A self-collaboration method and system applied to component traditional Chinese medicine design

Abstract

The application belongs to the technical field of intelligent design of traditional Chinese medicine, and provides a self-collaboration method and system applied to component traditional Chinese medicine design; the application collects multi-dimensional component traditional Chinese medicine data through a data collection layer, pre-processes a multi-dimensional component traditional Chinese medicine data set through a data processing layer, and performs correlation analysis, stores the data processed by the data processing layer in a distributed manner through a data storage layer, constructs component traditional Chinese medicine design multi-agent through a model optimization layer, and evaluates and adjusts the agent, then optimizes the prediction result based on a collaboration mechanism through a collaboration generation layer to generate the final prediction result, and finally performs data interaction on the prediction result through a user interaction layer to generate a visual interface; the application realizes full-process support through self-collaboration processing, and is helpful for efficient research and development of component traditional Chinese medicine.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent design of traditional Chinese medicine, and specifically relates to a self-collaborative method and system for component traditional Chinese medicine design. Background Art

[0002] Component TCM is a modern TCM developed under the guidance of TCM theory, based on the theory and principles of prescription compatibility, and combined with modern drug development methods and technologies, through the rational combination of active ingredients. This approach inherits the classical theories of TCM while incorporating scientific and innovative thinking. It is an important means to promote the modernization of TCM and one of the core paths for innovative TCM research and development. However, the research and development process of component TCM is extremely complex due to the huge amount of TCM data, the intricate combination of active ingredients and mechanisms of action, and the involvement of multidisciplinary knowledge such as chemistry, pharmacology, and bioinformatics. This makes the drug design of component TCM very complex, making it difficult to achieve efficient efficacy requirements when processing complex data, posing a huge challenge to the design and development of component TCM.

[0003] First, the data involved in component TCMs are of various types and come from a wide range of sources, including component data, efficacy target data, pharmacological activity data, etc. In addition, these data sources include high-throughput screening, literature data, clinical data, and experimental data. The sources and formats of the data vary greatly, which brings difficulties to data cleaning, standardization, and analysis, affecting subsequent analysis and decision-making. Secondly, the current drug design of component TCMs lacks intelligent means. Traditional component TCM design mostly relies on the professional knowledge and experience accumulation of researchers, and lacks a systematic algorithm model to formulate component TCM drug design strategies. When faced with complex component-target-disease interactions, traditional methods often find it difficult to quickly and comprehensively evaluate the effects of component combinations. This leads to low design efficiency and difficulty in accurately predicting drug efficacy, increasing the cost and time of drug development. Summary of the Invention

[0004] The present invention aims to address the problems of the existing technology and provides a self-collaborative processing method and system for component Chinese medicine design through data collection, processing, modeling, adjustment, user interaction and other links, which realizes the full process support of Chinese medicine component design from data collection to compatibility combination to result display, which is conducive to the efficient research and development of component Chinese medicine, can be widely used in the research and design of innovative drugs of component Chinese medicine, and reduce the cost and time of drug development.

[0005] In order to achieve the above technical objectives, the present invention adopts the following technical solutions:

[0006] The present invention provides a self-collaborative method for designing component traditional Chinese medicines, comprising the following steps:

[0007] S1. Collect multi-dimensional component Chinese medicine data;

[0008] S2. Preprocess the multi-dimensional component Chinese medicine dataset, perform correlation analysis on the preprocessed data, extract the features of the multi-dimensional component Chinese medicine dataset, and obtain the negative and positive efficacy indicators of the component Chinese medicines corresponding to the features;

[0009] S3, using the distributed storage system HDFS to distribute and securely manage the pre-processed data sets and the negative and positive efficacy indicators of the component Chinese medicines;

[0010] S4. Design a component Chinese medicine design multi-agent, obtain data stored in the distributed storage system HDFS, and input it into the component Chinese medicine design multi-agent. Combine the deep learning algorithm and the reinforcement learning algorithm to train the component Chinese medicine design multi-agent, and obtain a component Chinese medicine pharmacological substance prediction agent, a component Chinese medicine structure optimization agent, and a pharmacological substance compatibility prediction agent respectively. Multiple agents together constitute the trained component Chinese medicine design multi-agent;

[0011] S5. The user inputs the target Chinese medicine into the trained component Chinese medicine design multi-agent to generate prediction results of the component Chinese medicine active substance prediction agent, the component Chinese medicine structure optimization agent, and the active substance compatibility prediction agent. Based on the collaborative mechanism, the different prediction results are optimized to generate the final prediction result.

[0012] Preferably, in step S1, the multi-dimensional component Chinese medicine data includes Chinese medicine ingredient information, target database information, pharmacological property data, literature data, experimental data, chemical data of component Chinese medicine, biological data of component Chinese medicine and empirical data reported in literature.

[0013] Preferably, step S2 specifically includes:

[0014] S201, preprocessing the multi-dimensional component traditional Chinese medicine dataset, specifically cleaning the dataset, removing missing values, removing outliers, deduplication, and eliminating noise and redundant information in the dataset;

[0015] S202, converting data from different sources in the pre-processed data set into a unified format and unit, and performing standardization processing;

[0016] S203. Perform correlation analysis on the standardized data set, extract the characteristics of the multi-dimensional component traditional Chinese medicine data set, and obtain the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the characteristics.

[0017] Preferably, step S3 specifically includes:

[0018] S301, using the distributed storage system HDFS to store the preprocessed multi-dimensional component Chinese medicine dataset and the negative and positive efficacy indicators of the component Chinese medicines, and dividing the stored multi-dimensional component Chinese medicine dataset and the negative and positive efficacy indicators of the component Chinese medicines into semi-structured data, structured data, and data that needs to be archived in the distributed storage system HDFS;

[0019] S302. Use Apache Hive or Presto to index and query the semi-structured data and structured data stored in the distributed storage system HDFS, use Spark or MapReduce to extract the data that needs to be archived from the distributed storage system HDFS, and use MongoDB to create table structures and indexes for the extracted data that needs to be archived;

[0020] S303: Set user access control for data stored in the distributed storage system HDFS, build user security accounts, and perform security management.

