Medical content intelligent cooperative processing system and processing method

Abstract

The invention relates to the field of medical data processing, in particular to a medical content intelligent cooperative processing system and method. The system comprises a medical content intelligent cooperative processing system which is provided with a data acquisition module for acquiring medical field unstructured original data, a preprocessing module for format unification and quality filtering, and an intelligent processing module for performing semantic understanding and multi-dimensional tagging processing on the original data. The collaborative optimization module performs manual verification on a multi-dimensional tagging result and feeds back the result, and the data analysis and generation module performs content analysis and auxiliary creation based on the verified tagging data; the invention further comprises a processing method suitable for the system. The processing method comprises the steps of data acquisition, preprocessing, intelligent tagging, collaborative optimization, value output and the like. According to the method, the effects of effectively processing the unstructured original data in the medical field, realizing intelligent cooperative processing of medical contents, mining medical research hotspots and content gaps, generating medical literature review or science popularization manuscript first draft and the like are achieved.

Description

Technical Field

[0001] This application relates to the field of medical data processing, and in particular to a medical content intelligent collaborative processing system and processing method. Background Technology

[0002] In the medical field, with the rapid development of information technology, medical information has experienced explosive growth. This massive amount of medical data contains rich knowledge and valuable information, possessing immense value for medical research, clinical practice, and medical education. It helps doctors better diagnose diseases and develop treatment plans, and also assists researchers in conducting in-depth medical research, driving the advancement of medical science. However, how to effectively manage, accurately classify, and efficiently retrieve this complex and diverse medical information has become a key issue restricting its further development. Traditional technologies, when processing medical information, mainly rely on rigid classification methods based on single dimensions such as "information source and publication time." Keyword matching is commonly used for retrieving medical information. The processing of medical information lacks intelligent feedback mechanisms, relying primarily on manual processing and judgment. While these conventional methods can meet basic medical information management and retrieval needs to a certain extent, their limitations become increasingly apparent as the complexity, diversity, and specialization of medical information continue to increase. Existing technologies have a fundamental flaw: they cannot handle the complexity, diversity, and specialization of medical information. Keyword matching retrieval methods lack accuracy in medical scenarios, making it difficult to accurately find the information users truly need. Furthermore, existing technologies lack intelligent feedback mechanisms and rely excessively on manual processing, resulting in low processing efficiency and making it difficult to achieve system self-evolution and optimization. Summary of the Invention The purpose of this application is to overcome the above-mentioned technical problems and provide a medical content intelligent collaborative processing system and processing method.

[0003] To achieve the above objectives, the following technical solution is proposed: A medical content intelligent collaborative processing system includes: a data acquisition module for acquiring unstructured raw data in the medical field; a preprocessing module connected to the data acquisition module for format unification and quality filtering; an intelligent processing module connected to the preprocessing module for performing semantic understanding and multi-dimensional tagging processing on the raw data; a collaborative optimization module connected to the intelligent processing module for performing manual verification on the multi-dimensional tagging results and feeding back the verification results to the intelligent processing module; and a data analysis and generation module connected to the collaborative optimization module for performing content analysis and assisting in content creation based on the verified tagged data.

[0004] By adopting the above technical solutions, the data acquisition module acquires unstructured raw data in the medical field, the preprocessing module performs format unification and quality filtering, and the intelligent processing module performs semantic understanding and multi-dimensional tagging processing on the raw data to achieve precise structuring of medical information, solving the problem that existing technologies cannot handle the complexity, diversity, and professionalism of medical information. The collaborative optimization module performs manual verification on the multi-dimensional tagging results and provides feedback on the verification results, realizing closed-loop learning and collaborative optimization of the system, reducing reliance on manual intervention, and enabling the system to self-evolve. The data analysis and generation module performs content analysis and assisted creation based on the verified tagged data, realizing data-driven decision support and empowering medical content strategic planning.

[0005] Preferably, the data acquisition module includes a multi-source data acquisition unit, used to acquire the raw data from medical journal websites, clinical guideline databases, and authoritative medical platforms.

[0006] By adopting the above technical solutions, the multi-source data acquisition unit obtains raw data from medical journal websites, clinical guideline databases, and authoritative medical platforms, which can provide the system with more comprehensive and professional medical information. This helps to build a precise multi-dimensional intelligent tagging system and improve the quality and effectiveness of medical information processing.

