Data-driven Chinese herbal medicine relocation method
Through data-driven methods, machine learning technology and cluster analysis are used to build a correlation network between Chinese herbal medicine and disease, solving the problem of existing Chinese herbal relocation methods relying on experience and lack of scientific verification, and achieving the efficiency and accuracy of Chinese herbal medicine relocation.
Patent Information
- Application Number
- CN202411627927.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-14
- Publication Date
- 2025-02-25
AI Technical Summary
The existing Chinese herbal relocation methods have limitations, relying on experience and manual screening, lack of systematic analysis and accurate prediction methods, resulting in time-consuming, high cost, low success rate, and lack of scientific verification and model support.
Using a data-driven method, we use data-driven methods to collect Chinese herbal ingredients, target genes and disease information from large public databases, integrate and clean the data, build a drug similarity network and a disease similarity network, and use machine learning technology and cluster analysis to predict the association relationship between drugs and diseases.
It significantly improves the efficiency and accuracy of Chinese herbal repositioning, reduces manual intervention, improves analysis efficiency, provides scientific drug recommendations, and ensures its reliability and effectiveness in practical applications through rigorous model performance verification.
Smart Images
Figure CN119517451A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of drug repositioning, and more particularly to a data-driven Chinese herbal medicine repositioning method. Background Art
[0002] In traditional medicine, the use of Chinese herbal medicine is usually based on ancient experience and theory. Chinese herbal medicine repositioning is an important concept in the field of traditional Chinese medicine. Chinese herbal medicine repositioning refers to the exploration and development of new functions and new indications of Chinese herbal medicine through modern scientific and technological means, so as to find its potential value in the treatment of other diseases or new fields. Existing Chinese herbal medicine repositioning methods have deficiencies and shortcomings, such as: 1. Limitations of traditional methods: Traditional methods of repositioning Chinese herbal medicines mostly rely on experience and manual screening, lacking systematic analysis and accurate prediction methods. This process is usually time-consuming, costly, and has a low success rate, which limits the rapid development of new drugs.
[0003] 2. Insufficient data integration and analysis: Existing technologies lack efficient data mining and machine learning technical support when processing and integrating large amounts of complex data on Chinese herbal medicine ingredients and diseases. The quality and diversity of the data are poor, which affects the accuracy and reliability of the prediction results.
[0004] 3. Lack of scientific verification and model support: Many traditional Chinese herbal medicine repositioning methods lack sufficient scientific verification and quantitative analysis, resulting in insufficient reproducibility and credibility of the results. Existing research mostly remains at the theoretical discussion or small-scale experimental stage, lacking large-scale clinical verification and practical research.
[0005] 4. Insufficient technical means: The application of data mining technology and machine learning algorithms has not been widely promoted, resulting in low accuracy and efficiency of Chinese herbal medicine repositioning. The insufficient application of high-performance computing and big data analysis technology has restricted the innovation and development of Chinese herbal medicine repositioning methods.
[0006] 5. Limitations of the knowledge system: There is a large knowledge gap between Chinese herbal medicine and modern biomedicine. There is a lack of effective interdisciplinary research and technological integration. Research in related fields is mostly focused on a single drug ingredient or disease, and overall systematic research is insufficient.
[0007] Therefore, a data-driven Chinese herbal medicine repositioning method is urgently needed to solve the above problems. Summary of the invention
[0008] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a data-driven Chinese herbal medicine repositioning method. The present invention introduces advanced data-driven and machine learning technologies to significantly improve the efficiency and accuracy of Chinese herbal medicine repositioning, optimize data integration and processing capabilities, improve prediction models and algorithms, provide scientific drug recommendations, and through strict model performance verification, ensure its reliability and effectiveness in practical applications, so as to solve the problems raised in the above-mentioned background technology.