[0021] Preferably, step S4 specifically includes:

[0022] S401. Establish multiple AI-based component TCM agents, including a component TCM active substance prediction agent, a component TCM structure optimization agent, and an active substance compatibility prediction agent. Each agent undertakes different design tasks, forming a component TCM design multi-agent.

[0023] S402, obtaining the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the features in the distributed storage system HDFS, performing binary classification through one-hot processing, labeling the negative and positive efficacy as "0" and "1" respectively, and randomly dividing the processed multi-dimensional component traditional Chinese medicine dataset into a training set and a validation set at an 8:1 ratio;

[0024] S402, using the negative and positive efficacy indicators of the component Chinese medicine corresponding to the labeled features as a training set and training with a deep learning algorithm to generate an intelligent agent for predicting the efficacy of the component Chinese medicine;

[0025] S403. Use the validation set to evaluate the component TCM active substance prediction intelligent agent. The evaluation methods are accuracy and binary classification ROC-AUC. Adjust the structure of the component TCM active substance prediction intelligent agent based on the evaluation results.

[0026] Preferably, step S5 specifically includes:

[0027] S501. The user inputs the target component Chinese medicine into the component Chinese medicine active substance prediction intelligent agent to generate a prediction result;

[0028] S502: Based on the prediction result generated in step S501, the particle swarm strategy is used for optimization based on the collaborative mechanism, and adaptive adjustment is performed based on the feedback mechanism to generate the final prediction result.

[0029] Preferably, the method further includes performing data interaction on the final prediction result generated in step S5 to generate a visual interface.

[0030] The present invention also provides a self-collaborative system for component Chinese medicine design, which uses the self-collaborative method for component Chinese medicine design as described above, including a data acquisition layer, a data processing layer, a data storage layer, a model optimization layer, a collaborative generation layer, and a user interaction layer;

[0031] The data collection layer is used to collect multi-dimensional component traditional Chinese medicine data;

[0032] The data processing layer is used to preprocess the multi-dimensional component traditional Chinese medicine dataset generated by the data acquisition layer and perform correlation analysis on the preprocessed data;

[0033] The data storage layer is used to perform distributed storage of the data processed by the data processing layer;

[0034] The model optimization layer is used to design multiple component Chinese medicine design agents, form a component Chinese medicine design multi-agent, obtain data stored in the data storage layer, train the component Chinese medicine efficacy substance prediction agent, and evaluate and adjust the agent;

[0035] The collaborative generation layer is used to optimize the prediction results based on the collaborative mechanism and generate the final prediction results;

[0036] The user interaction layer is used to interact with the final prediction results and generate a visual interface.

[0037] Preferably, the model optimization layer performs the following:

[0038] A1. Establish multiple AI-based components of Chinese medicine (TCM) agents, including a component TCM active substance prediction agent, a component TCM structure optimization agent, and a component TCM active substance compatibility prediction agent. Each agent will be responsible for specific tasks, forming a multi-agent component TCM design system.

[0039] A2. Obtain data stored in the data storage layer and train it using a deep learning algorithm to obtain an intelligent agent for predicting the efficacy of Chinese herbal medicine components.

[0040] A3. Obtain the data stored in the data storage layer and use the reinforcement algorithm for training to obtain the component Chinese medicine efficacy structure optimization intelligent agent;

[0041] A4. Obtain data stored in the data storage layer and train it using a deep learning algorithm to obtain an intelligent agent for predicting the compatibility of medicinal substances.

[0042] A5. Evaluate the intelligent agent for predicting the effective substances of component traditional Chinese medicines and adjust the structure of the multi-agent design of component traditional Chinese medicines.

[0043] Preferably, the collaborative generation layer performs the following:

[0044] B1. Obtain the target component Chinese medicine and generate prediction results;

[0045] B2. Obtain the prediction results. Based on the collaborative mechanism, the particle swarm strategy is used for optimization to achieve data sharing among multiple component Chinese medicine agents, and adaptive adjustments are made based on the feedback mechanism to generate the final prediction results.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] (1) The present invention performs self-collaborative processing through data collection, processing, modeling, adjustment, user interaction and other links, and realizes the full process support of Chinese medicine component design from data collection to compatibility combination to result display, which is conducive to the efficient research and development of component Chinese medicines. It can be widely used in the research and design of innovative drugs based on component Chinese medicines, reducing the cost and time of drug development, and providing a new tool for the research and development of modern Chinese medicines;

[0048] (2) The present invention significantly improves the efficiency and quality of drug design by introducing multi-agents. Compared with the traditional manual design method that relies on expert experience, the present invention can automatically and accurately complete the component design of new Chinese medicines, reducing R&D time and improving the scientificity and reliability of the design.

[0049] (3) The present invention is based on a collaborative mechanism and adopts a particle swarm strategy, which enables each intelligent agent to share information in real time during the design process and adjust strategies with each other, continuously designing solutions. The collaborative mechanism gives the system a high degree of flexibility, ensuring that the design results meet the efficacy requirements and adapt to different application scenarios;

[0050] (4) The present invention provides intelligent support for the entire process from data collection, data processing and analysis to data storage, and multi-level data processing capabilities to ensure data quality and availability, laying the data foundation for subsequent intelligent agent training, adjustment, and maintenance.

[0051] (5) The present invention uses a multi-agent and collaborative mechanism to enable each agent to make collaborative adjustments based on shared information and respond to feedback data in drug design in real time. The resulting prediction results can achieve personalized and precise efficacy in the design of component traditional Chinese medicines, meeting the needs of personalized treatment.

[0052] (6) This invention combines deep learning, multi-agents, and collaborative mechanisms. Through multidisciplinary cross-disciplinary technologies, it improves the scientificity and accuracy of component Chinese medicine design. It also provides new ideas and methods for the future innovative development of Chinese medicine, which is conducive to promoting the modernization of Chinese medicine.

