Intelligent data mining method and system for clinical scientific research conversion

Abstract

The invention relates to the technical field of data mining, in particular to an intelligent data mining method and system for clinical scientific research conversion. The system comprises a data processing module, a data identification module and a data mining module. The data processing module is used for acquiring medical data and processing the medical data to obtain available medical data completely consistent with knowledge nodes in a preset medical map, and sending the available medical data to the data identification module; the data identification module is used for analyzing the available medical data, determining key variables in the available medical data, analyzing the key variables according to a preset model, determining a relationship change trend among the key variables, and sending the relationship change trend to the data mining module; and the data mining module is used for determining interaction among the key variables according to the relationship change trend and generating a data analysis report. The data processing efficiency is improved, the analysis precision is enhanced, personalized decision support is provided, and a more intelligent tool is provided for clinical diagnosis and treatment and patient management.

Description

Technical Field

[0001] The present application relates to the field of data mining technology, and in particular to an intelligent data mining method and system for clinical scientific research transformation. Background Art

[0002] With the explosive growth of data in the medical field, medical data has shown unprecedented diversity in both form and complexity. Electronic medical records (EMR), medical images, genomic data, laboratory test results and other types of data have penetrated into every corner of medicine.

[0003] Traditional medical research and clinical decision support systems usually rely on manual analysis and rule-based systems. However, these methods are incapable of dealing with large-scale, multi-dimensional and heterogeneous medical data. Therefore, how to efficiently mine the potential value of medical data and promote the transformation of clinical research and clinical practice has become a hot topic in current medical information technology research. Summary of the invention

[0004] The present application provides an intelligent data mining method and system for clinical scientific research transformation to solve the above problems.

[0005] In a first aspect, the present application provides an intelligent data mining system for clinical scientific research transformation, the system comprising: a data processing module, a data identification module, and a data mining module. The data processing module, the data identification module, and the data mining module are connected in sequence; The data processing module is used to acquire medical data and perform data processing on the medical data to obtain usable medical data that is completely consistent with the knowledge nodes in the preset medical atlas, and send the usable medical data to the data identification module; The data identification module is used to analyze the available medical data, determine the key variables in the available medical data, and analyze the key variables according to a preset model to determine the relationship change trend between the key variables, and send the relationship change trend to the data mining module; the key variables are used to characterize the data items in the medical data set that have an impact on disease diagnosis, treatment or prognosis, including the patient's genetic markers and biomarker levels; the relationship change trend is used to characterize the trend of changes between key variables over time or other factors; The data mining module is used to determine the interaction between key variables based on the relationship change trend, and generate a data analysis report based on the interaction; the interaction is used to characterize the mutual influence or synergy between two or more key variables; the data analysis report is used to characterize the document generated based on the data analysis results, including the analysis of the relationship change trend and the evaluation of the interaction.

[0006] Through this solution, data acquisition can ensure that the system can access and integrate medical data from different data sources to form a unified data view. Data cleaning can improve data quality by identifying and removing outliers, providing an accurate data basis for subsequent analysis. Missing value processing can process missing data through interpolation techniques such as VAE to reduce the impact of missing data on analysis results. Data standardization can eliminate dimensional differences between different data sets, making data more comparable during analysis. Data conversion can convert unstructured text data into structured data, making it easier to analyze with structured data. Key variable extraction can focus on information that is critical to disease diagnosis and treatment or clinical research by identifying key variables, thereby improving the pertinence of analysis. Entity recognition can identify key information in text data, such as disease names, drug names, etc., through NLP technology, providing a basis for subsequent analysis. Word vector generation can convert text data into word vectors, allowing text data to participate in the training of machine learning models together with structured data. Data splicing can splice word vectors of text data with structured data to form a more comprehensive data set, providing more information for model training. Redundancy analysis can remove unnecessary variables through redundancy analysis, simplify the model, and improve the efficiency and interpretability of the model. Relationship analysis can reveal hidden patterns and trends in medical data by analyzing the relationship between key variables, and provide support for clinical decision-making. Interaction determination can determine the interaction between key variables, providing a basis for understanding the mechanism of disease occurrence and treatment effect. Report generation can generate data analysis reports containing key variable interaction analysis, providing reference for clinical research and decision-making. View establishment can intuitively display data analysis results in the form of charts, graphs, etc., which is convenient for users to understand the data. View display can display data through visualization tools to improve data readability and decision-making efficiency. Interaction function determination can determine interactive functions according to user needs, such as data filtering, result sorting, etc., to improve user experience. Touch unit establishment can establish a touch unit, allowing users to interact with the system through devices such as touch screens, improving the convenience of operation. Data storage can store analysis results and reports on the cloud platform to ensure data security and accessibility. Data sharing can achieve data sharing through the cloud platform to promote multidisciplinary cooperation and cross-institutional research. In summary, it can effectively process and analyze medical data, extract key variables, analyze the relationship between variables, and intuitively present analysis results through visualization tools, thereby supporting clinical decision-making and promoting the transformation of scientific research results.

[0007] Optionally, when the data processing module processes the medical data to obtain usable medical data that is completely consistent with the knowledge nodes in the preset medical atlas, it is used to: The following formula is used to determine whether there are outliers in the medical data: ; in, represents mean square error; representing the medical data; Represents preset reconstruction data; If the mean square error is greater than a preset error threshold, it is determined that an abnormal value exists in the medical data, and the abnormal value is removed to obtain first data; Determine the missing part in the first data according to a preset missing marking method; Using VAE to interpolate the missing part to obtain second data; Retrieving a preset medical atlas, mapping the second data into the preset medical atlas, and determining, based on the mapping result, whether the second data is completely consistent with a knowledge node in the preset medical atlas; If they are completely consistent, the second data is used as the available medical data.

[0008] Through this solution, it is ensured that the system can access and integrate medical data from different data sources to form a unified data view, providing a comprehensive data foundation for subsequent analysis. By calculating the mean square error through formula (1), outliers in the data set can be effectively identified, thereby avoiding the impact of these outliers on subsequent analysis results. After removing the outliers, the data quality of the first data set is improved, which is conducive to improving the accuracy and reliability of subsequent analysis results. Using VAE technology to interpolate missing values ​​can effectively fill the missing parts in the data set and reduce the impact of missing data on the analysis results. Eliminating the dimensional differences between different data sets makes the data more comparable during analysis and improves the accuracy of subsequent analysis. Converting text data into word vectors allows text data to participate in the training of machine learning models together with structured data, thereby improving the comprehensiveness of model training. Splicing word vectors and standardized structured data to form a more comprehensive data set provides more information for model training and improves the accuracy and generalization ability of the model. By mapping the data to the preset medical atlas, the accuracy and consistency of the data can be verified to ensure that the data meets the requirements of medical knowledge and atlas. The second data set is confirmed to be usable medical data, which can be used in subsequent data identification modules and data mining modules to provide a high-quality data foundation for data analysis.

