Diagnostic dose key information extraction system based on multi-agent architecture

Abstract

The invention discloses a diagnostic drug dose key information extraction system based on a multi-agent architecture, which comprises a diagnostic drug dose key information extraction system and a diagnostic drug dose key information extraction system, the diagnostic drug dose key information extraction system comprises a data acquisition and preprocessing module, a natural language processing and information extraction module, a medical knowledge graph construction module, a diagnosis and drug dose recommendation module, a user interaction and feedback module and a safety management and privacy protection module. According to the system, key information about diagnosis and drug dosage can be automatically extracted, analyzed and processed from complex medical data, the system provides accurate and personalized diagnosis and drug dosage suggestions for patients through cooperative work of a plurality of intelligent modules, and meanwhile, the safety and privacy of the data are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of medical systems, and more specifically to a diagnostic drug dosage key information extraction system based on a multi-agent architecture. Background Art

[0002] Health and medical big data covers the entire life cycle of a person, involving the collection and aggregation of data on medical services, disease control, health protection, food safety, health care, etc., and is an important basic strategic resource for the country. The application and development of health and medical big data will bring about profound changes in the health and medical model, which will help stimulate the motivation and vitality of deepening the reform of the medical and health system, improve the efficiency and quality of health and medical services, expand resource supply, and continuously meet the multi-level and diversified health needs of the people, which will help cultivate new formats and economic growth points.

[0003] Drug therapy is an important means of treating diseases. When conducting drug therapy, the drug dosage is more important. Doctors need to adjust the dosage according to the patient's condition when prescribing drugs. The work intensity is high and requires strong medical experience support. Therefore, a new technical solution is needed to solve this problem. Summary of the invention

[0004] The purpose of the present invention is to provide a diagnostic drug dosage key information extraction system based on a multi-agent architecture, which solves the problem that drug therapy is an important means of disease treatment, and the drug dosage is more important during drug therapy. When prescribing drugs, doctors need to adjust the dosage according to the patient's condition, which has high work intensity and requires strong medical experience support.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a diagnostic drug dosage key information extraction system based on a multi-agent architecture, comprising: a diagnostic drug dosage key information extraction system, wherein the diagnostic drug dosage key information extraction system comprises a data acquisition and preprocessing module, a natural language processing and information extraction module, a medical knowledge graph construction module, a diagnosis and drug dosage recommendation module, a user interaction and feedback module, and a security management and privacy protection module;

[0006] The data acquisition and preprocessing module is responsible for collecting data from various data sources and performing preliminary data cleaning, formatting and normalization to ensure data quality and prepare for subsequent analysis;

[0007] The natural language processing and information extraction module extracts key information related to diagnosis and medicine from unstructured text data;

[0008] The medical knowledge graph construction module integrates information extracted from various data sources to construct a medical knowledge graph, which contains medical knowledge such as diseases, drugs, dosages, interactions, etc., to provide a basis for decision support;

[0009] The diagnosis and drug dosage recommendation module combines the patient's specific condition, physiological parameters, past medical history and other information, and uses machine learning or deep learning models to provide patients with personalized diagnosis suggestions and drug dosage recommendations;

[0010] The user interaction and feedback module provides a user-friendly interface, enabling doctors or patients to easily input queries, view diagnostic suggestions and drug dosage recommendations, and collect user feedback to optimize system performance;

[0011] The security management and privacy protection module ensures the security of the system and protects the privacy of patient data from being leaked.

[0012] As a preferred embodiment of the present invention, the data acquisition module in the data acquisition and preprocessing module collects data from electronic medical records, laboratory information, patient monitoring, medical images and external data sources, and the preprocessing module performs data cleaning, data formatting, data normalization and data integration operations.

[0013] As a preferred embodiment of the present invention, the natural language processing and information extraction module includes text preprocessing, named entity recognition, relationship extraction module and dosage information accurate extraction module, and the natural language processing and information extraction module also includes an integration and verification module.