[0053] (7) The present invention significantly reduces the repeated experiments and manual adjustments in the development of traditional Chinese medicine through intelligent and automated design processes, shortens the R&D cycle, and reduces the cost of drug development, thus having significant economic benefits and application promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of a self-collaborative method for component traditional Chinese medicine design according to an embodiment of the present invention;

[0055] Figure 2 Schematic diagram of a self-cooperation system applied to component traditional Chinese medicine design according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0057] Example 1

[0058] Combine Figure 1 As shown, the embodiment of the present invention provides a self-collaborative method for component traditional Chinese medicine design, comprising the following steps:

[0059] S1. Collect multi-dimensional component Chinese medicine data;

[0060] S2. Preprocess the multi-dimensional component Chinese medicine dataset, perform correlation analysis on the preprocessed data, extract the features of the multi-dimensional component Chinese medicine dataset, and obtain the negative and positive efficacy indicators of the component Chinese medicines corresponding to the features;

[0061] S3, using the distributed storage system HDFS to distribute and securely manage the pre-processed data sets and the negative and positive efficacy indicators of the component Chinese medicines;

[0062] S4. Design a component Chinese medicine design multi-agent, obtain data stored in the distributed storage system HDFS, and input it into the component Chinese medicine design multi-agent. Combine the deep learning algorithm and the reinforcement learning algorithm to train the component Chinese medicine design multi-agent, and obtain a component Chinese medicine pharmacological substance prediction agent, a component Chinese medicine structure optimization agent, and a pharmacological substance compatibility prediction agent respectively. Multiple agents together constitute the trained component Chinese medicine design multi-agent;

[0063] S5. The user inputs the target Chinese medicine into the trained component Chinese medicine design multi-agent to generate prediction results of the component Chinese medicine active substance prediction agent, the component Chinese medicine structure optimization agent, and the active substance compatibility prediction agent. Based on the collaborative mechanism, the different prediction results are optimized to generate the final prediction result.

[0064] Example 2

[0065] Combine Figure 1 As shown, the embodiment of the present invention provides a self-collaborative method for component traditional Chinese medicine design, comprising the following steps:

[0066] S1. Collect multi-dimensional component Chinese medicine data;

[0067] S2. Preprocess the multi-dimensional component Chinese medicine dataset, perform correlation analysis on the preprocessed data, extract the features of the multi-dimensional component Chinese medicine dataset, and obtain the negative and positive efficacy indicators of the component Chinese medicines corresponding to the features;

[0068] S3, using the distributed storage system HDFS to distribute and securely manage the pre-processed data sets and the negative and positive efficacy indicators of the component Chinese medicines;

[0069] S4. Design a component Chinese medicine design multi-agent, obtain data stored in the distributed storage system HDFS, and input it into the component Chinese medicine design multi-agent. Combine the deep learning algorithm and the reinforcement learning algorithm to train the component Chinese medicine design multi-agent, and obtain a component Chinese medicine pharmacological substance prediction agent, a component Chinese medicine structure optimization agent, and a pharmacological substance compatibility prediction agent respectively. Multiple agents together constitute the trained component Chinese medicine design multi-agent;

[0070] S5. The user inputs the target Chinese medicine into the trained component Chinese medicine design multi-agent, which generates prediction results for the component Chinese medicine active substance prediction agent, the component Chinese medicine structure optimization agent, and the active substance compatibility prediction agent. Based on the collaborative mechanism, the different prediction results are optimized to generate the final prediction result.

[0071] On this basis, in this embodiment, in step S1, the multi-dimensional component Chinese medicine data includes Chinese medicine ingredient information, target database information, pharmacological property data, literature data, experimental data, chemical data of component Chinese medicine, biological data of component Chinese medicine and empirical data reported in literature;

[0072] Furthermore, step S2 specifically includes:

[0073] S201, preprocessing the multi-dimensional component traditional Chinese medicine dataset, specifically cleaning the dataset, removing missing values, removing outliers, deduplication, and eliminating noise and redundant information in the dataset;

[0074] S202, converting data from different sources in the pre-processed data set into a unified format and unit, and performing standardization processing;

[0075] S203. Perform correlation analysis on the standardized data set, extract the characteristics of the multi-dimensional component traditional Chinese medicine data set, and obtain the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the characteristics.

[0076] Example 3

[0077] Combine Figure 1 As shown, the embodiment of the present invention provides a self-collaborative method for component traditional Chinese medicine design, comprising the following steps:

[0078] S1. Collect multi-dimensional component Chinese medicine data;

[0079] S2. Preprocess the multi-dimensional component Chinese medicine dataset, perform correlation analysis on the preprocessed data, extract the features of the multi-dimensional component Chinese medicine dataset, and obtain the negative and positive efficacy indicators of the component Chinese medicines corresponding to the features;

[0080] S3, using the distributed storage system HDFS to distribute and securely manage the pre-processed data sets and the negative and positive efficacy indicators of the component Chinese medicines;

[0081] S4. Design a component Chinese medicine design multi-agent, obtain data stored in the distributed storage system HDFS, and input it into the component Chinese medicine design multi-agent. Combine the deep learning algorithm and the reinforcement learning algorithm to train the component Chinese medicine design multi-agent, and obtain a component Chinese medicine pharmacological substance prediction agent, a component Chinese medicine structure optimization agent, and a pharmacological substance compatibility prediction agent respectively. Multiple agents together constitute the trained component Chinese medicine design multi-agent;

[0082] S5. The user inputs the target Chinese medicine into the trained component Chinese medicine design multi-agent to generate prediction results of the component Chinese medicine active substance prediction agent, the component Chinese medicine structure optimization agent, and the active substance compatibility prediction agent. Based on the collaborative mechanism, the different prediction results are optimized to generate the final prediction result.