[0009] Optionally, when the data identification module analyzes the available medical data to determine key variables in the available medical data, it is used to: Analyzing the available medical data to determine text data and structure data; Standardizing the structural data to obtain standard structural data; Performing entity recognition on the text data to determine the meaning of elements in the text data, and determining a word vector for each word in the text data based on the meaning of the elements; Based on the element meaning, concatenate the word vector and the standard structure data to obtain a concatenated variable; According to the meaning of the elements, redundancy analysis is performed on the splicing variables, redundant variables are determined and deleted, and key variables are obtained.

[0010] Through this solution, data is divided into text and structured data, which helps to apply different data processing and analysis techniques in a targeted manner and improve the efficiency and accuracy of analysis. Standardized structured data can eliminate dimensional differences between different data sets, so that data from different sources can be compared and analyzed under the same analysis framework. Entity recognition can help extract key medical concepts from text data, and word vector conversion can convert text data into numerical form, which is convenient for use in combination with structured data. Splicing variables can integrate the advantages of text data and structured data, provide a more comprehensive data perspective, and help discover more complex associations and patterns. Through redundant analysis, unnecessary data dimensions can be removed, the complexity of the model can be reduced, and the efficiency of model training and the accuracy of prediction can be improved. Through machine learning models, key variables that have an important impact on disease diagnosis, treatment and prognosis can be automatically identified, reducing the subjectivity of human judgment and improving the objectivity and scientificity of analysis.

[0011] Optionally, the data identification module includes a single factor analysis unit. When the data identification module analyzes the key variables according to a preset model to determine the relationship change trend between the key variables, the single factor analysis unit is used to: For any key variable, one key variable is used as the dependent variable and any other key variable is used as the independent variable; Determining a change reference object according to the element meaning of the independent variable and the element meaning of the dependent variable; The independent variable and the dependent variable are fitted according to the change reference object, and the change trend of the relationship between the independent variable and the dependent variable is determined according to the fitting result.

[0012] Through this scheme, by selecting key variables as dependent and independent variables, we can focus on studying the direct relationship between these variables, which helps to simplify the analysis process and may reveal the direct influence between variables. Determining the reference object of change according to the meaning of the element helps to ensure the medical rationality of the analysis results and can combine the analysis results with existing medical knowledge and clinical practice. By fitting the independent and dependent variables, a mathematical model can be established to describe the relationship between them, which is very important for understanding the interaction between variables and predicting future trends. Analyzing the fitting results can help us understand the degree and direction of the influence of independent variables on dependent variables, which is very valuable for making clinical decisions and designing treatment plans. Determining the trend of relationship changes helps to reveal the dynamic relationship between variables, which is very important for predicting disease progression, treatment response and patient prognosis. By visually displaying the fitting results and the trend of relationship changes, the analysis results can be made more intuitive and easy to understand, which helps non-data professionals (such as doctors) better understand and use this information. Iterative analysis of other key variable pairs can build a comprehensive relationship network, which is very important for understanding complex disease mechanisms and optimizing treatment strategies.

[0013] Optionally, a visualization module is also included. The visualization module is connected to the data processing module, the data identification module, and the data mining module respectively, and is used to analyze the output results of the data processing module, the data identification module, and the data mining module, determine the data characteristics of each output result, and establish a view for viewing based on the data characteristics.

[0014] Through this solution, the visualization module can present the output results of the data processing module, data identification module and data mining module to the user in an intuitive and easy-to-understand way. This not only helps users better understand and use data analysis results, but also supports clinical decision-making and promotes the transformation of scientific research results.

[0015] Optionally, when the data mining module generates a data analysis report based on the interaction, it is used to: Get the report template; Analyze the report template to determine the types of items included in the report; Determine the corresponding key variables according to the project type; According to the corresponding key variables, determine the interaction between each corresponding key variable and the remaining key variables; According to the interaction between each corresponding key variable and the other key variables, the project conclusion is determined, and a data analysis report is generated according to the project conclusion and the report template.

[0016] Through this program, the report template provides a standardized framework for the format and content of the report, ensuring the consistency and professionalism of the report. By analyzing the template, you can determine the types of items that the report needs to include, such as summary, methods, results, discussion, conclusions, etc. Based on the item types in the report template, you can determine which key variables are related to these items for detailed analysis in the report. Analyzing the interactions between key variables helps to reveal the relationship and impact between variables and provide a scientific basis for clinical decision-making. Based on the interactions between key variables, project conclusions can be drawn, which will guide clinical practice and policy making. Based on the project conclusions and the report template, a detailed data analysis report can be generated, which will include the analysis results of key variables and project conclusions.

[0017] Optionally, the data identification module includes a multi-factor analysis unit. When the data identification module analyzes the key variables according to a preset model to determine the relationship change trend between the key variables, the multi-factor analysis unit is used to: Analyze the key variables and determine the variable types of the key variables; According to the types of variables, several regression models are established; The key variables are analyzed using the several regression models to determine the changing trends of the relationships between the key variables.

[0018] Through this solution, appropriate analysis methods and models are selected according to the variable type to improve the analysis accuracy. The relationship between variables is analyzed through regression models to provide data support for clinical decision-making. The interaction between variables is revealed to provide a basis for clinical research and the formulation of treatment strategies. The dynamic relationship between variables is revealed to provide support for clinical prediction and prognosis evaluation.

[0019] Optionally, an interaction module is also included, which is connected to the visualization module. After the visualization module establishes a view, the view type is sent to the interaction module. The interaction module determines the interaction function according to the view type and establishes a touch unit according to the interaction function.

[0020] Through this solution, the visualization module creates views to display complex data in an intuitive way, which is convenient for users to understand and analyze. The interaction module determines how to implement user interaction based on the view type. The interaction module works together with the visualization module to provide users with a dynamic and interactive data exploration environment. Users can analyze data through intuitive interactive methods to better understand the characteristics and trends of the data. This helps to improve the efficiency and accuracy of data analysis, support clinical decision-making and promote the transformation of scientific research results.

[0021] Optionally, a cloud platform is also included, which is connected to the visualization module and the data mining module and is used to store the views established by the visualization module and the data analysis report generated by the data mining module to achieve data sharing.

[0022] Through this solution, the persistence and accessibility of the view are ensured, which is convenient for users to view and analyze data at any time. It is convenient for users and researchers to review and share analysis results, supporting scientific research and clinical decision-making. It promotes data sharing and knowledge dissemination, and improves the collaborative efficiency of medical research. Users can access data on different devices, improving the flexibility and availability of the system. The cloud platform becomes the core component for data storage, sharing and access in the process of intelligent positive data mining. It not only improves the availability and accessibility of data, but also enhances the security and maintainability of the system. The cloud platform allows different researchers and clinicians to share data and analysis reports, breaking down data silos and promoting interdisciplinary and cross-institutional collaborative research. It accelerates the communication and dissemination of scientific research results and improves scientific research efficiency. The cloud platform enables researchers and clinicians to access data and analysis reports at any location and at any time, without being restricted by geographical location. Through the cloud platform, data and analysis reports can be displayed openly and transparently to all authorized users, facilitating peer review and verification. The cloud platform can provide detailed content of the data analysis report to help clinicians transform research results into actual diagnosis and treatment plans. It shortens the conversion cycle from research to clinical application and improves the clinical value of scientific research results.

