Medical knowledge base verification method and system, medium and equipment

Abstract

The invention provides a medical knowledge base verification method and system, a medium and equipment, and the method comprises the steps: obtaining the feedback of a medical professional and a business rule provided by an expert, carrying out the verification and cleaning of structural data corresponding to diagnosis and treatment case data based on the feedback and the business rule, and generating the verified structural data; according to the verified structured data, identifying and extracting a plurality of first entities, determining a logical relationship among the plurality of first entities, and constructing a diagnosis and treatment case knowledge base; and verifying the diagnosis and treatment case knowledge base through a machine learning algorithm to generate a verified diagnosis and treatment case knowledge base. According to the invention, the efficiency of medical knowledge base verification is obviously improved; a large amount of data can be processed more quickly through automatic verification, and the time and labor cost required by manual verification are reduced; and in combination with a machine learning algorithm, the system can identify potential logic errors, inconsistency and abnormal data, so that the verification accuracy is improved, and error omission caused by human negligence is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of medical knowledge base verification, and in particular, to a medical knowledge base verification method, system, medium and equipment. Background Art

[0002] In the medical field, the accuracy and reliability of the knowledge base are crucial for clinical decision-making, disease diagnosis, and treatment plan formulation. Traditional medical knowledge base verification methods mostly rely on manual review, which has several significant defects. First, manual review is inefficient and difficult to cope with the growing amount of medical data and knowledge base size. With the continuous increase in data sources such as electronic health records (EHRs), medical journals, clinical guidelines, and expert opinions, the time and human resources required for manual review have increased dramatically, resulting in a longer verification cycle and difficulty in timely detection and correction of errors.

[0003] Secondly, manual audits are prone to missing errors. Due to limited human attention, long or intensive audit work can easily lead to fatigue and distraction, which in turn increases the risk of missing errors. In addition, differences in knowledge background and experience between different auditors may also lead to different judgment criteria for the same problem, affecting the consistency of verification results.

[0004] In addition, traditional verification methods are difficult to adapt to the dynamic update of medical knowledge. With the continuous progress of medical research and the emergence of new clinical practices, the medical knowledge base needs to be continuously updated and improved. However, manual review is difficult to track and reflect the latest medical research results and clinical practices in a timely manner, causing the knowledge base to lag behind actual needs. Summary of the invention

[0005] In view of the defects in the prior art, the purpose of the present invention is to provide a medical knowledge base verification method, system, medium and equipment.

[0006] According to a medical knowledge base verification method provided by the present invention, the method includes: obtaining feedback and business rules provided by medical professionals, and based on the feedback and the business rules, verifying and cleaning the structured data of the diagnosis and treatment case to generate verified structured data; identifying and extracting multiple entities from the verified structured data, determining the logical relationship between the multiple entities, and constructing a diagnosis and treatment case knowledge base; verifying the diagnosis and treatment case knowledge base through a machine learning algorithm to generate a verified diagnosis and treatment case knowledge base.

[0007] It is not difficult to understand that this application can effectively remove errors, redundancies, and inconsistencies in the data and significantly improve the quality and accuracy of the data by obtaining feedback and business rules from medical professionals and strictly verifying and cleaning the structured data of medical cases. By identifying and extracting multiple entities from the verified high-quality structured data and determining the logical relationships between these entities, a more accurate and comprehensive medical case knowledge base can be constructed. The knowledge base not only contains rich medical information, but also presents it in a structured form, which is convenient for computer processing and analysis, thereby enhancing its value in practical applications. By verifying the medical case knowledge base through machine learning algorithms, an automated and intelligent verification process can be achieved. Machine learning algorithms can learn and identify patterns, anomalies, and potential problems in data, thereby further improving the accuracy and reliability of the knowledge base.

[0008] Furthermore, the verification and cleaning of the structured data of the medical cases based on the feedback and the business rules includes: forming data verification rules according to the feedback and business rules; converting the data verification rules into executable verification logic or scripts; using the verification logic or scripts to verify the structured data of the medical cases; for structured data that fails the verification, recording error information, and marking, correcting or eliminating it as needed.