[0007] Preferably, the multi-source information acquisition unit is implemented through a distributed crawler architecture, including a timed task scheduling component and a parsing component.

[0008] By adopting the above technical solution and using a distributed crawler architecture to implement a multi-source data acquisition unit, unstructured raw data in the medical field can be obtained from medical journal websites, clinical guideline databases, and authoritative medical platforms. The scheduled task component can collect data according to the set time, and the parsing component can parse and process the collected data, thereby improving the efficiency and accuracy of data acquisition and providing a more comprehensive, timely, and accurate data foundation for subsequent medical content processing.

[0009] Preferably, the intelligent processing module includes: a semantic understanding unit for identifying medical entities and their relationships in the original data; a tag generation unit for generating a structured tag set based on a vector embedding model and a reordering model; and a tag storage unit for temporarily storing the structured tag set and establishing an association index with the original data.

[0010] By adopting the above technical solution, the semantic understanding unit identifies medical entities and their relationships in the original data, the tag generation unit generates a structured tag set based on the vector embedding model and the reordering model, and the tag storage unit temporarily stores the structured tag set and establishes an association index with the original data. This achieves accurate structuring of medical information, solves the problem that existing technologies cannot handle the complexity, diversity and professionalism of medical information, and at the same time achieves a leap from "keyword matching" to "semantic understanding" through semantic understanding, thereby improving the accuracy of retrieval.

[0011] Preferably, the tag generation unit uses a language model and a vector database to work together, determines tag weights through semantic similarity calculation, and performs priority screening of core tags.

[0012] By adopting the above technical solutions, the tag generation unit uses a language model and a vector database to work together, which can leverage the semantic understanding capabilities of the language model and the efficient retrieval capabilities of the vector database. By calculating the tag weights through semantic similarity, the relevance between the tags and the original data can be measured more accurately. By performing priority screening of core tags, important tags can be highlighted, achieving precise structuring of medical information, improving the accuracy and efficiency of tag generation, and thus improving the accuracy of the system's processing of medical information.

[0013] Preferably, the collaborative optimization module includes: a tag display unit, used to synchronously push the multi-dimensional tagging results and the original data to the human-computer interaction interface; and a verification feedback unit, used to receive manual verification instructions and update the tag library and error correction dataset.

[0014] By adopting the above technical solution, the label display unit pushes the multi-dimensional labeling results and the original data to the human-computer interaction interface in a synchronized manner, which is convenient for manual verification. The verification feedback unit receives manual verification instructions and updates the label library and error correction dataset, which can accumulate error correction data, provide high-quality labeled data for optimizing AI models, and realize closed-loop learning and collaborative optimization of the system, reduce reliance on manual intervention, and realize the system's self-evolution.

[0015] Preferably, the verification feedback unit transmits the manually corrected labels and the original error correction data to the intelligent processing module through a standardized interface for real-time optimization of the label generation model parameters.

[0016] By adopting the above technical solution, the system can transmit manually corrected labels and original error correction data to the intelligent processing module through a standardized interface, realize real-time optimization of label generation model parameters, reduce the system's dependence on manual labor, enable the system to evolve itself, and improve the accuracy of subsequent labels.

[0017] Preferably, the data analysis and generation module includes: a trend analysis unit, used to statistically analyze the distribution patterns of tags and identify research hotspots and content gaps in the medical field; and a content generation unit, used to generate initial drafts of medical literature reviews or popular science articles based on selected tags and topic directions.

[0018] By adopting the above technical solutions, the trend analysis unit can identify research hotspots and content gaps in the medical field by statistically analyzing the distribution patterns of tags, thereby enabling data-driven decision support and empowering strategic planning for medical content. The content generation unit can generate initial drafts of medical literature reviews or popular science articles based on selected tags and topics, which can assist in the creation of medical content.

[0019] Preferably, the trend analysis unit uses time series analysis and clustering algorithms to mine the correlation trends of disease types, treatment plans and clinical conclusions in the tag library.