[0023] In a second aspect, the present application provides an intelligent data mining method for clinical research transformation, the method comprising: Acquire medical data and perform data processing on the medical data to obtain usable medical data that is completely consistent with the knowledge nodes in the preset medical atlas; Analyze the available medical data to determine key variables in the available medical data, and analyze the key variables according to a preset model to determine the relationship change trend between the key variables; the key variables are used to characterize data items in the medical data set that have an impact on disease diagnosis, treatment or prognosis, including genetic markers and biomarker levels of patients; the relationship change trend is used to characterize the trend of changes between key variables over time or other factors; Based on the relationship change trend, the interaction between key variables is determined, and based on the interaction, a data analysis report is generated; the interaction is used to characterize the mutual influence or synergy between two or more key variables; the data analysis report is used to characterize the document generated based on the data analysis results, including the analysis of the relationship change trend and the evaluation of the interaction.

[0024] Optionally, the processing of the medical data to obtain usable medical data that is completely consistent with the knowledge nodes in the preset medical atlas includes: The following formula is used to determine whether there are outliers in the medical data: ; in, represents mean square error; representing the medical data; Represents preset reconstruction data; If the mean square error is greater than a preset error threshold, it is determined that an abnormal value exists in the medical data, and the abnormal value is removed to obtain first data; Determine the missing part in the first data according to a preset missing marking method; Using VAE to interpolate the missing part to obtain second data; Retrieving a preset medical atlas, mapping the second data into the preset medical atlas, and determining, based on the mapping result, whether the second data is completely consistent with a knowledge node in the preset medical atlas; If they are completely consistent, the second data is used as the available medical data.

[0025] Optionally, analyzing the available medical data to determine key variables in the available medical data includes: Analyzing the available medical data to determine text data and structure data; Standardizing the structural data to obtain standard structural data; Performing entity recognition on the text data to determine the meaning of elements in the text data, and determining a word vector for each word in the text data based on the meaning of the elements; Based on the element meaning, concatenate the word vector and the standard structure data to obtain a concatenated variable; According to the meaning of the elements, redundancy analysis is performed on the splicing variables, redundant variables are determined and deleted, and key variables are obtained.

[0026] Optionally, analyzing the key variables according to a preset model to determine the relationship change trend between the key variables includes: For any key variable, one key variable is used as the dependent variable and any other key variable is used as the independent variable; Determining a change reference object according to the element meaning of the independent variable and the element meaning of the dependent variable; The independent variable and the dependent variable are fitted according to the change reference object, and the change trend of the relationship between the independent variable and the dependent variable is determined according to the fitting result.

[0027] Optionally, the method further includes: The output results of the data processing module, the data identification module and the data mining module are analyzed to determine the data characteristics of each output result, and a view is established for viewing based on the data characteristics.

[0028] Optionally, generating a data analysis report according to the interaction includes: Get the report template; Analyze the report template to determine the types of items included in the report; Determine the corresponding key variables according to the project type; According to the corresponding key variables, determine the interaction between each corresponding key variable and the remaining key variables; According to the interaction between each corresponding key variable and the other key variables, the project conclusion is determined, and a data analysis report is generated according to the project conclusion and the report template.

[0029] Optionally, the method further includes: Analyze the key variables and determine the variable types of the key variables; According to the types of variables, several regression models are established; The key variables are analyzed using the several regression models to determine the changing trends of the relationships between the key variables.

[0030] Optionally, the method further includes, after establishing the view, determining an interactive function according to the view type, and establishing a touch control unit according to the interactive function.

[0031] Optionally, the method further includes storing the view established by the visualization module and the data analysis report generated by the data mining module to achieve data sharing. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0033] Figure 1 A schematic diagram of an application scenario provided for an embodiment of the present application; Figure 2 A schematic diagram of the structure of an intelligent data mining system for clinical research transformation provided in one embodiment of the present application; Figure 3A flow chart of an intelligent data mining method for clinical research transformation provided in one embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0035] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.

[0036] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0037] With the explosive growth of data in the medical field, medical data has shown unprecedented diversity in both form and complexity. Electronic medical records (EMR), medical images, genomic data, laboratory test results and other types of data have penetrated into every corner of medicine.

[0038] Traditional medical research and clinical decision support systems usually rely on manual analysis and rule-based systems. However, these methods are incapable of dealing with large-scale, multi-dimensional and heterogeneous medical data. Therefore, how to efficiently mine the potential value of medical data and promote the transformation of clinical research and clinical practice has become a hot topic in current medical information technology research.

[0039] Based on this, the present application provides an intelligent data mining method and system for clinical scientific research transformation, including a data processing module, a data identification module, and a data mining module, which are connected in sequence; the data processing module is used to acquire medical data and perform data processing on the medical data to obtain available medical data that is completely consistent with the knowledge nodes in the preset medical atlas, and send the available medical data to the data identification module; the data identification module is used to analyze the available medical data, determine the key variables in the available medical data, and analyze the key variables according to the preset model to determine the relationship change trend between the key variables, and send the relationship change trend to the data mining module; the key variables are used to characterize the data items in the medical data set that have an impact on disease diagnosis, treatment or prognosis, including the patient's genetic markers and biomarker levels; the relationship change trend is used to characterize the trend of changes between key variables over time or other factors; the data mining module is used to determine the interaction between key variables based on the relationship change trend, and generate a data analysis report based on the interaction; the interaction is used to characterize the mutual influence or synergy between two or more key variables; the data analysis report is used to characterize the document generated based on the data analysis results, including the analysis of the relationship change trend and the evaluation of the interaction. It can not only process multi-source and multi-type medical data, remove outliers and fill in missing data, but also extract valuable key variables from massive data, analyze the relationship between variables, and intuitively present the analysis results through visualization tools. In addition, the system can also generate personalized research reports based on clinical research needs, and promote the transformation of scientific research results into practical clinical decision support. By introducing artificial intelligence technology, especially emerging technologies such as deep learning, natural language processing (NLP), and variational autoencoders (VAE), data processing efficiency can be improved, analysis accuracy can be enhanced, and personalized decision support can be provided, providing more intelligent tools for medical research, clinical diagnosis and treatment, and patient management. The combination of these technologies helps to improve the accuracy and operability of medical data mining, promote the in-depth development of clinical scientific research, promote the effective transformation of clinical scientific research results, and ultimately improve the quality of medical services and clinical effects.

[0040] Figure 1A schematic diagram of an application scenario provided for this application, when analyzing a large amount of medical data and mining the data, the method provided by this application is applied. Specifically, the method provided by this application is applied to any server, and the server interacts with the medical data system. It can not only process multi-source and multi-type medical data, eliminate outliers and fill in missing data, but also extract valuable key variables from massive data, analyze the relationship between variables, and intuitively present the analysis results through visualization tools. In addition, the system can also generate personalized research reports according to clinical research needs, and promote the transformation of scientific research results into actual clinical decision support. The specific implementation method can refer to the following embodiments.