[0009] Furthermore, based on the feedback and business rules, forming data verification rules includes: collecting feedback from external users, experts or systems, analyzing the feedback, identifying problem types, error patterns and improvement suggestions; understanding business rules and standards related to the medical industry; defining data verification standards based on feedback and business rules; converting the defined data verification standards into data verification rules; verifying the data verification rules, and adjusting and optimizing the data verification rules based on the verification results; documenting the formed data verification rules; regularly reviewing and updating the data verification rules; converting the data verification rules into executable verification logic or scripts includes: parsing the data verification rules extracted from the feedback and business rules; abstracting the parsed data verification rules into logical expressions or conditional statements that can be understood by computers; writing verification scripts or logic codes based on the abstracted logical expressions or conditional statements; parameterizing fixed values ​​in data verification rules; writing test cases, testing verification scripts, and recording verification results; optimizing verification scripts based on test results and fixing problems; integrating the written and tested verification scripts into the medical knowledge base verification system; and writing verification script usage documents and rule descriptions.

[0010] It is not difficult to understand that this application can significantly improve the accuracy and completeness of data by collecting feedback and business rules from medical professionals, forming detailed data verification rules, and strictly verifying and cleaning the structured data of diagnosis and treatment cases. This helps to reduce problems such as misdiagnosis and missed diagnosis caused by data errors or inconsistencies, and improve the quality and safety of medical services. Further, this application converts data verification rules into executable verification logic or scripts to realize the automation of the verification process. This not only improves the verification efficiency, but also reduces manual intervention and reduces the risk of human error. At the same time, the diagnosis and treatment case knowledge base is verified by machine learning algorithms, which further improves the intelligent level of verification, can discover potential data patterns or anomalies, and provide more reliable support for medical decision-making. Further, this application has a high degree of flexibility and scalability. By parameterizing the fixed values ​​in the data verification rules, the verification standards can be easily adjusted according to actual needs. In addition, with the continuous development of medical technology and the updating of business rules, new verification rules can be easily added or existing rules can be modified to adapt to new application scenarios and needs. Furthermore, this application documents the data validation rules and regularly reviews and updates the rules, which helps ensure the standardization and consistency of the validation process. At the same time, the validation script usage documentation and rule descriptions provide clear guidance and reference for other developers or maintainers, reducing the system maintenance cost and learning curve.

[0011] Furthermore, the identifying and extracting multiple entities from the verified structured data includes: converting each character of the structured data into a word vector and a word vector to which each character belongs, and concatenating the word vector and the word vector to obtain a concatenated vector; determining a concatenated feature vector by performing feature extraction on the concatenated vector, and obtaining a probability that each character corresponds to a different label based on the concatenated feature vector; determining a label sequence of the structured data based on the probability that each character corresponds to a different label; and determining the entities in the structured data based on the label sequence of the structured data.

[0012] Furthermore, determining the logical relationship between multiple entities includes: using an information extraction algorithm to extract entity attributes of the entities to obtain entity attributes of the entities; mapping the entities to data models of corresponding categories based on the categories of preset data models and the entity attributes of the entities to obtain logical relationships between entities.

[0013] It is not difficult to understand that this application uses the splicing of word vectors and word vectors, as well as feature extraction technology, to accurately identify and extract multiple entities from the verified structured data. This method fully considers the contextual information of the text and improves the accuracy of entity recognition. The attributes of the entities are extracted through the information extraction algorithm, and the entities are mapped to the data model of the corresponding category to construct the logical relationship between the entities. This step makes the information in the diagnosis and treatment case knowledge base richer and more structured, which is convenient for subsequent query and analysis.

[0014] Furthermore, the method for generating the verified diagnosis and treatment case knowledge base includes: verifying each structured data in the diagnosis and treatment case knowledge base through a machine learning algorithm to obtain a corresponding verification score, making a judgment based on the verification score and a preset value, and then concluding that the corresponding structured data is normal data or abnormal data, processing the abnormal data in the diagnosis and treatment case knowledge base, and generating a verified diagnosis and treatment case knowledge base.

[0015] Furthermore, the method for obtaining the preset value includes: first discretizing the continuous floating-point eigenvalues ​​of the medical case data into k integers, and constructing a histogram with a width of k; traversing the medical case data, and accumulating statistics in the histogram based on the discretized values ​​as indexes; and traversing to find the optimal segmentation point based on the discrete values ​​of the histogram, and the optimal segmentation point is the preset value.