[0083] In step S1, the multi-dimensional component Chinese medicine data includes Chinese medicine ingredient information, target database information, pharmacological property data, literature data, experimental data, chemical data of component Chinese medicine, biological data of component Chinese medicine and empirical data reported in literature;

[0084] Step S2 specifically includes:

[0085] S201, preprocessing the multi-dimensional component traditional Chinese medicine dataset, specifically cleaning the dataset, removing missing values, removing outliers, deduplication, and eliminating noise and redundant information in the dataset;

[0086] S202, converting data from different sources in the pre-processed data set into a unified format and unit, and performing standardization processing;

[0087] S203, performing correlation analysis on the standardized data set, extracting the characteristics of the multi-dimensional component traditional Chinese medicine data set, and obtaining the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the characteristics;

[0088] On this basis, in this embodiment, step S3 specifically includes:

[0089] S301, using the distributed storage system HDFS to store the preprocessed multi-dimensional component Chinese medicine dataset and the negative and positive efficacy indicators of the component Chinese medicines, and dividing the stored multi-dimensional component Chinese medicine dataset and the negative and positive efficacy indicators of the component Chinese medicines into semi-structured data, structured data, and data that needs to be archived in the distributed storage system HDFS;

[0090] S302. Use Apache Hive or Presto to index and query the semi-structured data and structured data stored in the distributed storage system HDFS, use Spark or MapReduce to extract the data that needs to be archived from the distributed storage system HDFS, and use MongoDB to create table structures and indexes for the extracted data that needs to be archived;

[0091] S303. Set user access control for data stored in the distributed storage system HDFS, build user security accounts, and perform security management to support fast access, query, and management of data, support efficient storage and reading operations, and ensure data security, access speed, and ease of use and stability in subsequent analysis.

[0092] Example 4

[0093] Combine Figure 1 As shown, the embodiment of the present invention provides a self-collaborative method for component traditional Chinese medicine design, comprising the following steps:

[0094] S1. Collect multi-dimensional component Chinese medicine data;

[0095] S2. Preprocess the multi-dimensional component Chinese medicine dataset, perform correlation analysis on the preprocessed data, extract the features of the multi-dimensional component Chinese medicine dataset, and obtain the negative and positive efficacy indicators of the component Chinese medicines corresponding to the features;

[0096] S3, using the distributed storage system HDFS to distribute and securely manage the pre-processed data sets and the negative and positive efficacy indicators of the component Chinese medicines;

[0097] S4. Design a component Chinese medicine design multi-agent, obtain data stored in the distributed storage system HDFS, and input it into the component Chinese medicine design multi-agent. Combine the deep learning algorithm and the reinforcement learning algorithm to train the component Chinese medicine design multi-agent, and obtain a component Chinese medicine pharmacological substance prediction agent, a component Chinese medicine structure optimization agent, and a pharmacological substance compatibility prediction agent respectively. Multiple agents together constitute the trained component Chinese medicine design multi-agent;

[0098] S5. The user inputs the target Chinese medicine into the trained component Chinese medicine design multi-agent to generate prediction results of the component Chinese medicine active substance prediction agent, the component Chinese medicine structure optimization agent, and the active substance compatibility prediction agent. Based on the collaborative mechanism, the different prediction results are optimized to generate the final prediction result.

[0099] In step S1, the multi-dimensional component Chinese medicine data includes Chinese medicine ingredient information, target database information, pharmacological property data, literature data, experimental data, chemical data of component Chinese medicine, biological data of component Chinese medicine and empirical data reported in literature;

[0100] Step S2 specifically includes:

[0101] S201, preprocessing the multi-dimensional component traditional Chinese medicine dataset, specifically cleaning the dataset, removing missing values, removing outliers, deduplication, and eliminating noise and redundant information in the dataset;

[0102] S202, converting data from different sources in the pre-processed data set into a unified format and unit, and performing standardization processing;

[0103] S203, performing correlation analysis on the standardized data set, extracting the characteristics of the multi-dimensional component traditional Chinese medicine data set, and obtaining the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the characteristics;

[0104] On this basis, in this embodiment, step S4 specifically includes:

[0105] S401. Establish multiple AI-based component TCM agents, including a component TCM active substance prediction agent, a component TCM structure optimization agent, and an active substance compatibility prediction agent. Each agent undertakes different design tasks, forming a component TCM design multi-agent.

[0106] S402, obtaining the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the features in the distributed storage system HDFS, performing binary classification through one-hot processing, labeling the negative and positive efficacy as "0" and "1" respectively, and randomly dividing the processed multi-dimensional component traditional Chinese medicine dataset into a training set and a validation set at an 8:1 ratio;

[0107] S403, using the negative and positive efficacy indicators of the component Chinese medicine corresponding to the labeled features as a training set and training with a deep learning algorithm to generate a component Chinese medicine efficacy substance prediction intelligent agent;

[0108] S404. Use the validation set to evaluate the component TCM pharmacological substance prediction agent. The evaluation methods are accuracy and binary classification ROC-AUC. Adjust the structure of the component TCM pharmacological substance prediction agent based on the evaluation results. Specifically, ROC-AUC>0.9 indicates that the agent has excellent prediction ability and can predict subsequent component TCM pharmacological substances.

[0109] Example 5

[0110] Combine Figure 1 As shown, the embodiment of the present invention provides a self-collaborative method for component traditional Chinese medicine design, comprising the following steps:

[0111] S1. Collect multi-dimensional component Chinese medicine data;

[0112] S2. Preprocess the multi-dimensional component Chinese medicine dataset, perform correlation analysis on the preprocessed data, extract the features of the multi-dimensional component Chinese medicine dataset, and obtain the negative and positive efficacy indicators of the component Chinese medicines corresponding to the features;

[0113] S3, using the distributed storage system HDFS to distribute and securely manage the pre-processed data sets and the negative and positive efficacy indicators of the component Chinese medicines;

[0114] S4. Design a component Chinese medicine design multi-agent, obtain data stored in the distributed storage system HDFS, and input it into the component Chinese medicine design multi-agent. Combine the deep learning algorithm and the reinforcement learning algorithm to train the component Chinese medicine design multi-agent, and obtain a component Chinese medicine pharmacological substance prediction agent, a component Chinese medicine structure optimization agent, and a pharmacological substance compatibility prediction agent respectively. Multiple agents together constitute the trained component Chinese medicine design multi-agent;

[0115] S5. The user inputs the target Chinese medicine into the trained component Chinese medicine design multi-agent to generate prediction results of the component Chinese medicine active substance prediction agent, the component Chinese medicine structure optimization agent, and the active substance compatibility prediction agent. Based on the collaborative mechanism, the different prediction results are optimized to generate the final prediction result.