[0020] By adopting the above technical solutions, the data acquisition module of the intelligent collaborative processing system for medical content acquires unstructured raw data in the medical field. After format unification and quality filtering by the preprocessing module, the intelligent processing module performs semantic understanding and multi-dimensional tagging. The collaborative optimization module performs manual verification and provides feedback on the results. The data analysis and generation module performs content analysis and assists in creation based on the verified data. Among them, the trend analysis unit uses time series analysis and clustering algorithms to mine the correlation and change trends of disease types, treatment plans and clinical conclusions in the tag library, which can realize macro-analysis of medical content, accurately identify research trends and potential hotspots, and discover content gaps in the medical field.

[0021] A method for intelligent collaborative processing of medical content, applicable to the intelligent collaborative processing system for medical content as described in any one of claims 1-9, includes the following steps: S1, a data acquisition module acquires unstructured medical information; S2, a preprocessing module performs format unification and quality filtering; S3, an intelligent tagging module generates multi-dimensional structured tags; S4, a collaborative optimization module performs manual verification and error correction data feedback; S5, a value output module generates data analysis results and a draft of the content.

[0022] By adopting the above technical solutions, the data acquisition module obtains unstructured medical information, acquiring data from multiple sources to provide a wide range of data for subsequent processing; the preprocessing module performs format unification and quality filtering, which can standardize the data format and improve its quality; the intelligent tagging module generates multi-dimensional structured tags to achieve precise structuring of medical information and overcome the shortcomings of traditional single classification; the collaborative optimization module performs manual verification and error correction data feedback, which can reduce reliance on manual labor, enable the system to self-evolve, improve tag accuracy, and achieve personalized content recommendation; the value output module generates data analysis results and initial content drafts, which can perform macro-analysis of medical content based on the multi-dimensional tagging system, identify research trends, potential hotspots, and content gaps, and empower medical content strategic planning.

[0023] The beneficial effects of this invention are: 1. The intelligent processing module performs multi-dimensional tagging on raw data, achieving precise structuring of medical information and solving the problem of existing technologies being unable to handle the complexity, diversity, and specialization of medical information. 2. The intelligent processing module introduces semantic understanding, combined with the collaborative optimization module for manual verification and feedback, transforming from "keyword matching" to "semantic understanding," significantly improving retrieval accuracy. 3. The collaborative optimization module integrates a collaborative workflow of "AI preliminary processing + manual professional verification + result feedback optimization," accumulating error correction data, iteratively updating the model, learning preferences, and personalizing adaptations, reducing reliance on manual intervention and enabling the system to self-evolve. 4. The data analysis and generation module performs content analysis based on the verified tagged data, enabling tag co-occurrence analysis, time series analysis, and blank spot identification, expanding application value and empowering medical content strategic planning. Attached Figure Description

[0024] Figure 1 This is a diagram showing the connection relationships between the system modules of the present invention.

[0025] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0026] The technical solutions in the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. The described embodiments are only possible technical implementations of the present invention, but are not limited thereto. Other embodiments obtained by those skilled in the art in conjunction with the embodiments of the present invention without creative effort are also within the protection scope of the present invention.

[0027] Example 1 The intelligent collaborative processing system for medical content provided in this application includes a data acquisition module, a preprocessing module, an intelligent processing module, a collaborative optimization module, and a data analysis and generation module. The data acquisition module is connected to the preprocessing module, the preprocessing module is connected to the intelligent processing module, the intelligent processing module is connected to the collaborative optimization module, and the collaborative optimization module is connected to the data analysis and generation module. This achieves the effect of effective management, accurate classification, and efficient retrieval of medical content because each module works collaboratively in sequence to process and optimize the medical data step by step.

[0028] Specifically, the data acquisition module includes a multi-source data acquisition unit. The multi-source data acquisition unit is characterized by its ability to obtain raw data from multiple authoritative sources, such as medical journal websites, clinical guideline databases, and authoritative medical platforms. In terms of replaceability, data can also be obtained from medical academic conference materials, professional medical forums, and other sources. The multi-source data acquisition unit is implemented through a distributed crawler architecture, which includes a scheduled task component and a parsing component. The scheduled task component executes data acquisition tasks at preset time intervals, ensuring data real-time performance and update frequency; for example, data acquisition can be scheduled for every morning. The parsing component performs preliminary parsing of the acquired data, converting unstructured data into a format suitable for subsequent processing. Regarding replaceability, the scheduled task component can be replaced with an event-triggered scheduling component, which immediately initiates acquisition when a new data update event occurs; the parsing component can also employ different parsing algorithms or tools. The combined logic of the multi-source data acquisition unit is that the scheduled task component triggers data acquisition, and the parsing component parses the acquired data, thus efficiently and accurately obtaining raw medical data from multiple data sources.