[0041] Figure 2 A schematic diagram of the structure of an intelligent data mining system for clinical research transformation provided in one embodiment of the present application is shown in FIG. Figure 2 As shown, the visualized intelligent data mining system 200 for clinical research transformation of this embodiment includes: a data processing module 201, a data identification module 202, and a data mining module 203. The data processing module 201, the data identification module 202, and the data mining module 203 are connected in sequence; the data processing module 201 is used to acquire medical data and perform data processing on the medical data to obtain available medical data that is completely consistent with the knowledge nodes in the preset medical atlas, and send the available medical data to the data identification module 202; the data identification module 202 is used to analyze the available medical data, determine the key variables in the available medical data, and analyze the key variables according to the preset model to determine the relationship change trend between the key variables, and send the relationship change trend to the data mining module 203; the key variables are used to characterize the data items in the medical data set that have an impact on disease diagnosis, treatment or prognosis, including the patient's genetic markers and biomarker levels; the relationship change trend is used to characterize the trend of changes between key variables over time or other factors; the data mining module 203 is used to determine the interaction between key variables based on the relationship change trend, and generate a data analysis report based on the interaction; the interaction is used to characterize the mutual influence or synergy between two or more key variables; the data analysis report is used to characterize the document generated based on the data analysis results, including the analysis of the relationship change trend and the evaluation of the interaction.

[0042] Medical data can be various types of data collected from medical research and clinical practice, such as electronic medical records (EMR), medical images (such as X-rays, CT scans, MRI, etc.), genomic data, laboratory test results, patient demographic information, medical history, medication records, surgical records, etc.

[0043] Data processing can be a series of operations to clean, convert, standardize and integrate raw medical data to facilitate subsequent data analysis and mining, such as outlier detection, missing value processing, data format conversion, data quality assessment, etc.

[0044] Available medical data can be medical data that has been processed, has had outliers removed, missing values ​​filled, and necessary format conversion and standardization performed, making it suitable for data analysis and mining.

[0045] Key variables can be data items in medical data sets that have an important impact on disease diagnosis, treatment or prognosis, such as the patient's age, gender, genetic markers, biomarker levels, drug response, etc.

[0046] A preset model can be an analysis model that is pre-designed based on medical knowledge and data characteristics and is used to analyze and predict medical data.

[0047] Relationship change trends can be patterns and trends of changes between key variables over time or other factors. By analyzing these trends, the potential laws of disease development or changes in treatment effects can be revealed.

[0048] Interaction can be the mutual influence or synergy between two or more key variables. In medical data mining, identifying interactions between key variables helps to understand the complexity of the disease and develop more precise treatment strategies.

[0049] A data analysis report is a document generated based on the results of data analysis, which usually includes a description of key variables, an analysis of relationship trends, an assessment of interactions, and recommendations for clinical decision-making. The purpose of the report is to provide decision support for medical professionals.

[0050] Specifically, the complexity of medical data is reflected in multiple aspects, mainly including huge data volume, diverse data types, and large differences in data quality. A large amount of medical data has problems such as missing, noise, and outliers. If these problems are not handled, they may lead to inaccurate data analysis results or mislead decision-making. Therefore, it is necessary to first obtain original medical data from different data sources (such as EMR, medical imaging, genomics data, etc.). Statistical analysis methods are used to identify and eliminate outliers. For the missing parts of the data, VAE technology is used for interpolation processing to estimate the missing data points. Structured data is standardized to eliminate the impact of dimensions for subsequent analysis. Entity recognition is performed on text data and converted into word vectors to be combined with structured data. By analyzing the standardized data, key variables related to disease diagnosis and treatment or clinical research are identified. Entity recognition is performed on text data to determine the meaning of elements in text data. According to the meaning of elements, words in text data are converted into word vectors. Word vectors and standard structure data are spliced ​​to form spliced ​​variables. Redundancy analysis is performed on spliced ​​variables, redundant variables are eliminated, and key variables are retained. According to the preset model, the relationship between key variables is analyzed. Determine the interaction between variables based on the changing trend of the relationship between key variables. Generate a detailed data analysis report based on the interaction. Analyze the output results of the data processing module, data identification module, and data mining module to determine the data characteristics. According to the data characteristics, establish intuitive views for viewing, such as charts, graphs, etc. Determine the corresponding interactive functions based on the view type established by the visualization module. According to the interactive functions, establish touch units to facilitate user interaction with the system. Store the views established by the visualization module and the data analysis reports generated by the data mining module. Realize data sharing on the cloud platform to facilitate cross-platform access.

[0051] Through this solution, data acquisition can ensure that the system can access and integrate medical data from different data sources to form a unified data view. Data cleaning can improve data quality by identifying and removing outliers, providing an accurate data basis for subsequent analysis. Missing value processing can process missing data through interpolation techniques such as VAE to reduce the impact of missing data on analysis results. Data standardization can eliminate dimensional differences between different data sets, making data more comparable during analysis. Data conversion can convert unstructured text data into structured data, making it easier to analyze with structured data. Key variable extraction can focus on information that is critical to disease diagnosis and treatment or clinical research by identifying key variables, thereby improving the pertinence of analysis. Entity recognition can identify key information in text data, such as disease names, drug names, etc., through NLP technology, providing a basis for subsequent analysis. Word vector generation can convert text data into word vectors, allowing text data to participate in the training of machine learning models together with structured data. Data splicing can splice word vectors of text data with structured data to form a more comprehensive data set, providing more information for model training. Redundancy analysis can remove unnecessary variables through redundancy analysis, simplify the model, and improve the efficiency and interpretability of the model. Relationship analysis can reveal hidden patterns and trends in medical data by analyzing the relationship between key variables, and provide support for clinical decision-making. Interaction determination can determine the interaction between key variables, providing a basis for understanding the mechanism of disease occurrence and treatment effect. Report generation can generate data analysis reports containing key variable interaction analysis, providing reference for clinical research and decision-making. View establishment can intuitively display data analysis results in the form of charts, graphs, etc., which is convenient for users to understand the data. View display can display data through visualization tools to improve data readability and decision-making efficiency. Interaction function determination can determine interactive functions according to user needs, such as data filtering, result sorting, etc., to improve user experience. Touch unit establishment can establish a touch unit, allowing users to interact with the system through devices such as touch screens, improving the convenience of operation. Data storage can store analysis results and reports on the cloud platform to ensure data security and accessibility. Data sharing can achieve data sharing through the cloud platform to promote multidisciplinary cooperation and cross-institutional research. In summary, it can effectively process and analyze medical data, extract key variables, analyze the relationship between variables, and intuitively present analysis results through visualization tools, thereby supporting clinical decision-making and promoting the transformation of scientific research results.