[0014] As a preferred embodiment of the present invention, the medical knowledge graph construction module constructs a disease ontology graph, a drug ontology graph, a patient information graph, a diagnostic process graph, a drug dosage guide graph and a clinical pathway graph.

[0015] As a preferred embodiment of the present invention, the diagnosis and drug dosage recommendation module includes a diagnosis reasoning module, a dosage calculation module and a risk assessment module, wherein the diagnosis reasoning module includes knowledge graph query, reasoning and evaluation, and personalized factor verification, the dosage calculation module includes data integration, dosage calculation and verification and review, and the risk assessment module includes side effect assessment, drug interaction assessment and comprehensive decision support.

[0016] As a preferred implementation of the present invention, the user interaction and feedback module includes a user management module, a feedback collection module and a data storage module.

[0017] As a preferred implementation mode of the present invention, the security management and privacy protection module includes a data encryption module, an access control module and a security audit module.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] The key information extraction system for diagnosis and drug dosage of the present invention includes a data acquisition and preprocessing module, a natural language processing and information extraction module, a medical knowledge graph construction module, a diagnosis and drug dosage recommendation module, a user interaction and feedback module, and a security management and privacy protection module. It can automatically extract, analyze and process key information about diagnosis and drug dosage from complex medical data. The system provides patients with accurate and personalized diagnosis and drug dosage recommendations through the collaborative work of multiple intelligent modules, while ensuring the security and privacy of the data. The data acquisition and preprocessing module is responsible for collecting data from various data sources, and performing preliminary data cleaning, formatting and normalization processing to ensure data quality in preparation for subsequent analysis. The natural language processing and information extraction module extracts key information about diagnosis and drug dosage from unstructured text data such as doctor's notes, patient subjective The medical knowledge graph construction module integrates the information extracted from various data sources to construct a medical knowledge graph, which contains medical knowledge such as diseases, drugs, dosages, interactions, etc., providing a basis for decision support. The diagnosis and drug dosage recommendation module combines the patient's specific condition, physiological parameters, past medical history and other information, and uses machine learning or deep learning models to provide patients with personalized diagnosis suggestions and drug dosage recommendations. The user interaction and feedback module provides a friendly user interface, allowing doctors or patients to easily enter queries, view diagnosis suggestions and drug dosage recommendations, and collect user feedback to optimize system performance. The security management and privacy protection module ensures the security of the system and protects the privacy of patient data from being leaked. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0021] Figure 2 This is a flow chart of the data acquisition and preprocessing module of the present invention;

[0022] Figure 3 This is a flow chart of the natural language processing and information extraction module of the present invention;

[0023] Figure 4 This is a schematic diagram of the module flow for constructing the medical knowledge graph of the present invention;

[0024] Figure 5 This is a schematic diagram of the process flow of the diagnosis and drug dosage recommendation module of the present invention;

[0025] Figure 6 The user interaction and feedback module process intention of the present invention;

[0026] Figure 7 It is a flow chart of the security management and privacy protection module of the present invention. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] See also Figure 1-7 , the present invention provides a technical solution: a diagnostic drug dosage key information extraction system based on a multi-agent architecture, comprising: a diagnostic drug dosage key information extraction system, wherein the diagnostic drug dosage key information extraction system includes a data acquisition and preprocessing module, a natural language processing and information extraction module, a medical knowledge graph construction module, a diagnosis and drug dosage recommendation module, a user interaction and feedback module, and a security management and privacy protection module;

[0029] The data acquisition and preprocessing module is responsible for collecting data from various data sources and performing preliminary data cleaning, formatting and normalization to ensure data quality and prepare for subsequent analysis;

[0030] The natural language processing and information extraction module extracts key information related to diagnosis and medicine from unstructured text data;

[0031] The medical knowledge graph construction module integrates information extracted from various data sources to construct a medical knowledge graph, which contains medical knowledge such as diseases, drugs, dosages, interactions, etc., to provide a basis for decision support;

[0032] The diagnosis and drug dosage recommendation module combines the patient's specific condition, physiological parameters, past medical history and other information, and uses machine learning or deep learning models to provide patients with personalized diagnosis suggestions and drug dosage recommendations;

[0033] The user interaction and feedback module provides a user-friendly interface, enabling doctors or patients to easily input queries, view diagnostic suggestions and drug dosage recommendations, and collect user feedback to optimize system performance;

[0034] The security management and privacy protection module ensures the security of the system and protects the privacy of patient data from being leaked.