[0009] To achieve the above object, the present invention provides the following technical solution: a data-driven Chinese herbal medicine repositioning method, comprising the following steps: Step 1: Data collection and processing: collect information on Chinese herbal medicine ingredients, target genes, and diseases from large public databases such as TCMSP, SymMap, and UniProt; Step 2: Data integration, including data cleaning, format unification, data removal, and data conversion, to ensure data consistency and integrity; Step 3: Feature construction, using multidimensional binary vectors to represent drug features, including chemical structure, protein domain, and gene ontology information, calculating the similarity between drugs and diseases through the Jaccard similarity coefficient, and constructing a similarity matrix; Step 4: Construct drug similarity network and disease similarity network, use the method of combining path length and node degree to calculate the distance between drug module and disease module, so as to quantify the interaction between drug and disease. Drug module and disease module are located at different positions in the network. Drug small molecules and disease genes represent the corresponding nodes in the network respectively. The distance between drug module and disease module is calculated by combining path length, node degree and path length to construct drug-disease network. The formula is as follows: (4-1); (4-2); Where d(T, S) is the distance between the drug target t in the drug module T and the disease gene s in the disease module S, G is the network; Dmax(G) refers to the diameter of the graph G, dis(t, s) refers to the shortest path length between nodes t and s, degt refers to the degree of node t, and degG refers to the maximum degree in the graph G; Step 5: Calculation of similarity of Chinese herbal medicines. In order to improve the relevance and accuracy of the calculation results, the model constructs a reference distance distribution, which simulates the expected distance between two groups of randomly selected proteins. The selection of these proteins is based on the similarity of connectivity with target disease-related or drug target proteins. The distance between the two groups of proteins is randomly calculated 1000 times and the distance values are stored. The size of the Z value is calculated by using the mean and standard deviation of the reference distance distribution. This randomization process allows the biological significance of the network connection between any pair of nodes to be retained. The formula is as follows: (5-1); Where d(T, S) is the drug-disease distance value in formula (3.1) and formula (3.2), μd(T, S) is the mean, and σ_(d(T,S)) is the variance; The specific steps for calculating the Z value are roughly divided into the following five steps: (1) Sort all gene nodes by their connectivity, from high to low; (2) Gene nodes are randomly selected in the ranking, and the selection range includes the first 100 and last 100 nodes of each node; (3) Calculate the distance between the selected drug gene node and the disease gene node; (4) Steps (2) and (3) are repeated 1000 times and the results are stored in an array; (5) For each drug-disease pair, there are 1000 data points. Calculate the mean and variance, and get the Z value according to formula (5-1); Step 6: Cluster analysis, enhance the predictive ability of the model, using the ClusterONE clustering method. This step aims to strengthen the cohesion within the disease-related protein cluster. First, the similarity data between the drug module and the disease module obtained by formula (5-1) is constructed into a weighted graph, where each node represents a drug or disease, each edge represents a possible interaction relationship, and the weight of the edge represents the confidence of the interaction. Then, the network is preprocessed by specifying parameter settings, and a greedy strategy-based algorithm is used to search for high-quality clusters in the network, gradually expanding and merging nodes to form clusters. In this process, ClusterONE considers the connection density and mutual connectivity between nodes to determine the optimal clustering structure. Finally, the score f(v) of each disease module is calculated by evaluating the quality of the clustering, which represents the cohesion of the module. The formula is as follows: (6-1); Among them, Win(v) represents the total weight of all "edges" in the disease module, Wout(v) represents the total weight of the edges connecting the disease module with the rest of the network, and P(v) is the penalty term, that is, the node density. When the drug and the disease are in the same cluster, this model will adjust its similarity measure to more accurately reflect the potential connection between the drug and the disease. Through improvement, the model not only improves the cohesion of disease-related clusters, but also enhances the accuracy of identifying potential drug therapeutic effects, thereby improving the accuracy of drug-disease association. The formula for adjusting the drug-disease similarity measure is as follows: (6-2); Among them, sim' is the adjusted similarity; Finally, in order to quantify the probability of association between each drug and disease, this model uses the sigmoid function. The sigmoid function is a commonly used mathematical function, which is usually used to map input values to a range between 0 and 1. The image of the sigmoid function presents an S-shaped curve, which is characterized by that when the x-axis approaches positive infinity, the y-axis approaches 1, and when the x-axis approaches negative infinity, the y-axis approaches 0. This function shape makes it often used to convert input values of any range into probability values. This method changes the parameters of the sigmoid function, and the formula for calculating the probability of association between each drug and disease is as follows: (6-3); Where x is the drug-disease similarity value, b is the midpoint of the sigmoid function curve, the value of b is half of the maximum value of x, and c is the steepness of the sigmoid function curve, c=10; Step 7: Repositioning of Chinese herbal medicines. The model predicts the potential drug-disease association and generates a prediction probability list. The resulting probability list can be viewed as a weighted bipartite network, in which one group represents drugs and the other group represents diseases. In this network, there is a weighted edge between each pair of nodes, and the weight reflects the degree of association between the drug and the disease. The existence of these weighted edges conveys the probability of interaction between drug small molecules and disease modules, and provides detailed related information, which helps to deeply understand the interaction patterns between drugs and diseases and their potential effects.