[0116] In step S1, the multi-dimensional component Chinese medicine data includes Chinese medicine ingredient information, target database information, pharmacological property data, literature data, experimental data, chemical data of component Chinese medicine, biological data of component Chinese medicine and empirical data reported in literature;

[0117] Step S2 specifically includes:

[0118] S201, preprocessing the multi-dimensional component traditional Chinese medicine dataset, specifically cleaning the dataset, removing missing values, removing outliers, deduplication, and eliminating noise and redundant information in the dataset;

[0119] S202, converting data from different sources in the pre-processed data set into a unified format and unit, and performing standardization processing;

[0120] S203, performing correlation analysis on the standardized data set, extracting the characteristics of the multi-dimensional component traditional Chinese medicine data set, and obtaining the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the characteristics;

[0121] On this basis, in this embodiment, step S5 specifically includes:

[0122] S501: The user inputs the target component Chinese medicine into the adjusted component Chinese medicine active substance prediction agent to generate a prediction result;

[0123] S502. Based on the prediction results generated in step S501, the particle swarm strategy is used for optimization based on the collaborative mechanism, and adaptive adjustment is performed based on the feedback mechanism to generate the final prediction results. Through the collaborative mechanism, each intelligent agent makes coordinated adjustments based on shared information, responds to the feedback data in drug design in real time, and finally generates the final prediction results.

[0124] Furthermore, it also includes data interaction for the prediction results generated in step S5 to generate a visualization interface; specifically, first, the prediction results are used to generate a visualization interface through the user interaction layer to display the pharmacological substance screening results, compatibility combination prediction results and component Chinese medicine structure results of the component Chinese medicine, providing intuitive graphical support; then, when the user adjusts the component Chinese medicine design parameters of the prediction results through the visualization interface, the system records the component Chinese medicine design parameters adjusted by the user in real time, and inputs the component Chinese medicine design parameters adjusted by the user into the intelligent body for iteration to generate a personalized component Chinese medicine design plan that meets the user's needs.

[0125] Example 6

[0126] Combine Figure 1 As shown, the embodiment of the present invention provides a self-collaborative method for component traditional Chinese medicine design, comprising the following steps:

[0127] S1. Collect multi-dimensional component Chinese medicine data;

[0128] S2. Preprocess the multi-dimensional component Chinese medicine dataset, perform correlation analysis on the preprocessed data, extract the features of the multi-dimensional component Chinese medicine dataset, and obtain the negative and positive efficacy indicators of the component Chinese medicines corresponding to the features;

[0129] S3, using the distributed storage system HDFS to distribute and securely manage the pre-processed data sets and the negative and positive efficacy indicators of the component Chinese medicines;

[0130] S4. Design a component Chinese medicine design multi-agent, obtain data stored in the distributed storage system HDFS, and input it into the component Chinese medicine design multi-agent. Combine the deep learning algorithm and the reinforcement learning algorithm to train the component Chinese medicine design multi-agent, and obtain a component Chinese medicine pharmacological substance prediction agent, a component Chinese medicine structure optimization agent, and a pharmacological substance compatibility prediction agent respectively. Multiple agents together constitute the trained component Chinese medicine design multi-agent;

[0131] S5. The user inputs the target Chinese medicine into the trained component Chinese medicine design multi-agent, which generates prediction results for the component Chinese medicine active substance prediction agent, the component Chinese medicine structure optimization agent, and the active substance compatibility prediction agent. Based on the collaborative mechanism, the different prediction results are optimized to generate the final prediction result.

[0132] Furthermore, in step S1, the multi-dimensional component Chinese medicine data includes Chinese medicine ingredient information, target database information, pharmacological property data, literature data, experimental data, chemical data of component Chinese medicines, biological data of component Chinese medicines and empirical data reported in literature;

[0133] Furthermore, step S2 specifically includes:

[0134] S201, preprocessing the multi-dimensional component traditional Chinese medicine dataset, specifically cleaning the dataset, removing missing values, removing outliers, deduplication, and eliminating noise and redundant information in the dataset;

[0135] S202, converting data from different sources in the pre-processed data set into a unified format and unit, and performing standardization processing;

[0136] S203, performing correlation analysis on the standardized data set, extracting the characteristics of the multi-dimensional component traditional Chinese medicine data set, and obtaining the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the characteristics;

[0137] Furthermore, step S3 specifically includes:

[0138] S301, using the distributed storage system HDFS to store the preprocessed multi-dimensional component Chinese medicine dataset and the negative and positive efficacy indicators of the component Chinese medicines, and dividing the stored multi-dimensional component Chinese medicine dataset and the negative and positive efficacy indicators of the component Chinese medicines into semi-structured data, structured data, and data that needs to be archived in the distributed storage system HDFS;

[0139] S302. Use Apache Hive or Presto to index and query the semi-structured data and structured data stored in the distributed storage system HDFS, use Spark or MapReduce to extract the data that needs to be archived from the distributed storage system HDFS, and use MongoDB to create table structures and indexes for the extracted data that needs to be archived;

[0140] S303: Set user access control for data stored in the distributed storage system HDFS, build user security accounts, and perform security management;

[0141] Furthermore, step S4 specifically includes:

[0142] S401. Establish multiple AI-based component TCM agents, including a component TCM active substance prediction agent, a component TCM structure optimization agent, and an active substance compatibility prediction agent. Each agent undertakes different design tasks, forming a component TCM design multi-agent.