[0052] In some embodiments, the following formula (1) is used to determine whether there are outliers in the medical data: (1) in, represents mean square error; Represents medical data; Represents preset reconstruction data; If the mean square error is greater than the preset error threshold, it is determined that there are outliers in the medical data, and the outliers are eliminated to obtain the first data; according to the preset missing marking method, the missing part in the first data is determined; the missing part is interpolated using VAE to obtain the second data; the preset medical atlas is retrieved, the second data is mapped to the preset medical atlas, and according to the mapping result, it is determined whether the second data is completely consistent with the knowledge nodes in the preset medical atlas; if they are completely consistent, the second data is used as available medical data.

[0053] Outliers can be data points in a data set that are significantly different from other data points.

[0054] The mean square error can be a statistical indicator that measures the difference between each data point in the data set and the preset reconstructed data.

[0055] The preset reconstruction data may be a data set that is pre-calculated according to the statistical characteristics of the data set or domain knowledge during the data preprocessing stage, and is stored in a preset database for use when needed.

[0056] The preset error threshold may be a standard for determining whether the mean square error is too large, and is stored in a preset database for recall when needed.

[0057] The first data may be a preliminarily processed data set after outliers are removed.

[0058] The preset missing value marking method can be a method for marking missing values ​​in a data set, which is stored in a preset database and can be called when needed.

[0059] Missing parts can be data points that are not recorded or are incompletely recorded in the data set.

[0060] VAE Chinese name: Variational Autoencoder, is a generative adversarial network that can learn data distribution and reconstruct data through encoder and decoder.

[0061] Interpolation is the process of estimating and filling in missing values ​​in a dataset using mathematical methods.

[0062] A preset medical atlas can be a framework that includes medical knowledge and data structures, which is used to guide the analysis and interpretation of medical data and is stored in a preset database for use when needed.

[0063] The mapping result may be a data point corresponding to a medical knowledge node obtained after mapping the data set to a preset medical atlas.

[0064] A knowledge node can be a medical concept or entity defined in a preset medical graph.

[0065] Specifically, use formula (1) to calculate the mean square error To detect outliers, It is the original medical data. is the preset reconstruction data. Set a preset error threshold. If the calculated mean square error If it is greater than this threshold, the data is considered to contain outliers. The detected outliers are removed from the data set to obtain the first data set. According to the preset missing marking method, the missing parts in the first data set are identified. The missing parts are interpolated using VAE (variational autoencoder) technology to generate the second data set. The structured data in the second data set is standardized to eliminate the dimensional differences between different data sets. Entity recognition is performed on the text data in the second data set to determine the meaning of the elements in the text data and convert them into word vectors. The word vectors and the standardized structured data are spliced ​​to form a complete and analyzable data set. The preset medical atlas is retrieved and the spliced ​​data is mapped to the atlas. According to the mapping results, check whether the data is completely consistent with the knowledge nodes in the atlas. If the data is completely consistent with the knowledge nodes in the medical atlas, the second data set is confirmed to be usable medical data and can be used in subsequent data identification and data mining modules.

[0066] Through this solution, it is ensured that the system can access and integrate medical data from different data sources to form a unified data view, providing a comprehensive data foundation for subsequent analysis. By calculating the mean square error through formula (1), outliers in the data set can be effectively identified, thereby avoiding the impact of these outliers on subsequent analysis results. After removing the outliers, the data quality of the first data set is improved, which is conducive to improving the accuracy and reliability of subsequent analysis results. Using VAE technology to interpolate missing values ​​can effectively fill the missing parts in the data set and reduce the impact of missing data on the analysis results. Eliminating the dimensional differences between different data sets makes the data more comparable during analysis and improves the accuracy of subsequent analysis. Converting text data into word vectors allows text data to participate in the training of machine learning models together with structured data, thereby improving the comprehensiveness of model training. Splicing word vectors and standardized structured data to form a more comprehensive data set provides more information for model training and improves the accuracy and generalization ability of the model. By mapping the data to the preset medical atlas, the accuracy and consistency of the data can be verified to ensure that the data meets the requirements of medical knowledge and atlas. The second data set is confirmed to be usable medical data, which can be used in subsequent data identification modules and data mining modules to provide a high-quality data foundation for data analysis.

[0067] In some embodiments, available medical data is analyzed to determine text data and structural data; the structural data is standardized to obtain standard structural data; entity recognition is performed on the text data to determine the meaning of elements in the text data, and based on the element meaning, the word vector of each word in the text data is determined; based on the element meaning, the word vector and the standard structural data are spliced ​​to obtain spliced ​​variables; based on the element meaning, the spliced ​​variables are subjected to redundancy analysis, redundant variables are determined and deleted, and key variables are obtained.

[0068] Text data can be medical information in text form, such as clinical medical records, medical literature, patient reports, etc.

[0069] Structured data can be medical information in a structured form, such as a patient's age, gender, height, weight, laboratory test results, imaging measurements, etc.

[0070] Normalization can be the process of converting structured data into data with the same dimensions to facilitate comparison and analysis between different data sets.

[0071] Standard structured data can be structured data that has been standardized, with unified dimensions and formats, and is suitable for subsequent data analysis and machine learning model training.

[0072] Entity recognition can be the process of identifying medical entities with specific meanings from text data, such as disease names, drug names, symptom descriptions, etc.

[0073] Element meaning can be the medical meaning of each word or phrase in the text data. During entity recognition, each identified medical entity is assigned a specific meaning to facilitate subsequent data analysis and interpretation.

[0074] Word embedding can be a vector that converts each word or phrase in text data into a high-dimensional space. These vectors can capture the semantic relationship between words, making text data useful for machine learning models.

[0075] The concatenated variables can be a new variable set formed by combining the standardized results of structured data with the word vectors of text data.

[0076] Redundancy analysis can be used to identify and remove redundant variables in a data set, that is, those variables that are highly correlated with other variables.

[0077] Specifically, available medical data is divided into text data and structured data. Text data may include clinical medical records, medical literature, etc., while structured data may include the patient's age, gender, laboratory test results, etc. Standardize the structured data, for example, standardize continuous variables such as age and weight, and encode categorical variables such as gender and disease type. Use natural language processing (NLP) technology to perform entity recognition on text data to identify key medical concepts in the text, such as disease names, drug names, symptom descriptions, etc. According to the results of entity recognition, determine the meaning of each word in the text data, and use word embedding technology (such as Word2Vec, GloVe or BERT) to convert the text data into word vectors. Concatenate the standardized results of structured data with the word vectors of text data to form concatenated variables. This can combine the rich semantic information in text data and the precise numerical information of structured data. Perform redundancy analysis on concatenated variables to identify and delete redundant variables. Redundancy analysis can be achieved through feature selection algorithms (such as principal component analysis PCA, linear discriminant analysis LDA, etc.) to reduce data dimensions and improve the generalization ability of the model. The processed data is trained through machine learning models (such as random forest, support vector machine SVM, gradient boosting tree GBDT, etc.) to identify key variables related to disease diagnosis or clinical research. The model can determine which variables are most critical to disease prediction based on the importance score of the features.