[0010] In a preferred embodiment, the formula (4-1) in step 4 adopts a path length calculation method, that is, calculating the distance between the drug target gene and the closest disease protein. All target genes of each drug correspond to a closest distance value, and finally calculating the average value of the closest distances corresponding to all target genes. The closer the calculation result is to 1, the closer the distance between the drug module and the disease module is.
[0011] In a preferred embodiment, the formula (4-2) in step 4 adopts a calculation method that combines node degree and path length. Node degree refers to the number of connections between a node and other nodes, which is an important concept in complex network analysis. In a complex network, each node has a degree, that is, the number of edges directly connected to the node. The degree of a node can help understand the importance and connection of nodes in the network by analyzing the degree of the node and the distribution of the node degree, as well as better understanding the structure and characteristics of the complex network, identifying the key nodes in the network and studying the robustness and robustness of the network. The calculation result of the node degree method used in formula (4-2) has a value range of [0,2]. Therefore, the closer the calculated result is to 2, the closer the distance between the drug module and the disease module is.
[0012] Technical effects and advantages of the present invention: The present invention improves the efficiency and accuracy of Chinese herbal medicine repositioning. By introducing data-driven and machine learning technologies, the efficiency and accuracy of Chinese herbal medicine repositioning are significantly improved. Through big data and machine learning technologies, accurate matching of Chinese herbal medicine ingredients and diseases is achieved, the need for manual intervention is reduced, and the analysis efficiency is improved. By collecting and integrating Chinese herbal medicine ingredients, target genes and disease information in multiple databases, a comprehensive data set is constructed, and the quality and consistency of the data are improved. The present invention adopts advanced data cleaning, preprocessing and integration methods to ensure the high quality and consistency of the data, providing a solid foundation for subsequent analysis. According to the multi-component and multi-target characteristics of Chinese herbal medicine, a special prediction algorithm is developed, which significantly improves the prediction ability of the model. Through the integrated learning model, the advantages of multiple machine learning algorithms are combined to further improve the accuracy of the prediction. The model of the present invention has high probability and reliability in predicting the effect of drugs on diseases, and provides a scientific basis for drug repositioning. The AUC value of the model of the present invention reaches 0.923, which shows its high performance and reliability in predicting the relationship between Chinese herbal medicine and disease. The model is verified on multiple independent data sets and shows high consistency and robustness. In summary, the present invention significantly improves the efficiency and accuracy of Chinese herbal medicine repositioning by introducing advanced data-driven and machine learning technologies, optimizes data integration and processing capabilities, improves prediction models and algorithms, provides scientific drug recommendations, and ensures its reliability and effectiveness in practical applications through strict model performance verification. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0014] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0015] Example 1 As attached Figure 1 As shown, the present invention provides a data-driven Chinese herbal medicine repositioning method, comprising the following steps: Step 1: Data collection and processing: collect information on Chinese herbal medicine ingredients, target genes, and diseases from large public databases such as TCMSP, SymMap, and UniProt; Step 2: Data integration, including data cleaning, format unification, data removal, and data conversion, to ensure data consistency and integrity; Step 3: Feature construction, using multidimensional binary vectors to represent drug features, including chemical structure, protein domain, and gene ontology information, calculating the similarity between drugs and diseases through the Jaccard similarity coefficient, and constructing a similarity matrix; Step 4: Construct drug similarity network and disease similarity network, use the method of combining path length and node degree to calculate the distance between drug module and disease module, so as to quantify the interaction between drug and disease. Drug module and disease module are located at different positions in the network. Drug small molecules and disease genes represent the corresponding nodes in the network respectively. The distance between drug module and disease module is calculated by combining path length, node degree and path length to construct drug-disease network. The formula is as follows: (4-1); (4-2); Where d(T, S) is the distance between the drug target t in the drug module T and the disease gene s in the disease module S, G is the network; Dmax(G) refers to the diameter of the graph G, dis(t, s) refers to the shortest path length between nodes t and s, degt refers to the degree of node t, and degG refers to the maximum degree in the graph