[0029] Specifically, the preprocessing module performs format unification and quality filtering. Format unification converts medical data from different sources and in different formats into a unified format, such as converting data in PDF and DOC formats into text format, facilitating subsequent processing. Quality filtering removes noise, duplicate information, and errors from the data, improving data quality. Regarding replaceable features, format unification can employ different conversion tools or algorithms; quality filtering can add more filtering rules, such as evaluating and screening data for completeness and accuracy. The combined effect of the preprocessing modules is to provide high-quality, uniformly formatted data for subsequent intelligent processing, reducing the complexity and errors of subsequent processing.

[0030] Specifically, the intelligent processing module includes a semantic understanding unit, a tag generation unit, and a tag storage unit. The semantic understanding unit identifies medical entities and their relationships in the raw data. It uses natural language processing (NLP) technology to analyze medical text, identify medical entities such as disease names, drug names, and treatment methods, and determine the relationships between them, such as a drug being used to treat a certain disease. For replaceable features, different NLP models or algorithms can be used, such as a deep learning-based semantic understanding model. The tag generation unit generates a structured tag set based on vector embedding and re-ranking models. It converts medical data into vector representations, determines tag weights through semantic similarity calculations, and performs priority filtering of core tags. For example, a language model can work in conjunction with a vector database to compare the similarity of medical text with tags in a tag library, selecting the most relevant tags. For replaceable features, the vector embedding and re-ranking models can use different architectures and parameter settings. The tag storage unit temporarily stores the structured tag set and establishes an index linking it to the raw data. It stores the generated tags in a database and establishes a correspondence between tags and raw data for easy subsequent querying and use. Regarding replaceable features, the tag storage unit can employ different database management systems. The combined logic of the intelligent processing module is as follows: the semantic understanding unit first analyzes the data to determine medical entities and relationships; the tag generation unit generates tags based on this information; and the tag storage unit stores the tags and establishes an index, thereby achieving semantic understanding and multi-dimensional tagging processing of medical data.

[0031] Specifically, the collaborative optimization module includes a label display unit and a verification feedback unit. The label display unit synchronously pushes the multi-dimensional labeling results and raw data to the human-computer interaction interface. It can intuitively display the generated labels and corresponding raw medical data to human verifiers for easy viewing and verification. Regarding replaceable features, the human-computer interaction interface can adopt different design styles and layouts to improve user experience. The verification feedback unit receives human verification instructions and updates the label library and error correction dataset. It transmits the manually corrected labels and raw error correction data to the intelligent processing module through a standardized interface for real-time optimization of the label generation model parameters. For example, when a human verifier finds a label inaccurate, the verification feedback unit feeds back the correct label and related error correction data to the intelligent processing module, which can then optimize the label generation model based on this data. Regarding replaceable features, the standardized interface can adopt different protocols and specifications. The combined effect of the collaborative optimization module is to continuously optimize the label generation model through human verification and feedback, thereby improving the accuracy and quality of the labels.

[0032] Specifically, the data analysis and generation module includes a trend analysis unit and a content generation unit. The trend analysis unit is used to statistically analyze tag distribution patterns and identify research hotspots and content gaps in the medical field. It uses time-series analysis and clustering algorithms to mine the changing trends of correlations between disease types, treatment plans, and clinical conclusions in the tag library. For example, it analyzes the changing trends of treatment plans for a certain disease over different time periods, or discovers the correlations between certain disease types. Regarding replaceable features, the time-series analysis and clustering algorithms can employ different algorithms and parameter settings. The content generation unit is used to generate initial drafts of medical literature reviews or popular science articles based on selected tags and topic directions. It can extract relevant information from the tag library and raw data according to the user's selected tags and topic directions to generate preliminary medical literature reviews or popular science articles. Regarding replaceable features, the content generation unit can employ different text generation models and algorithms. The combined logic of the data analysis and generation module is that the trend analysis unit first analyzes the tag data to identify research hotspots and content gaps, and the content generation unit generates relevant medical content based on this information and the user's topic direction, thereby supporting medical research and popular science.