[0078] Through this solution, data is divided into text and structured data, which helps to apply different data processing and analysis techniques in a targeted manner and improve the efficiency and accuracy of analysis. Standardized structured data can eliminate dimensional differences between different data sets, so that data from different sources can be compared and analyzed under the same analysis framework. Entity recognition can help extract key medical concepts from text data, and word vector conversion can convert text data into numerical form, which is convenient for use in combination with structured data. Splicing variables can integrate the advantages of text data and structured data, provide a more comprehensive data perspective, and help discover more complex associations and patterns. Through redundant analysis, unnecessary data dimensions can be removed, the complexity of the model can be reduced, and the efficiency of model training and the accuracy of prediction can be improved. Through machine learning models, key variables that have an important impact on disease diagnosis, treatment and prognosis can be automatically identified, reducing the subjectivity of human judgment and improving the objectivity and scientificity of analysis.

[0079] In some embodiments, for any key variable, one key variable is used as the dependent variable, and any other key variable is used as the independent variable; the change reference object is determined based on the element meaning of the independent variable and the element meaning of the dependent variable; based on the change reference object, the independent variable and the dependent variable are fitted, and based on the fitting results, the change trend of the relationship between the independent variable and the dependent variable is determined.

[0080] Key variables can be data items in medical data sets that have important effects on disease diagnosis, treatment or prognosis.

[0081] The change reference object can be an object used for comparison and reference when analyzing the change trend of the relationship between key variables.

[0082] Specifically, a key variable is selected from the data set as the dependent variable, and another key variable is selected as the independent variable. According to the element meaning of the independent variable and the dependent variable, the reference object of the change is determined. The reference object can be a specific node in the medical knowledge graph or a set of related medical concepts. The independent and dependent variables are fitted using statistical or machine learning models (such as linear regression, polynomial regression, time series analysis, etc.). The purpose of fitting is to build a model that describes how the independent variable affects the dependent variable. The parameters and statistical indicators of the fitting model (such as R-squared, p-value, etc.) are analyzed to determine the strength and direction of the relationship between the independent and dependent variables. According to the fitting results, the trend of the relationship between the independent and dependent variables is determined. This may be manifested as a linear relationship, a nonlinear relationship, a periodic change, etc. The fitting results and the trend of the relationship change are visualized through charts or graphs for easy understanding and interpretation. The same analysis process is performed on other key variable pairs to fully understand the relationship network among all key variables.

[0083] Through this scheme, by selecting key variables as dependent and independent variables, we can focus on studying the direct relationship between these variables, which helps to simplify the analysis process and may reveal the direct influence between variables. Determining the reference object of change according to the meaning of the element helps to ensure the medical rationality of the analysis results and can combine the analysis results with existing medical knowledge and clinical practice. By fitting the independent and dependent variables, a mathematical model can be established to describe the relationship between them, which is very important for understanding the interaction between variables and predicting future trends. Analyzing the fitting results can help us understand the degree and direction of the influence of independent variables on dependent variables, which is very valuable for making clinical decisions and designing treatment plans. Determining the trend of relationship changes helps to reveal the dynamic relationship between variables, which is very important for predicting disease progression, treatment response and patient prognosis. By visually displaying the fitting results and the trend of relationship changes, the analysis results can be made more intuitive and easy to understand, which helps non-data professionals (such as doctors) better understand and use this information. Iterative analysis of other key variable pairs can build a comprehensive relationship network, which is very important for understanding complex disease mechanisms and optimizing treatment strategies.

[0084] In some embodiments, the intelligent data mining system 200 for clinical scientific research transformation also includes a visualization module 204, which is respectively connected to the data processing module 201, the data identification module 202, and the data mining module 203, and is used to analyze the output results of the data processing module 201, the data identification module 202, and the data mining module 203, determine the data characteristics of each output result, and establish a view for viewing based on the data characteristics.

[0085] Data characteristics can be a set of features and attributes of a data set that together describe the basic nature, structure, and quality of the data.

[0086] Specifically, the visualization module 204 receives output results from the data processing module 201, the data identification module 202, and the data mining module 203, including processed data, identified key variables, and analysis results. The received output results are analyzed to determine the characteristics of the data, such as the distribution, trend, pattern, etc. of the data. This can be achieved through statistical analysis, data exploration, and visualization exploration. According to the characteristics of the data, appropriate visualization techniques are selected to establish views. This may include charts, graphs, maps, network diagrams, etc., so as to intuitively display the data and analysis results. A user-friendly interactive interface is designed to allow users to interact with the view through operations such as clicking, sliding, and zooming, so as to explore the data and analysis results in more depth. Front-end development technologies (such as HTML, CSS, JavaScript) and visualization libraries (such as D3.js, Plotly, Highcharts, etc.) are used to realize the display of the view, ensuring that users can easily view and understand the data. An export function is provided to allow users to export the visualized data and analysis results into pictures, PDFs, or other formats for easy sharing and further analysis. An integrated feedback mechanism is provided to allow users to provide feedback on the visualized data and analysis results to improve the visualization and analysis process.

[0087] Through this solution, the visualization module 204 can present the output results of the data processing module 201, the data identification module 202 and the data mining module 203 to the user in an intuitive and easy-to-understand manner. This not only helps users better understand and use data analysis results, but also supports clinical decision-making and promotes the transformation of scientific research results.

[0088] In some embodiments, a report template is obtained; the report template is analyzed to determine the project types included in the report; based on the project types, corresponding key variables are determined; based on the corresponding key variables, the interaction between each corresponding key variable and the remaining key variables is determined; based on the interaction between each corresponding key variable and the remaining key variables, the project conclusion is determined, and a data analysis report is generated based on the project conclusion and the report template.

[0089] A report template can be a preset document structure that defines the format, content, and organization of a report.

[0090] Item types can be different content sections or chapters in a report.

[0091] The project conclusion can be the final summary based on the data analysis results in the report, which usually includes an explanation of the research results, recommendations for clinical practice, and prospects for future research.

[0092] Specifically, design or select a template suitable for medical data mining reports, which should contain all necessary information modules, such as summary, methods, results, discussion, conclusions, etc. Analyze the template to determine the types of items and content structure that need to be included in the report. According to the types of items in the report template, identify the key variables associated with these items from the data set. Use statistical analysis and machine learning techniques to analyze the interactions between key variables, such as correlation analysis, regression analysis, network analysis, etc. Based on the interactions between key variables, draw project conclusions that will guide clinical practice and policy making. Based on the project conclusions and report template, use natural language generation technology to automatically fill in the report content and generate a detailed data analysis report.

[0093] Through this program, the report template provides a standardized framework for the format and content of the report, ensuring the consistency and professionalism of the report. By analyzing the template, you can determine the types of items that the report needs to include, such as summary, methods, results, discussion, conclusions, etc. Based on the item types in the report template, you can determine which key variables are related to these items for detailed analysis in the report. Analyzing the interactions between key variables helps to reveal the relationship and impact between variables and provide a scientific basis for clinical decision-making. Based on the interactions between key variables, project conclusions can be drawn, which will guide clinical practice and policy making. Based on the project conclusions and the report template, a detailed data analysis report can be generated, which will include the analysis results of key variables and project conclusions.

