between network nodes based on the physical data of normal samples, connecting network nodes with Pearson correlation coefficients greater than the correlation threshold as network edges, and constructing a normal dynamic network.
7. The method for modeling drug-time-effect relationship based on dynamic network pharmacology according to claim 1, characterized in that: The specific steps of step S3 include: Step S31: constructing a basic evolution model based on the disease dynamic network, and embedding the drug to be tested into the basic evolution model; Step S32: Initialize the basic evolution model and the parameters of the drug to be tested, and generate the state data of the disease dynamic network at different times after drug administration through numerical simulation; Step S33: extracting the baseline characteristic data of the normal dynamic network, and calculating the difference between the state data at different moments and the baseline characteristic data.
8. The method for modeling drug-time-effect relationship based on dynamic network pharmacology according to claim 7, characterized in that: The specific steps of calculating the difference value in step S33 are: for each time point of the simulation, calculating the network node activity error and network edge weight distance between the disease dynamic network after medication and the normal dynamic network as local difference values, calculating the Euclidean distance of the network topology structure between the disease dynamic network after medication and the normal dynamic network, and obtaining the global difference value based on the Euclidean distance.
9. The method for modeling drug-time-effect relationship based on dynamic network pharmacology according to claim 8, characterized in that: When calculating the Euclidean distance between the network topology structures of the disease dynamic network and the normal dynamic network after medication, the core topology parameters are first extracted according to the network topology structures of the disease dynamic network and the normal dynamic network, and the respective topological feature vectors are formed according to the core topological parameters. The Euclidean distance of the topological feature vectors is calculated as the global difference value.
10. The method for modeling drug-time-effect relationship based on dynamic network pharmacology according to claim 9, characterized in that: The specific steps of step S4 include: Step S41: obtaining key indicators of the mechanism of action of the drug to be tested and drawing a graph of its action changing over time; Step S42: superimpose the time-varying curves of the local difference value and the global difference value on the action curve graph; Step S43: When the changes of the local difference value curve and the global difference value curve are synchronized with the action curve graph, it is determined that the drug is effective.