G; Step 5: Calculation of similarity of Chinese herbal medicines. In order to improve the relevance and accuracy of the calculation results, the model constructs a reference distance distribution, which simulates the expected distance between two groups of randomly selected proteins. The selection of these proteins is based on the similarity of connectivity with target disease-related or drug target proteins. The distance between the two groups of proteins is randomly calculated 1000 times and the distance values are stored. The size of the Z value is calculated by using the mean and standard deviation of the reference distance distribution. This randomization process allows the biological significance of the network connection between any pair of nodes to be retained. The formula is as follows: (5-1); Where d(T, S) is the drug-disease distance value in formula (3.1) and formula (3.2), μd(T, S) is the mean, and σ_(d(T,S)) is the variance; The specific steps for calculating the Z value are roughly divided into the following five steps: (1) Sort all gene nodes by their connectivity, from high to low; (2) Gene nodes are randomly selected in the ranking, and the selection range includes the first 100 and last 100 nodes of each node; (3) Calculate the distance between the selected drug gene node and the disease gene node; (4) Steps (2) and (3) are repeated 1000 times and the results are stored in an array; (5) For each drug-disease pair, there are 1000 data points. Calculate the mean and variance, and get the Z value according to formula (5-1); Step 6: Cluster analysis, enhance the predictive ability of the model, using the ClusterONE clustering method. This step aims to strengthen the cohesion within the disease-related protein cluster. First, the similarity data between the drug module and the disease module obtained by formula (5-1) is constructed into a weighted graph, where each node represents a drug or disease, each edge represents a possible interaction relationship, and the weight of the edge represents the confidence of the interaction. Then, the network is preprocessed by specifying parameter settings, and a greedy strategy-based algorithm is used to search for high-quality clusters in the network, gradually expanding and merging nodes to form clusters. In this process, ClusterONE considers the connection density and mutual connectivity between nodes to determine the optimal clustering structure. Finally, the score f(v) of each disease module is calculated by evaluating the quality of the clustering, which represents the cohesion of the module. The formula is as follows: (6-1); Among them, Win(v) represents the total weight of all "edges" in the disease module, Wout(v) represents the total weight of the edges connecting the disease module with the rest of the network, and P(v) is the penalty term, that is, the node density. When the drug and the disease are in the same cluster, this model will adjust its similarity measure to more accurately reflect the potential connection between the drug and the disease. Through improvement, the model not only improves the cohesion of disease-related clusters, but also enhances the accuracy of identifying potential drug therapeutic effects, thereby improving the accuracy of drug-disease association. The formula for adjusting the drug-disease similarity measure is as follows: (6-2); Among them, sim' is the adjusted similarity; Finally, in order to quantify the probability of association between each drug and disease, this model uses the sigmoid function. The sigmoid function is a commonly used mathematical function, which is usually used to map input values to a range between 0 and 1. The image of the sigmoid function presents an S-shaped curve, which is characterized by that when the x-axis approaches positive infinity, the y-axis approaches 1, and when the x-axis approaches negative infinity, the y-axis approaches 0. This function shape makes it often used to convert input values of any range into probability values. This method changes the parameters of the sigmoid function, and the formula for calculating the probability of association between each drug and disease is as follows: (6-3); Where x is the drug-disease similarity value, b is the midpoint of the sigmoid function curve, the value of b is half of the maximum value of x, and c is the steepness of the sigmoid function curve, c=10; Step 7: Repositioning of Chinese herbal medicines. The model predicts the potential drug-disease association and generates a prediction probability list. The resulting probability list can be viewed as a weighted bipartite network, in which one group represents drugs and the other group represents diseases. In this network, there is a weighted edge between each pair of nodes, and the weight reflects the degree of association between the drug and the disease. The existence of these weighted edges conveys the probability of interaction between drug small molecules and disease modules, and provides detailed related information, which helps to deeply understand the interaction patterns between drugs and diseases and their potential effects.
[0016] The formula (4-1) in step 4 adopts the path length calculation method, that is, calculating the distance between the drug target gene and the closest disease protein. All target genes of each drug correspond to a closest distance value. Finally, the average value of the closest distances corresponding to all target genes is calculated. The closer the calculation result is to 1, the closer the distance between the drug module and the disease module is.