[0143] S402, obtaining the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the features in the distributed storage system HDFS, performing binary classification through one-hot processing, labeling the negative and positive efficacy as "0" and "1" respectively, and randomly dividing the processed multi-dimensional component traditional Chinese medicine dataset into a training set and a validation set at an 8:1 ratio;

[0144] S403, using the negative and positive efficacy indicators of the component Chinese medicine corresponding to the labeled features as a training set and training with a deep learning algorithm to generate a component Chinese medicine efficacy substance prediction intelligent agent;

[0145] S404. Evaluate the intelligent agent for predicting the effective substances of Chinese herbal medicine components using the validation set. The evaluation methods are accuracy and binary classification ROC-AUC. Adjust the structure of the intelligent agent for predicting the effective substances of Chinese herbal medicine components based on the evaluation results.

[0146] Furthermore, step S5 specifically includes:

[0147] S501: The user inputs the target component Chinese medicine into the adjusted intelligent agent to generate a prediction result;

[0148] S502: Based on the prediction result generated in step S501, the particle swarm strategy is used for optimization based on the collaborative mechanism, and adaptive adjustment is performed based on the feedback mechanism to generate the final prediction result.

[0149] Example 7

[0150] Combine Figure 2 As shown, an embodiment of the present invention further provides a self-collaborative system for component Chinese medicine design, which uses the self-collaborative method for component Chinese medicine design as described above, including a data acquisition layer, a data processing layer, a data storage layer, a model optimization layer, a collaborative generation layer, and a user interaction layer;

[0151] The data collection layer is used to collect multi-dimensional component traditional Chinese medicine data;

[0152] The data processing layer is used to preprocess the multi-dimensional component traditional Chinese medicine dataset generated by the data acquisition layer and perform correlation analysis on the preprocessed data;

[0153] The data storage layer is used to perform distributed storage of the data processed by the data processing layer;

[0154] The model optimization layer is used to construct a multi-agent design of component Chinese medicines, intelligently analyze the structure, combination, and dosage of component Chinese medicines, obtain data stored in the data storage layer, construct an agent for predicting the efficacy of the component Chinese medicine's active substances, and evaluate and adjust the agent;

[0155] The collaborative generation layer is used to optimize the agent prediction results based on the collaborative mechanism and generate the final prediction results;

[0156] The user interaction layer is used to interact with the final prediction results and generate a visual interface.

[0157] Example 8

[0158] An embodiment of the present invention further provides a self-collaborative system for component Chinese medicine design, which uses the self-collaborative method for component Chinese medicine design as described above, including a data acquisition layer, a data processing layer, a data storage layer, a model optimization layer, a collaborative generation layer, and a user interaction layer;

[0159] The data collection layer is used to collect multi-dimensional data on component Chinese medicines. The data collection layer can collect multi-dimensional data on component Chinese medicines from various sources and integrate them to ensure the comprehensiveness, accuracy, timeliness and relevance of the data.

[0160] Specifically, the data collection layer includes data collection devices and big data mining devices deployed in the laboratory. The data collection devices include microplate readers, high-content imaging equipment, real-time fluorescence detection equipment, etc., which can collect biological activity data, pharmacological parameters, component detection results, etc. of component traditional Chinese medicines in various experimental links; the big data mining device is responsible for extracting information related to component traditional Chinese medicines from various big data sources such as existing literature databases, genome databases, and drug target databases, and aggregating and integrating data information from different sources. All collected data will be uploaded to the data processing layer in real time;

[0161] The data processing layer is used to preprocess the multi-dimensional component traditional Chinese medicine dataset generated by the data acquisition layer and perform correlation analysis on the preprocessed data; the data processing layer can ensure the consistency and accuracy of the data;

[0162] Specifically, the data processing layer includes a data cleaning unit, a data standardization unit, and a data analysis unit. The data cleaning unit performs denoising and deduplication on the data collected from the data acquisition layer. The unit can automatically identify and remove noise, outliers, and redundant information in the data to ensure the integrity and accuracy of the data, so as to ensure the reliability of subsequent analysis results; the data standardization unit converts data from different sources into a unified format and unit to facilitate subsequent cross-data source analysis and ensure the compatibility of data between different modules. Standardization can significantly reduce analysis bias caused by inconsistent data sources; the data analysis unit conducts in-depth mining of the pre-processed data through big data analysis technology, and uses analysis methods such as correlation analysis, cluster analysis, and regression analysis to extract the key features and efficacy indicators of the component Chinese medicine dataset, which can provide accurate component Chinese medicine input data for the model optimization layer;

[0163] The data storage layer is used for distributed storage of data processed by the data processing layer. The data storage layer can store and manage pre-processed data securely and efficiently, ensuring that the data can be efficiently accessed, queried, and managed. It can also perform regular data backups at the data storage layer to prevent data loss and provide version control so that you can go back to a specific version of data when needed.

[0164] Specifically, the data storage layer includes a distributed storage unit, a database management unit, and a data security management unit. The distributed storage unit distributes data across multiple storage nodes to improve storage scalability and fault tolerance. The database management unit provides functions such as storage, query, and update of component Chinese medicine data. The data security management unit ensures the security of stored data during transmission, storage, and access through data encryption, identity authentication, permission control, and audit logs.

[0165] The model optimization layer is used to design multiple component Chinese medicine design agents, form a component Chinese medicine design multi-agent, obtain the data stored in the data storage layer, train the component Chinese medicine efficacy substance prediction agent, and evaluate and adjust the component Chinese medicine efficacy substance prediction agent;

[0166] Specifically, the model optimization layer includes a model training unit and a model evaluation unit; the model training unit constructs a multi-agent for component Chinese medicine design, intelligently performs structure, combination and dosage of component Chinese medicine, and trains the agent based on the key feature data and efficacy indicators of the component Chinese medicine data set stored in the data storage layer using deep learning, reinforcement learning algorithms or machine learning algorithms, and then automatically adjusts the model hyperparameters using grid search or Bayesian methods to obtain the best prediction performance; the model evaluation unit uses accuracy and binary classification ROC-AUC to evaluate and adjust the agent. In addition, evaluation indicators such as mean square error and determination coefficient accuracy can also be used to evaluate the agent to ensure that the agent has good generalization performance on the test set. In addition, cross-validation and adaptive validation methods are used to ensure the stability and accuracy of the agent;

[0167] The collaborative generation layer is used to optimize the prediction results of the component Chinese medicine design agent based on the collaborative mechanism and generate the final prediction results;