[0035] Further improvements, such as Figure 2 As shown: the data acquisition module in the data acquisition and preprocessing module collects data from electronic medical records, laboratory information, patient monitoring, medical images and external data sources, and the preprocessing module performs data cleaning, data formatting, data normalization and data integration operations;

[0036] Electronic medical records contain patients’ medical history, diagnostic information, doctor’s orders, laboratory test results, etc.;

[0037] Laboratory information provides detailed laboratory test results, such as blood tests, urine analysis, etc.;

[0038] Patient monitoring such as heart rate monitors, blood glucose meters, etc., to record the patient’s physiological parameters in real time;

[0039] Medical imaging stores patients’ X-rays, CT, MRI and other medical imaging data;

[0040] External data sources include datasets from medical research institutions, public health databases, etc.

[0041] Data cleaning identifies and corrects missing values, outliers, and duplicate data. Data formatting converts the cleaned data into a unified format for subsequent processing. Data normalization scales the data that requires numerical calculations to a unified range for subsequent model training and use. Data integration integrates data from different data sources into a unified storage system for unified access and processing by subsequent modules.

[0042] Further improvements, such as Figure 3 As shown: the natural language processing and information extraction module includes text preprocessing, named entity recognition, relationship extraction module and dosage information accurate extraction module, and the natural language processing and information extraction module also includes an integration and verification module;

[0043] Text preprocessing divides the text into basic units such as words, phrases or sentences, marks the part of speech for each word or phrase, removes common but meaningless words in the text to reduce noise, and converts the text into a numerical form that can be processed by computers;

[0044] Named entity recognition uses predefined dictionaries and rules to identify entities in text, such as disease names, drug names, etc. It automatically identifies named entities in text by training a large amount of labeled data, trains the named entity recognition model so that it can identify key entities such as disease names and drug names in the text, and applies the model to the text to be processed to extract all relevant named entities;

[0045] The relation extraction module determines the type of relation to be extracted, such as "disease-treatment drug", "drug-recommended dose", etc., trains or applies the relation extraction model, and identifies and extracts relevant relation triplets from the text, such as (disease A, treatment, drug B), (drug B, recommended dose, 10 mg);

[0046] The accurate extraction module of dosage information is usually closely related to the specific drug name and may contain multiple units (such as mg, ml, g, etc.) and possible ranges (such as 5-10 mg). Therefore, it is necessary to further analyze and verify the dosage information based on the relationship extraction;

[0047] The integration and verification module integrates the named entities, relation triples, and dosage information extracted from the text to form a structured data representation and verify the accuracy and consistency of the extracted information. Additional verification steps may be required, such as comparison with the medical knowledge base.

[0048] Further improvements, such as Figure 4 As shown: the medical knowledge graph construction module constructs a disease ontology graph, a drug ontology graph, a patient information graph, a diagnostic process graph, a drug dosage guide graph and a clinical pathway graph;

[0049] The disease ontology map includes disease name, disease symptoms, disease stage, and complications;

[0050] The drug ontology map includes drug name, chemical composition, drug specifications, manufacturer

[0051] The patient information map includes patient ID, name, age, gender, allergy history, genetic disease history, and physiological indicators;

[0052] The diagnostic flowchart includes diagnostic steps, diagnostic tests, and diagnostic criteria;

[0053] The drug dosage guide map includes drug names, dosage units, and patient attributes;

[0054] The clinical pathway map includes clinical pathway steps, diseases, treatment plans, and examination items.