[0017] Formula (4-2) in step 4 uses a calculation method that combines node degree and path length. Node degree refers to the number of connections between a node and other nodes, which is an important concept in complex network analysis. In a complex network, each node has a degree, which is the number of edges directly connected to the node. The degree of a node can help understand the importance and connection of nodes in the network. By analyzing the degree of nodes and the distribution of node degrees, as well as better understanding the structure and characteristics of complex networks, identifying key nodes in the network, and studying the robustness and robustness of the network, the calculation result of the node degree method used in formula (4-2) has a value range of [0, 2]. Therefore, the closer the calculated result is to 2, the closer the distance between the drug module and the disease module is.
[0018] The present invention improves the efficiency and accuracy of Chinese herbal medicine repositioning. By introducing data-driven and machine learning technologies, the efficiency and accuracy of Chinese herbal medicine repositioning are significantly improved. Through big data and machine learning technologies, accurate matching of Chinese herbal medicine ingredients and diseases is achieved, the need for manual intervention is reduced, and the analysis efficiency is improved. By collecting and integrating Chinese herbal medicine ingredients, target genes and disease information in multiple databases, a comprehensive data set is constructed, and the quality and consistency of the data are improved. The present invention adopts advanced data cleaning, preprocessing and integration methods to ensure the high quality and consistency of the data, providing a solid foundation for subsequent analysis. According to the multi-component and multi-target characteristics of Chinese herbal medicine, a special prediction algorithm is developed, which significantly improves the prediction ability of the model. Through the integrated learning model, the advantages of multiple machine learning algorithms are combined to further improve the accuracy of the prediction. The model of the present invention has high probability and reliability in predicting the effect of drugs on diseases, and provides a scientific basis for drug repositioning. The AUC value of the model of the present invention reaches 0.923, which shows its high performance and reliability in predicting the relationship between Chinese herbal medicine and disease. The model is verified on multiple independent data sets and shows high consistency and robustness. In summary, the present invention significantly improves the efficiency and accuracy of Chinese herbal medicine repositioning by introducing advanced data-driven and machine learning technologies, optimizes data integration and processing capabilities, improves prediction models and algorithms, provides scientific drug recommendations, and ensures its reliability and effectiveness in practical applications through strict model performance verification.
[0019] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A data-driven Chinese herbal medicine repositioning method, characterized in that: The following steps are involved: Step 1: Data collection and processing: collect information on Chinese herbal medicine ingredients, target genes, and diseases from large public databases such as TCMSP, SymMap, and UniProt; Step 2: Data integration, including data cleaning, format unification, data removal, and data conversion, to ensure data consistency and integrity; Step 3: Feature construction, using multidimensional binary vectors to represent drug features, including chemical structure, protein domain, and gene ontology information, calculating the similarity between drugs and diseases through the Jaccard similarity coefficient, and constructing a similarity matrix; Step 4: Construct drug similarity network and disease similarity network, use the method of combining path length and node degree to calculate the distance between drug module and disease module, so as to quantify the interaction between drug and disease. Drug module and disease module are located at different positions in the network. Drug small molecules and disease genes represent the corresponding nodes in the network respectively. The distance between drug module and disease module is calculated by combining path length, node degree and path length to construct drug-disease network. The formula is as follows: （4-1）； （4-2）； Where d(T, S) is the distance between the drug target t in the drug module T and the disease gene s in the disease module S, G is the network; Dmax(G) refers to the diameter of the graph G, dis(t, s) refers to the shortest path length between nodes t and s, degt refers to the degree of node t, and degG refers to the maximum degree in the graph G; Step 5: Calculation of similarity of Chinese herbal medicines. In order to improve the relevance and accuracy of the calculation results, the model constructs a reference distance distribution, which simulates the expected distance between two groups of randomly selected proteins. The selection of these proteins is based on the similarity of connectivity with target disease-related or drug target proteins. The distance between the two groups of proteins is randomly calculated 1000 times and the distance values are stored. The size of the Z value is calculated by using the mean and standard deviation of the reference distance distribution. This randomization process allows the biological significance of the network connection between any pair of nodes to be retained. The formula is as follows: （5-1）； Where d(T, S) is the drug-disease distance value in formula (3.1) and formula (3.2), μd(T, S) is