[0168] Specifically, the collaborative generation layer includes a component Chinese medicine result generation unit, a strategy generation unit and a knowledge reasoning unit; the user inputs the target component Chinese medicine into the component Chinese medicine result generation unit, and the component Chinese medicine result generation unit inputs the target component Chinese medicine into the multi-agent obtained in the model optimization layer. The component Chinese medicine result generation unit generates a component combination scheme with potential efficacy. By integrating traditional Chinese medicine theory and modern deep learning algorithms, the unit can intelligently recommend matching schemes for different components to ensure that the combination has high efficacy and safety; the strategy generation unit adopts a particle swarm strategy combined with the results generated by the component Chinese medicine result generation unit. In addition, the particle swarm strategy can also be replaced by a multi-objective algorithm to further generate a compatibility combination prediction strategy, a component Chinese medicine pharmacological substance screening result and a component Chinese medicine structure result to maximize the drug's potency and minimize side effects; the knowledge reasoning unit integrates the theoretical knowledge of Chinese medicine, modern pharmacology data and the results generated by the strategy generation unit to infer the potential mechanisms of different combinations, explain the scientificity and effectiveness of the compatibility combination, and provide a theoretical basis for generating the scheme;

[0169] The user interaction layer is used to interact with the prediction results and generate a visual interface.

[0170] Specifically, the user interaction layer includes a user interface unit, a parameter input unit and a real-time feedback unit; the user interface unit provides a web graphical interface for users to intuitively view and operate system functions, and the interface includes a data display panel, a prediction result viewing panel, a strategy recommendation panel, etc., which supports users to browse the drug efficacy combination scheme, model results and the working status of each module; the parameter input unit allows users to customize the input parameters according to actual needs, and enter their own preferences or restrictions in this unit. The system will perform personalized calculations and provide drug efficacy substance screening results, compatibility combination schemes and component Chinese medicine structure results that are more in line with the needs based on the parameters entered by the user; the real-time feedback unit is used to show the user the calculation and generation progress of the system at each stage, including data processing progress, model status, drug efficacy combination generation results, etc. Through real-time feedback, users can understand the system operation status at any time and adjust the input or selection parameters in time to obtain the final results;

[0171] On this basis, in this embodiment, the model optimization layer performs the following:

[0172] A1. Establish multiple AI-based TCM component agents, including a TCM component efficacy substance prediction agent, a TCM component structure optimization agent, and a TCM component compatibility agent. Each agent will undertake specific tasks, forming a multi-agent component optimization design agent.

[0173] A2. Obtain data stored in the data storage layer, use deep learning algorithms to train multiple agents, and generate a component Chinese medicine efficacy substance prediction agent;

[0174] A3. Obtain the data stored in the data storage layer and use the reinforcement algorithm for training to obtain the component Chinese medicine efficacy structure optimization intelligent agent;

[0175] A4. Obtain data stored in the data storage layer and train it using a deep learning algorithm to obtain an intelligent agent for predicting the compatibility of medicinal substances.

[0176] A5. Evaluate the intelligent agent for predicting the effective substances of component traditional Chinese medicines and adjust the structure of the multi-agent design of component traditional Chinese medicines.

[0177] Example 8

[0178] An embodiment of the present invention further provides a self-collaborative system for component Chinese medicine design, which uses the self-collaborative method for component Chinese medicine design as described above, including a data acquisition layer, a data processing layer, a data storage layer, a model optimization layer, a collaborative generation layer, and a user interaction layer;

[0179] The data collection layer is used to collect multi-dimensional component traditional Chinese medicine data;

[0180] The data processing layer is used to preprocess the multi-dimensional component traditional Chinese medicine dataset generated by the data acquisition layer and perform correlation analysis on the preprocessed data;

[0181] The data storage layer is used to perform distributed storage of the data processed by the data processing layer;

[0182] The model optimization layer is used to build a multi-agent design of component Chinese medicines, intelligently analyze the structure, combination, and dosage of component Chinese medicines, obtain data stored in the data storage layer, build a component Chinese medicine efficacy substance prediction agent, and evaluate and adjust the agent;

[0183] The collaborative generation layer is used to optimize the agent prediction results based on the collaborative mechanism and generate the final prediction results;

[0184] The user interaction layer is used to interact with the final prediction results and generate a visual interface;

[0185] On this basis, in this embodiment, the collaborative generation layer performs the following:

[0186] B1. Obtain the target component Chinese medicine and generate the intelligent agent prediction results;

[0187] B2. Obtain the prediction results of the intelligent agent, optimize it based on the collaborative mechanism, use the particle swarm strategy to achieve data sharing among multiple component Chinese medicine intelligent agents, and make adaptive adjustments based on the feedback mechanism to generate the final prediction results.

[0188] The above description is only an embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the scope of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A self-collaborative method for designing component traditional Chinese medicines, characterized in that: The following steps are involved: S1. Collect multi-dimensional component Chinese medicine data; S2. Preprocess the multi-dimensional component Chinese medicine dataset, perform correlation analysis on the preprocessed data, extract the features of the multi-dimensional component Chinese medicine dataset, and obtain the negative and positive efficacy indicators of the component Chinese medicines corresponding to the features; S3, using the distributed storage system HDFS to distribute and securely manage the pre-processed data sets and the negative and positive efficacy indicators of the component Chinese medicines; S4. Design a component Chinese medicine design multi-agent, obtain data stored in the distributed storage system HDFS, and input it into the component Chinese medicine design multi-agent. Combine the deep learning algorithm and the reinforcement learning algorithm to train the component Chinese medicine design multi-agent, and obtain a component Chinese medicine pharmacological substance prediction agent, a component Chinese medicine structure optimization agent, and a pharmacological substance compatibility prediction agent respectively. Multiple agents together constitute the trained component Chinese medicine design multi-agent; S5. The user inputs the target Chinese medicine into the trained component Chinese medicine design multi-agent to generate prediction results of the component Chinese medicine active substance prediction agent, the component Chinese medicine structure optimization agent, and the active substance compatibility prediction agent. Based on the collaborative mechanism, the different prediction results are optimized to generate the final prediction result.