[0094] In some embodiments, the data identification module 202 includes a multi-factor analysis unit 2022. When the data identification module 202 analyzes key variables according to a preset model and determines the relationship change trend between the key variables, the multi-factor analysis unit 2022 is used to: analyze the key variables and determine the variable types of the key variables; establish several regression models based on the variable types; use several regression models to analyze the key variables and determine the relationship change trend between the key variables.

[0095] The variable type can be the nature and characteristics of the variables in the data set, such as continuous variables, discrete variables, categorical variables, ordinal variables, nominal variables, proportional variables, etc.

[0096] Specifically, statistical analysis is performed on key variables to determine their variable types (such as continuous variables, categorical variables, time series variables, etc.). According to the types of key variables, corresponding regression models are established (such as linear regression, logistic regression, time series regression, etc.). The established regression model is used to analyze key variables to evaluate the relationship and impact of variables. By analyzing the output results of the regression model, the changing trend of the relationship between key variables is determined.

[0097] Through this solution, appropriate analysis methods and models are selected according to the variable type to improve the analysis accuracy. The relationship between variables is analyzed through regression models to provide data support for clinical decision-making. The interaction between variables is revealed to provide a basis for clinical research and the formulation of treatment strategies. The dynamic relationship between variables is revealed to provide support for clinical prediction and prognosis evaluation.

[0098] In some embodiments, the intelligent data mining system 200 for clinical scientific research transformation also includes an interaction module 205, which is connected to the visualization module 204. After the visualization module establishes a view, the view type is sent to the interaction module. The interaction module determines the interaction function based on the view type and establishes a touch unit based on the interaction function.

[0099] The view type can be in the form of graphs or charts used for data visualization, such as bar charts, line charts, scatter plots, heat maps, maps, etc.

[0100] Interactive functions can refer to the actions that users can perform when interacting with the data visualization interface, such as zooming, filtering, sorting, searching, annotating, downloading, etc.

[0101] A touch unit may be an element in a user interface that allows a user to interact with it, such as a button, a slide bar, a drop-down menu, an input box, a check box, a radio button, etc.

[0102] Specifically, the visualization module 204 selects a suitable visualization type (such as a chart, graph, map, etc.) to display the data according to the output results of the data processing module 201, the data identification module 202, and the data mining module 203. The visualization module 204 sends the created view type information to the interaction module 205. The interaction module 205 determines the interactive functions that the user may need, such as zooming, filtering, searching, sorting, etc., according to the received view type. The interaction module 205 designs and implements touch units according to the determined interactive functions, and these units can be buttons, sliders, selection boxes, etc. The touch units are integrated into the visualization interface, and the user can interact with the view through these units, such as clicking a button to filter data, dragging a slider to adjust the view size, etc. According to the user's interactive behavior, feedback information is collected, and the visualization or interactive functions may be adjusted to better meet user needs.

[0103] Through this solution, the visualization module creates a view to display complex data in an intuitive way, which is convenient for users to understand and analyze. The interaction module 205 determines how to implement user interaction based on the view type. The interaction module 205 works in conjunction with the visualization module 204 to provide users with a dynamic and interactive data exploration environment. Users can analyze data through intuitive interactive methods to better understand the characteristics and trends of the data. This helps to improve the efficiency and accuracy of data analysis, support clinical decision-making and promote the transformation of scientific research results.

[0104] In some embodiments, the intelligent data mining system 200 for clinical scientific research transformation also includes a cloud platform 206, which is connected to the visualization module 204 and the data mining module 203, and is used to store the views established by the visualization module 204 and the data analysis reports generated by the data mining module 203 to achieve data sharing.

[0105] Specifically, various views (such as charts, graphs, etc.) generated by the visualization module 204 are stored on the cloud platform 206. Data analysis reports generated by the data mining module 203 are stored on the cloud platform 206. Data sharing services are provided through the cloud platform 206, allowing authorized users to access and download views and reports. Ensure that the cloud platform 206 supports multiple devices and operating systems, such as PCs, tablets, smartphones, etc., as well as operating systems such as Windows, MacOS, iOS, and Android.

[0106] Through this solution, the persistence and accessibility of the view are ensured, which is convenient for users to view and analyze data at any time. It is convenient for users and researchers to review and share analysis results, and support scientific research and clinical decision-making. It promotes data sharing and knowledge dissemination, and improves the collaborative efficiency of medical research. Users can access data on different devices to improve the flexibility and availability of the system. The cloud platform 206 becomes the core component for data storage, sharing and access in the process of intelligent positive data mining. It not only improves the availability and accessibility of data, but also enhances the security and maintainability of the system. The cloud platform 206 allows different researchers and clinicians to share data and analysis reports, breaking the data island and promoting interdisciplinary and inter-institutional collaborative research. It accelerates the communication and dissemination of scientific research results and improves scientific research efficiency. The cloud platform 206 enables researchers and clinicians to access data and analysis reports at any place and at any time, without being restricted by geographical location. Through the cloud platform 206, data and analysis reports can be displayed openly and transparently to all authorized users, which is convenient for peer review and verification. The cloud platform 206 can provide detailed content of the data analysis report to help clinicians transform research results into actual diagnosis and treatment plans. It shortens the transformation cycle from research to clinical application and improves the clinical value of scientific research results.

[0107] Figure 3 This is a flow chart of an intelligent data mining method for clinical research transformation provided in an embodiment of the present application. The method of this embodiment can be applied to the server in the above scenario. Figure 3 As shown, the method includes: S301, acquiring medical data and performing data processing on the medical data to obtain usable medical data that is completely consistent with the knowledge nodes in the preset medical atlas; S302, analyzing available medical data, determining key variables in the available medical data, and analyzing the key variables according to a preset model to determine the relationship change trend between the key variables; the key variables are used to characterize data items in the medical data set that have an impact on disease diagnosis, treatment or prognosis, including genetic markers and biomarker levels of patients; the relationship change trend is used to characterize the trend of changes between key variables over time or other factors; S303. Determine the interaction between key variables based on the relationship change trend, and generate a data analysis report based on the interaction; the interaction is used to characterize the mutual influence or synergy between two or more key variables; the data analysis report is used to characterize the document generated based on the data analysis results, including the analysis of the relationship change trend and the evaluation of the interaction.

[0108] Optionally, the following formula is used to determine whether there are outliers in the medical data: ; in, represents mean square error; Represents medical data; Represents preset reconstruction data; If the mean square error is greater than the preset error threshold, it is determined that there are outliers in the medical data, and the outliers are eliminated to obtain the first data; according to the preset missing marking method, the missing part in the first data is determined; the missing part is interpolated using VAE to obtain the second data; the preset medical atlas is retrieved, the second data is mapped to the preset medical atlas, and according to the mapping result, it is determined whether the second data is completely consistent with the knowledge nodes in the preset medical atlas; if they are completely consistent, the second data is used as available medical data.