the mean, and σ_(d(T,S)) is the variance; The specific steps for calculating the Z value are roughly divided into the following five steps: (1) Sort all gene nodes by their connectivity, from high to low; (2) Gene nodes are randomly selected in the ranking, and the selection range includes the first 100 and last 100 nodes of each node; (3) Calculate the distance between the selected drug gene node and the disease gene node; (4) Steps (2) and (3) are repeated 1000 times and the results are stored in an array; (5) For each drug-disease pair, there are 1000 data points. Calculate the mean and variance, and get the Z value according to formula (5-1); Step 6: Cluster analysis, enhance the predictive ability of the model, using the ClusterONE clustering method. This step aims to strengthen the cohesion within the disease-related protein cluster. First, the similarity data between the drug module and the disease module obtained by formula (5-1) is constructed into a weighted graph, where each node represents a drug or disease, each edge represents a possible interaction relationship, and the weight of the edge represents the confidence of the interaction. Then, the network is preprocessed by specifying parameter settings, and a greedy strategy-based algorithm is used to search for high-quality clusters in the network, gradually expanding and merging nodes to form clusters. In this process, ClusterONE considers the connection density and mutual connectivity between nodes to determine the optimal clustering structure. Finally, the score f(v) of each disease module is calculated by evaluating the quality of the clustering, which represents the cohesion of the module. The formula is as follows: （6-1）； Among them, Win(v) represents the total weight of all "edges" in the disease module, Wout(v) represents the total weight of the edges connecting the disease module with the rest of the network, and P(v) is the penalty term, that is, the node density. When the drug and the disease are in the same cluster, this model will adjust its similarity metric to more accurately reflect the potential connection between the drug and the disease. Through improvement, the model not only improves the cohesion of disease-related clusters, but also enhances the accuracy of identifying potential drug therapeutic effects, thereby improving the accuracy of drug-disease association. The formula for adjusting the drug-disease similarity metric is as follows: （6-2）； Among them, sim' is the adjusted similarity; Finally, in order to quantify the probability of association between each drug and disease, this model uses the sigmoid function. The sigmoid function is a commonly used mathematical function, which is usually used to map input values to a range between 0 and 1. The image of the sigmoid function presents an S-shaped curve, which is characterized by that when the x-axis approaches positive infinity, the y-axis approaches 1, and when the x-axis approaches negative infinity, the y-axis approaches 0. This function shape makes it often used to convert input values of any range into probability values. This method changes the parameters of the sigmoid function, and the formula for calculating the probability of association between each drug and disease is as follows: （6-3）； Where x is the drug-disease similarity value, b is the midpoint of the sigmoid function curve, the value of b is half of the maximum value of x, and c is the steepness of the sigmoid function curve, c=10; Step 7: Repositioning of Chinese herbal medicines. The model predicts the potential drug-disease association and generates a prediction probability list. The resulting probability list can be viewed as a weighted bipartite network, in which one group represents drugs and the other group represents diseases. In this network, there is a weighted edge between each pair of nodes, and the weight reflects the degree of association between the drug and the disease. The existence of these weighted edges conveys the probability of interaction between drug small molecules and disease modules, and provides detailed related information, which helps to deeply understand the interaction patterns between drugs and diseases and their potential effects.
2. A data-driven Chinese herbal medicine repositioning method according to claim 1, characterized in that: The formula (4-1) in step 4 adopts the path length calculation method, that is, calculating the distance between the drug target gene and the closest disease protein. All target genes of each drug correspond to a closest distance value. Finally, the average value of the closest distances corresponding to all target genes is calculated. The closer the calculation result is to 1, the closer the distance between the drug module and the disease module is.
3. A data-driven Chinese herbal medicine repositioning method according to claim 1, characterized in that: Formula (4-2) in step 4 uses a calculation method that combines node degree and path length. Node degree refers to the number of connections between a node and other nodes, which is an important concept in complex network analysis. In a complex network, each node has a degree, which is the number of edges directly connected to the node. The degree of a node can help understand the importance and connection of nodes in the network. By analyzing the degree of nodes and the distribution of node degrees, as well as better understanding the structure and characteristics of complex networks, identifying key nodes in the network, and studying the robustness and robustness of the network, the calculation result of the node degree method used in formula (4-2) has a value range of [0, 2]. Therefore, the closer the calculated result is to 2, the closer the distance between the drug module and the disease module is.