[0033] The implementation principle of this embodiment is as follows: This intelligent collaborative processing system for medical content collects raw medical data from multiple authoritative channels through the collaborative work of its various modules. After preprocessing to improve data quality, it performs semantic understanding and multi-dimensional tagging. The tags are then continuously optimized through manual verification, and finally, data analysis and content generation are performed. This multi-module collaborative approach effectively solves the problems of existing technologies in processing medical information, such as their inability to handle complexity, diversity, and specialization, insufficient retrieval accuracy, and lack of intelligent feedback. It improves the efficiency of medical information management, classification, and retrieval, providing strong support for medical research, clinical practice, and public education.

[0034] The difference between this embodiment and the previous embodiments lies in the fact that the multi-source data acquisition unit of the data acquisition module adopts a different acquisition architecture. This embodiment uses a centralized acquisition architecture for its multi-source data acquisition unit, which includes a data collection component and a data integration component. The data collection component is responsible for collecting raw medical data from various data sources. It can connect to medical journal websites, clinical guideline databases, and authoritative medical platforms via network interfaces to periodically acquire data. Regarding replaceable features, the data collection component can use different network protocols and interface methods. The data integration component integrates the collected data, removes duplicate data, and performs preliminary format conversion. Regarding replaceable features, the data integration component can use different data cleaning and transformation algorithms.

[0035] The implementation principle of this embodiment is as follows: the multi-source data acquisition unit of the centralized acquisition architecture can also acquire raw medical data from multiple data sources, and improve the quality and availability of data through data integration. Compared with the distributed crawler architecture, the centralized acquisition architecture may be more convenient in terms of data management and maintenance, and can flexibly adjust the acquisition strategy and data processing method according to actual needs. It can also provide effective data support for subsequent medical content processing, solving the problems of existing technologies in medical information acquisition.

[0036] The intelligent collaborative processing method for medical content provided in this application includes the following steps: S1, the data acquisition module acquires unstructured medical information. The multi-source data acquisition unit of the data acquisition module obtains raw medical data from sources such as medical journal websites, clinical guideline databases, and authoritative medical platforms. It utilizes a timed task scheduling component within a distributed crawler architecture to collect data according to preset times, and a parsing component performs preliminary analysis of the acquired data.

[0037] S2, the preprocessing module performs format unification and quality filtering. The preprocessing module converts the collected medical data in different formats into a unified text format, removing noise, duplicate information, and errors to improve data quality.

[0038] S3, the intelligent tagging module generates multi-dimensional structured tags. The semantic understanding unit in the intelligent processing module identifies medical entities and their relationships in the original data, the tag generation unit generates a set of structured tags based on vector embedding and reordering models, and the tag storage unit temporarily stores the set of structured tags and establishes an association index with the original data.

[0039] S4, the collaborative optimization module performs manual verification and error correction data feedback. The label display unit of the collaborative optimization module synchronously pushes the multi-dimensional labeling results and the original data to the human-computer interaction interface. Manual verifiers verify the labels, and the verification feedback unit receives the manual verification instructions and updates the label library and error correction dataset. Through a standardized interface, the manually corrected labels and the original error correction data are transmitted to the intelligent processing module to optimize the label generation model parameters in real time.

[0040] S5 generates data analysis results and initial content drafts through the value output module. The trend analysis unit of the data analysis and generation module statistically analyzes the distribution patterns of tags and identifies research hotspots and content gaps in the medical field. The content generation unit generates initial drafts of medical literature reviews or popular science articles based on selected tags and topic directions.

[0041] The implementation principle of this embodiment is as follows: the intelligent collaborative processing method for medical content sequentially collects, preprocesses, intelligently tags, manually verifies, and analyzes and generates content from medical data. Through clear steps and processes, this method achieves effective management and processing of medical content, solves the problems existing in medical information processing in current technologies, improves the utilization efficiency and value of medical information, and provides a scientific and efficient processing method for research and practice in the medical field.