[0055] Further improvements, such as Figure 5 As shown: the diagnosis and drug dosage recommendation module includes a diagnosis reasoning module, a dosage calculation module and a risk assessment module, wherein the diagnosis reasoning module includes knowledge graph query, reasoning and evaluation and personalized factor verification, the dosage calculation module includes data integration, dosage calculation and verification and review, and the risk assessment module includes side effect evaluation, drug interaction evaluation and comprehensive decision support;

[0056] Knowledge graph query uses the medical knowledge graph as a reference to query disease information related to the patient's symptoms, signs, and test results. Through the query, a list of disease candidates that may match the patient's condition can be obtained;

[0057] Reasoning and evaluation: Based on the patient's specific situation, such as the severity, duration, and accompanying symptoms of the symptoms, combined with the disease information in the knowledge graph, logical reasoning and evaluation are performed. Using algorithms and models, such as Bayesian networks, decision trees, support vector machines, or deep learning models, the agent can evaluate the probability of each candidate disease and rule out unlikely diagnoses.

[0058] Personalized factor verification During the reasoning process, the agent will also consider the patient's personalized factors, such as age, gender, genetic factors, previous medical history, allergy history, etc. These factors may affect the probability of disease occurrence and clinical manifestations;

[0059] Data integration collects basic information such as the patient's age, gender, weight, height, renal function, liver function, and key medical information such as current disease status, allergy history, and previous medication history, and obtains detailed information on the target drug from the medical knowledge graph, including the conventional dosage range, dosage adjustment recommendations for different patient groups, and the route and frequency of administration;

[0060] Based on the standards and best practices of the medical industry, develop detailed data specifications, including data formats, naming rules, coding standards, etc., to ensure data consistency during the integration and structuring process. According to the importance and purpose of the data, set key quality indicators such as data completeness, accuracy, consistency ratio, etc. to evaluate data quality. Establish a quality control team composed of data engineers, data analysts and medical experts to be responsible for quality control work throughout the information integration and structuring process. Conduct regular training for quality control team members to enhance their professional skills and quality control awareness. The training content may include medical knowledge, data standards, and quality control methods.

[0061] The dosage calculation is based on the patient's age, weight, renal function and other basic parameters, combined with the recommended dosage range of the drug, to calculate the basic dosage. The basic dosage is adjusted individually considering factors such as the severity of the patient's disease, complications, and interactions with other drugs. For drugs that require continuous monitoring and dosage adjustment, the system can dynamically adjust the dosage based on the patient's real-time physiological parameters.

[0062] Verification and review ensure that the calculated dose is within the safe and effective range of the drug to avoid over- or under-dosing. For high-risk drugs or complex situations, the dose recommendation can be submitted to an expert for manual review;

[0063] Introduce an expert review or two-person review mechanism. The expert review team is composed of doctors or medical experts with rich clinical experience. They have the ability to conduct in-depth analysis and evaluation of treatment records. According to the norms and standards of the medical industry, specific standards and processes for treatment record review are formulated, including requirements for data integrity, accuracy, logical consistency, etc. The treatment records automatically extracted by the system are first screened and sorted to remove obvious errors or redundant information; the screened treatment records are submitted to the expert review team for review, focusing on the accuracy, completeness and consistency of the data with the patient's condition; problems found by experts during the review process are fed back to the data extraction module through the system, and the relevant personnel make corrections and resubmit for review until they pass;

[0064] The two-person review mechanism assigns the work of extracting treatment records to two different staff members, who perform the initial extraction and review work respectively. The initial extractor is responsible for extracting the original treatment records from the medical system and conducting preliminary organization and proofreading; the reviewer is responsible for conducting a comprehensive review of the initially extracted records to ensure the accuracy and completeness of the data. The initial extractor extracts the treatment records from the medical system in accordance with established procedures and specifications. After receiving the initially extracted records, the reviewer checks them one by one, focusing on the accuracy of the data, logical consistency, and whether there are any missing items. After the review is correct, the reviewer confirms and submits the final treatment record data for subsequent analysis.