2. The self-collaborative method for component Chinese medicine design according to claim 1, characterized in that: In step S1, the multi-dimensional component Chinese medicine data includes Chinese medicine ingredient information, target database information, pharmacological property data, literature data, experimental data, chemical data of component Chinese medicine, biological data of component Chinese medicine and empirical data reported in literature.

3. The self-collaborative method for component Chinese medicine design according to claim 2, characterized in that: Step S2 specifically includes: S201, preprocessing the multi-dimensional component traditional Chinese medicine dataset, specifically cleaning the dataset, removing missing values, removing outliers, deduplication, and eliminating noise and redundant information in the dataset; S202, converting data from different sources in the pre-processed data set into a unified format and unit, and performing standardization processing; S203. Perform correlation analysis on the standardized data set, extract the characteristics of the multi-dimensional component traditional Chinese medicine data set, and obtain the negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the characteristics.

4. The self-collaborative method for component Chinese medicine design according to claim 3, characterized in that: Step S3 specifically includes: S301, using the distributed storage system HDFS to store the preprocessed multi-dimensional component Chinese medicine dataset and the negative and positive efficacy indicators of the component Chinese medicines, and dividing the stored multi-dimensional component Chinese medicine dataset and the negative and positive efficacy indicators of the component Chinese medicines into semi-structured data, structured data, and data that needs to be archived in the distributed storage system HDFS; S302. Use Apache Hive or Presto to index and query the semi-structured data and structured data stored in the distributed storage system HDFS, use Spark or MapReduce to extract the data that needs to be archived from the distributed storage system HDFS, and use MongoDB to create table structures and indexes for the extracted data that needs to be archived; S303: Set user access control for data stored in the distributed storage system HDFS, build user security accounts, and perform security management.

5. The self-collaborative method for component traditional Chinese medicine design according to claim 3, characterized in that: Step S4 specifically includes: S401. Establish multiple AI-based component TCM agents, including a component TCM active substance prediction agent, a component TCM structure optimization agent, and an active substance compatibility prediction agent. Each agent undertakes different design tasks, forming a component TCM design multi-agent. S402, obtaining negative and positive efficacy indicators of the component traditional Chinese medicine corresponding to the features in the distributed storage system HDFS, performing binary classification through one-hot processing, labeling negative and positive efficacy as "0" and "1" respectively, and randomly dividing the processed multi-dimensional component traditional Chinese medicine dataset into a training set and a validation set at an 8:1 ratio; S403, using the negative and positive efficacy indicators of the component Chinese medicine corresponding to the labeled features as a training set and training with a deep learning algorithm to generate a component Chinese medicine efficacy substance prediction intelligent agent; S404. Use the validation set to evaluate the component TCM active substance prediction intelligent agent. The evaluation methods are accuracy and binary classification ROC-AUC. Adjust the structure of the component TCM active substance prediction intelligent agent based on the evaluation results.

6. The self-collaborative method for component Chinese medicine design according to claim 3, characterized in that: Step S5 specifically includes: S501: The user inputs the target component Chinese medicine into the adjusted component Chinese medicine active substance prediction agent to generate a prediction result; S502: Based on the prediction result generated in step S501, the particle swarm strategy is used for optimization based on the collaborative mechanism, and adaptive adjustment is performed based on the feedback mechanism to generate the final prediction result.

7. The self-collaborative method for component Chinese medicine design according to claim 1, characterized in that: It also includes data interaction on the final prediction result generated in step S5 to generate a visualization interface.

8. A self-cooperation system for designing component Chinese medicines, characterized in that: Using the self-collaborative method for component traditional Chinese medicine design according to any one of claims 1 to 7, comprising a data acquisition layer, a data processing layer, a data storage layer, a model optimization layer, a collaborative generation layer and a user interaction layer; The data collection layer is used to collect multi-dimensional component traditional Chinese medicine data; The data processing layer is used to preprocess the multi-dimensional component traditional Chinese medicine dataset generated by the data acquisition layer and perform correlation analysis on the preprocessed data; The data storage layer is used to perform distributed storage of the data processed by the data processing layer; The model optimization layer is used to design multiple component Chinese medicine design agents, form a component Chinese medicine design multi-agent, obtain the data stored in the data storage layer, train the component Chinese medicine efficacy substance prediction agent, and evaluate and adjust the component Chinese medicine efficacy substance prediction agent; The collaborative generation layer is used to optimize the prediction results based on the collaborative mechanism and generate the final prediction results; The user interaction layer is used to interact with the final prediction results and generate a visual interface.

9. The self-cooperation system for component Chinese medicine design according to claim 8, characterized in that: The model optimization layer performs the following: A1. Establish multiple AI-based components of Chinese medicine (TCM) agents, including a component TCM active substance prediction agent, a component TCM structure optimization agent, and a component TCM active substance compatibility prediction agent. Each agent will be responsible for specific tasks, forming a multi-agent component TCM design system. A2. Obtain data stored in the data storage layer and train it using a deep learning algorithm to obtain an intelligent agent for predicting the efficacy of Chinese herbal medicine components. A3. Obtain the data stored in the data storage layer and use the reinforcement algorithm for training to obtain the component Chinese medicine efficacy structure optimization intelligent agent; A4. Obtain data stored in the data storage layer and train it using a deep learning algorithm to obtain an intelligent agent for predicting the compatibility of medicinal substances. A5. Evaluate the intelligent agent for predicting the effective substances of component traditional Chinese medicines and adjust the structure of the multi-agent design of component traditional Chinese medicines.

10. The self-cooperation system for component Chinese medicine design according to claim 9 is characterized in that: The collaborative generation layer performs the following tasks: B1. Obtain the target component Chinese medicine and generate prediction results; B2. Obtain the prediction results. Based on the collaborative mechanism, the particle swarm strategy is used for optimization to achieve data sharing among multiple component Chinese medicine agents, and adaptive adjustments are made based on the feedback mechanism to generate the final prediction results.

Images