[0109] Optionally, analyze available medical data to determine text data and structural data; standardize the structural data to obtain standard structural data; perform entity recognition on the text data to determine the meaning of elements in the text data, and determine the word vector of each word in the text data based on the element meaning; based on the element meaning, concatenate the word vector and the standard structural data to obtain concatenated variables; perform redundancy analysis on the concatenated variables based on the element meaning, determine redundant variables and delete them to obtain key variables.

[0110] Optionally, for any key variable, one key variable is used as the dependent variable, and any other key variable is used as the independent variable; the change reference object is determined according to the element meaning of the independent variable and the element meaning of the dependent variable; the independent variable and the dependent variable are fitted according to the change reference object, and the change trend of the relationship between the independent variable and the dependent variable is determined according to the fitting result.

[0111] Optionally, the output results of the data processing module, the data identification module, and the data mining module are analyzed to determine the data characteristics of each output result, and a view is established for viewing based on the data characteristics.

[0112] Optionally, obtain a report template; analyze the report template to determine the project types included in the report; determine the corresponding key variables based on the project types; determine the interaction between each corresponding key variable and the remaining key variables based on the corresponding key variables; determine the project conclusions based on the interaction between each corresponding key variable and the remaining key variables, and generate a data analysis report based on the project conclusions and the report template.

[0113] Optionally, analyze key variables and determine their variable types; establish several regression models based on the variable types; use several regression models to analyze key variables and determine the changing trend of the relationship between key variables.

[0114] Optionally, after the view is established, the interactive function is determined according to the view type, and the touch unit is established according to the interactive function.

[0115] Optionally, the views created by the visualization module and the data analysis reports generated by the data mining module are stored to achieve data sharing.

[0116] The method of this embodiment can be applied to the system of any of the above embodiments. Its implementation principle and technical effects are similar and will not be described in detail here.

Claims

1. An intelligent data mining system for clinical research transformation, characterized in that: Including data processing module, data identification module, data mining module, The data processing module, the data identification module, and the data mining module are connected in sequence; The data processing module is used to acquire medical data and perform data processing on the medical data to obtain usable medical data that is completely consistent with the knowledge nodes in the preset medical atlas, and send the usable medical data to the data identification module; The data identification module is used to analyze the available medical data, determine the key variables in the available medical data, analyze the key variables according to a preset model, determine the relationship change trend between the key variables, and send the relationship change trend to the data mining module; The key variables are used to characterize data items in medical data sets that have an impact on disease diagnosis, treatment or prognosis, including genetic markers and biomarker levels of patients; the relationship change trend is used to characterize the trend of changes between key variables over time or other factors; The data mining module is used to determine the interaction between key variables according to the relationship change trend, and generate a data analysis report according to the interaction; the interaction is used to characterize the mutual influence or synergy between two or more key variables; The data analysis report is used to characterize the document generated based on the data analysis results, including the analysis of relationship change trends and the evaluation of interactions.

2. The system according to claim 1, characterized in that When the data processing module processes the medical data to obtain available medical data that is completely consistent with the knowledge nodes in the preset medical atlas, it is used to: The following formula is used to determine whether there are outliers in the medical data: ； in, represents mean square error; representing the medical data; Represents preset reconstruction data; If the mean square error is greater than a preset error threshold, it is determined that an abnormal value exists in the medical data, and the abnormal value is removed to obtain first data; Determine the missing part in the first data according to a preset missing marking method; Using VAE to interpolate the missing part to obtain second data; Mapping the second data to the preset medical atlas, and determining whether the second data is completely consistent with the knowledge nodes in the preset medical atlas according to the mapping result; If they are completely consistent, the second data is used as the available medical data.

3. The system according to claim 2, characterized in that When the data identification module analyzes the available medical data and determines the key variables in the available medical data, it is used to: Analyzing the available medical data to determine text data and structure data; Standardizing the structural data to obtain standard structural data; Performing entity recognition on the text data to determine the meaning of elements in the text data, and determining a word vector for each word in the text data based on the meaning of the elements; Based on the element meaning, concatenate the word vector and the standard structure data to obtain a concatenated variable; According to the meaning of the elements, redundancy analysis is performed on the splicing variables, redundant variables are determined and deleted, and key variables are obtained.

4. The system according to claim 3, characterized in that The data identification module includes a single factor analysis unit. When the data identification module analyzes the key variables according to a preset model and determines the relationship change trend between the key variables, the single factor analysis unit is used to: For any key variable, one key variable is used as the dependent variable and any other key variable is used as the independent variable; Determining a change reference object according to the element meaning of the independent variable and the element meaning of the dependent variable; The independent variable and the dependent variable are fitted according to the change reference object, and the change trend of the relationship between the independent variable and the dependent variable is determined according to the fitting result.

5. The system according to claim 1, characterized in that Also includes visualization modules, The visualization module is connected to the data processing module, the data identification module, and the data mining module respectively, and is used to analyze the output results of the data processing module, the data identification module, and the data mining module, determine the data characteristics of each output result, and establish a view for viewing based on the data characteristics.

6. The system according to claim 1, characterized in that When the data mining module generates a data analysis report based on the interaction, it is used to: Get the report template; Analyze the report template to determine the types of items included in the report; Determine the corresponding key variables according to the project type; According to the corresponding key variables, determine the interaction between each corresponding key variable and the remaining key variables; According to the interaction between each corresponding key variable and the other key variables, the project conclusion is determined, and a data analysis report is generated according to the project conclusion and the report template.

7. The system according to claim 1, characterized in that The data identification module includes a multi-factor analysis unit. When the data identification module analyzes the key variables according to a preset model and determines the relationship change trend between the key variables, the multi-factor analysis unit is used to: Analyze the key variables and determine the variable types of the key variables; According to the types of variables, several regression models are established; The key variables are analyzed using the several regression models to determine the changing trends of the relationships between the key variables.

8. The system according to any one of claims 5 to 7, characterized in that: It also includes an interaction module connected to the visualization module. After the visualization module establishes a view, the view type is sent to the interaction module. The interaction module determines the interaction function according to the view type and establishes a touch unit according to the interaction function.

9. The system according to claim 5, characterized in that It also includes a cloud platform connected to the visualization module and the data mining module, and is used to store the views established by the visualization module and the data analysis report generated by the data mining module to achieve data sharing.

10. An intelligent data mining method for clinical research transformation, characterized in that: include: Acquire medical data and perform data processing on the medical data to obtain usable medical data that is completely consistent with the knowledge nodes in the preset medical atlas; Analyze the available medical data to determine key variables in the available medical data, and analyze the key variables according to a preset model to determine the relationship change trend between the key variables; The key variables are used to characterize data items in medical data sets that have an impact on disease diagnosis, treatment or prognosis, including genetic markers and biomarker levels of patients; The relationship change trend is used to characterize the change trend between key variables over time or other factors; Determine the interaction between key variables according to the relationship change trend, and generate a data analysis report according to the interaction; the interaction is used to characterize the mutual influence or synergy between two or more key variables; The data analysis report is used to characterize the document generated based on the data analysis results, including the analysis of relationship change trends and the evaluation of interactions.

Images