[0065] Side effect assessment collects known side effect information of drugs from medical literature, clinical trial data and other sources, builds a side effect database, and assesses the patient's sensitivity and potential risk of side effects of specific drugs based on factors such as age, gender, genetic background, and disease status. For each patient, a personalized side effect warning list is generated, and medical staff are reminded to pay attention to monitoring and prevention;

[0066] Drug interaction assessment collects information on interactions between different drugs, especially interactions that may result in changes in drug efficacy, increased toxicity, or reduced efficacy. It compares the patient's current medication regimen with the interaction library to identify potential interaction risks. For discovered drug interaction risks, it provides recommendations for adjusting the medication regimen or increasing monitoring measures.

[0067] Comprehensive decision support comprehensively evaluates the dosage calculation results and risk assessment results to generate a risk score for each patient, reflecting the overall risk level of their medication. Based on the risk score and the patient's specific situation, it provides medical staff with personalized diagnostic advice and drug dosage recommendations, while reminding them to pay attention to potential medication risks.

[0068] Further improvements, such as Figure 6As shown: the user interaction and feedback module includes a user management module, a feedback collection module and a data storage module, designs and implements a user-friendly interactive interface, collects feedback from users on system performance, accuracy, ease of use, etc., and makes optimization suggestions to other modules based on user feedback to continuously improve the system;

[0069] Establish a dedicated user feedback portal, such as online forms, emails, customer service hotlines, etc., to facilitate users and data users to report data quality issues at any time. Establish a problem tracking system within the team to report and track data quality issues among team members. Set a fixed time period (such as weekly, monthly) to actively collect user feedback and internal reports. For urgent or serious data quality issues, establish an immediate response mechanism to ensure that problems are handled quickly. Categorize and organize the collected feedback, such as missing data, format errors, logical contradictions, etc., prioritize the feedback according to the severity and scope of the problem, ensure that the most important problems are solved first, and deeply analyze the root cause of the problem to determine whether it is due to data source problems, processing logic errors or other reasons.

[0070] Further improvements, such as Figure 7 As shown: the security management and privacy protection module includes a data encryption module, an access control module and a security audit module, which encrypts the storage and transmission of sensitive data, implements strict access control policies, ensures that only authorized users can access the data, records and monitors security events of the system, and promptly discovers and responds to potential security threats;

[0071] The data encryption module includes transmission encryption: using SSL / TLS protocol to encrypt all network communications to ensure data transmission security during data collection, communication between agents and data storage;

[0072] Storage encryption: Encrypt sensitive data stored in the database, such as using the AES-256 bit encryption standard;

[0073] The access control module uses role-based access control to define different user roles (such as doctors, system administrators, data analysts) and their corresponding access rights to ensure that only authorized users can access relevant data.

[0074] Principle of least privilege: Assign each user role the minimum set of permissions required to complete its tasks to reduce the risk of data leakage.

[0075] Working principle: The diagnostic and drug dosage key information extraction system includes data collection and preprocessing module, natural language processing and information extraction module, medical knowledge graph construction module, diagnosis and drug dosage recommendation module, user interaction and feedback module and security management and privacy protection module. It can automatically extract, analyze and process key information about diagnosis and drug dosage from complex medical data. The system provides patients with accurate and personalized diagnosis and drug dosage recommendations through the collaborative work of multiple intelligent modules, while ensuring the security and privacy of data. The data collection and preprocessing module is responsible for collecting data from various data sources, and performing preliminary data cleaning, formatting and normalization to ensure data quality in preparation for subsequent analysis. The natural language processing and information extraction module extracts data from unstructured text data such as doctor's notes, patient The chief complaint extracts key information related to diagnosis and drugs, such as disease name, symptom description, drug name, dosage, etc. The medical knowledge graph construction module integrates information extracted from various data sources to construct a medical knowledge graph, which contains medical knowledge such as diseases, drugs, dosages, interactions, etc., providing a basis for decision support. The diagnosis and drug dosage recommendation module combines the patient's specific condition, physiological parameters, past medical history and other information, and uses machine learning or deep learning models to provide patients with personalized diagnostic suggestions and drug dosage recommendations. The user interaction and feedback module provides a friendly user interface, allowing doctors or patients to easily enter queries, view diagnostic suggestions and drug dosage recommendations, and collect user feedback to optimize system performance. The security management and privacy protection module ensures the security of the system and protects the privacy of patient data from being leaked.