[0042] Example 2 In addition to using a single LLM, this application can also employ a "dual-brain" coding guidance model. For example, one AI model is dedicated to extracting medical entities (diseases, drugs) from text, while another model focuses on determining the relationships between these entities based on clinical logic and assigning classification codes. Working together, the two can potentially improve the accuracy of processing complex medical record content.

[0043] In the retrieval module, a hybrid retrieval mechanism can be designed. When a user makes a query, the system performs traditional keyword matching based on tags and text, as well as semantic similarity matching based on embedding vectors, in parallel. Finally, the two sets of results are fused and reranked to balance retrieval recall and precision, ensuring relevance while not overlooking results with highly matched keywords.

[0044] The collected medical literature and other information are deconstructed according to preset dimensions (such as disease, symptoms, treatment, and prognosis) and populated into a unified "health profile" template. This highly standardized data structure is very convenient for subsequent comparison, analysis, and trend mining.

[0045] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A medical content intelligent collaborative processing system, characterized in that, include: The data acquisition module is used to acquire unstructured raw data in the medical field; A preprocessing module, connected to the data acquisition module, is used for format unification and quality filtering; An intelligent processing module, connected to the preprocessing module, is used to perform semantic understanding and multi-dimensional tagging processing on the raw data; A collaborative optimization module, connected to the intelligent processing module, is used to perform manual verification on the multi-dimensional labeled results and feed back the verification results to the intelligent processing module. The data analysis and generation module, connected to the collaborative optimization module, is used to perform content analysis and assist in creation based on the verified tagged data.

2. The intelligent collaborative processing system for medical content according to claim 1, characterized in that, The data acquisition module includes a multi-source data acquisition unit, used to obtain the raw data from medical journal websites, clinical guideline databases, and authoritative medical platforms.

3. The intelligent collaborative processing system for medical content according to claim 2, characterized in that, The multi-source information acquisition unit is implemented through a distributed crawler architecture, which includes a timed task scheduling component and a parsing component.

4. The intelligent collaborative processing system for medical content according to claim 1, characterized in that, The intelligent processing module includes: A semantic understanding unit is used to identify medical entities and their relationships in the raw data; The tag generation unit is used to generate a set of structured tags based on the vector embedding model and the reordering model. The tag storage unit is used to temporarily store the structured tag set and establish an association index with the original data.

5. The intelligent collaborative processing system for medical content according to claim 4, characterized in that, The tag generation unit uses a language model and a vector database to work together, and determines the tag weights through semantic similarity calculation and performs priority screening of core tags.

6. The intelligent collaborative processing system for medical content according to claim 1, characterized in that, The collaborative optimization module includes: The tag display unit is used to synchronously push the multi-dimensional tagging results and the original data to the human-computer interaction interface. The verification feedback unit is used to receive manual verification instructions and update the tag library and error correction dataset.

7. The intelligent collaborative processing system for medical content according to claim 6, characterized in that, The verification feedback unit transmits the manually corrected labels and the original error correction data to the intelligent processing module through a standardized interface, which is used to optimize the label generation model parameters in real time.

8. The intelligent collaborative processing system for medical content according to claim 1, characterized in that, The data analysis and generation module includes: The trend analysis unit is used to statistically analyze the distribution patterns of tags and identify research hotspots and content gaps in the medical field. The content generation unit is used to generate initial drafts of medical literature reviews or popular science articles based on selected tags and topic directions.

9. The intelligent collaborative processing system for medical content according to claim 8, characterized in that, The trend analysis unit uses time series analysis and clustering algorithms to mine the correlation trends of disease types, treatment plans and clinical conclusions in the tag library.

10. A method for intelligent collaborative processing of medical content, applicable to the intelligent collaborative processing system for medical content as described in any one of claims 1-9, characterized in that, Includes the following steps: S1. The data acquisition module acquires unstructured medical information; S2, The preprocessing module performs format unification and quality filtering; S3, the intelligent tagging module generates multi-dimensional structured tags; S4. The collaborative optimization module performs manual verification and error correction data feedback. S5. Generate data analysis results and initial draft content through the value output module.

Images