[0076] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

[0077] Finally, a few points should be explained: First, in the description of this application, it should be noted that, unless otherwise specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, and can be mechanical or electrical connections, or internal connectivity between two components, or direct connections. "Up", "down", "left", "right", etc. are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may change.

[0078] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A diagnostic drug dosage key information extraction system based on a multi-agent architecture, characterized by: include: A diagnostic drug dosage key information extraction system, wherein the diagnostic drug dosage key information extraction system includes a data acquisition and preprocessing module, a natural language processing and information extraction module, a medical knowledge graph construction module, a diagnosis and drug dosage recommendation module, a user interaction and feedback module, and a security management and privacy protection module; The data acquisition and preprocessing module is responsible for collecting data from various data sources and performing preliminary data cleaning, formatting and normalization to ensure data quality and prepare for subsequent analysis; The natural language processing and information extraction module extracts key information related to diagnosis and medicine from unstructured text data; The medical knowledge graph construction module integrates information extracted from various data sources to construct a medical knowledge graph, which contains medical knowledge such as diseases, drugs, dosages, interactions, etc., to provide a basis for decision support; The diagnosis and drug dosage recommendation module combines the patient's specific condition, physiological parameters, past medical history and other information, and uses machine learning or deep learning models to provide patients with personalized diagnosis suggestions and drug dosage recommendations; The user interaction and feedback module provides a user-friendly interface, enabling doctors or patients to easily input queries, view diagnostic suggestions and drug dosage recommendations, and collect user feedback to optimize system performance; The security management and privacy protection module ensures the security of the system and protects the privacy of patient data from being leaked.

2. According to claim 1, a diagnostic drug dosage key information extraction system based on a multi-agent architecture is characterized by: The data acquisition module in the data acquisition and preprocessing module collects data from electronic medical records, laboratory information, patient monitoring, medical images and external data sources, and the preprocessing module performs data cleaning, data formatting, data normalization and data integration operations.

3. According to claim 1, a diagnostic drug dosage key information extraction system based on a multi-agent architecture is characterized by: The natural language processing and information extraction module includes text preprocessing, named entity recognition, relationship extraction module and dosage information accurate extraction module, and the natural language processing and information extraction module also includes an integration and verification module.

4. The diagnostic drug dosage key information extraction system based on a multi-agent architecture according to claim 1 is characterized by: The medical knowledge graph construction module constructs a disease ontology graph, a drug ontology graph, a patient information graph, a diagnosis process graph, a drug dosage guide graph and a clinical pathway graph.

5. The diagnostic drug dosage key information extraction system based on multi-agent architecture according to claim 1 is characterized by: The diagnosis and drug dosage recommendation module includes a diagnosis reasoning module, a dosage calculation module and a risk assessment module, wherein the diagnosis reasoning module includes knowledge graph query, reasoning and evaluation, and personalized factor verification; the dosage calculation module includes data integration, dosage calculation, verification and review; and the risk assessment module includes side effect assessment, drug interaction assessment, and comprehensive decision support.

6. The diagnostic drug dosage key information extraction system based on multi-agent architecture according to claim 1 is characterized by: The user interaction and feedback module includes a user management module, a feedback collection module and a data storage module.

7. The diagnostic drug dosage key information extraction system based on multi-agent architecture according to claim 1 is characterized by: The security management and privacy protection module includes a data encryption module, an access control module and a security audit module.

Images