[0016] It is not difficult to understand that this application not only constructs a diagnosis and treatment case knowledge base based on verified data, but also innovatively introduces a machine learning algorithm for secondary verification, further ensuring the quality of the knowledge base; by setting a preset value as the verification standard, and using histogram statistics and optimal segmentation point technology to automatically determine the value, the verification process is more scientific and efficient; the final generated verified diagnosis and treatment case knowledge base not only reduces the interference of abnormal data, but also improves the overall quality of medical data, providing more solid data support for clinical decision-making, and helping to improve the level of medical services and patient satisfaction.

[0017] The present invention also provides a medical knowledge base verification system, the system comprising the following modules:

[0018] Module M1: Obtain feedback and business rules provided by medical professionals, and based on the feedback and the business rules, verify and clean the structured data corresponding to the diagnosis and treatment case data to generate verified structured data;

[0019] Module M2: Identify and extract multiple entities from the verified structured data, determine the logical relationship between multiple entities, and build a diagnosis and treatment case knowledge base;

[0020] Module M3: Verify the diagnosis and treatment case knowledge base through machine learning algorithms to generate a verified diagnosis and treatment case knowledge base.

[0021] The present invention also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the medical knowledge base verification method described above are implemented.

[0022] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of the above-mentioned medical knowledge base verification method when executed by the processor.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] The present invention significantly improves the efficiency of medical knowledge base verification through automated verification processes; compared with traditional manual review methods, automated verification can process large amounts of data more quickly, reducing the time and labor costs required for manual review; combined with machine learning algorithms, the system can identify potential logical errors, inconsistencies and abnormal data, improve the accuracy of verification, and reduce errors and omissions caused by human negligence; at the same time, the present invention has a built-in feedback mechanism that can collect feedback from medical professionals for verification, correction and updating of verification rules; this dynamic optimization process ensures that the verification rules are always consistent with the latest medical knowledge and industry standards, thereby improving the timeliness and reliability of the knowledge base. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0026] Figure 1 is a flow chart of the medical knowledge base verification method of the present invention;

[0027] Figure 2 A schematic diagram of the logical relationship between multiple entities in the embodiment;

[0028] Figure 3 This is a flow chart of extracting entity attributes from entities in structured data using an information extraction algorithm in this embodiment;

[0029] Figure 4 This is a flow chart of verifying the diagnosis and treatment case knowledge base through a machine learning algorithm in this embodiment;

[0030] Figure 5 A flowchart of a method for determining a preset value in an embodiment;

[0031] Figure 6This is a system block diagram of the medical knowledge base verification system of the present invention. DETAILED DESCRIPTION

[0032] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0033] The present application provides a medical knowledge base verification method, which includes: obtaining feedback and business rules provided by medical professionals, and based on the feedback and the business rules, verifying and cleaning the structured data of medical cases to generate verified structured data; identifying and extracting multiple entities from the verified structured data, determining the logical relationship between the multiple entities, and constructing a medical case knowledge base; verifying the medical case knowledge base through a machine learning algorithm to generate a verified medical case knowledge base.

[0034] It is not difficult to understand that this application, by integrating the feedback of medical professionals and business rules, deeply verifies and cleans the structured data of diagnosis and treatment cases, ensuring the accuracy and completeness of the data, which not only effectively improves the data quality, but also lays a solid foundation for the subsequent construction of a diagnosis and treatment case knowledge base. Furthermore, this application also uses a machine learning algorithm to further verify the constructed diagnosis and treatment case knowledge base; it not only improves the accuracy and reliability of the knowledge base, but also enhances its practicality and robustness in practical applications.

[0035] like Figure 1 As shown, a medical knowledge base verification method includes:

[0036] Step S0: Collect and standardize the diagnosis and treatment case data.

[0037] Specifically, information from electronic health records (EHRs), medical journals, clinical guidelines, and expert opinions is collected; medical case data includes patient demographic information, symptoms, signs, laboratory test results, imaging reports, diagnoses, treatment plans, and follow-up results. A medical coding system is used to standardize case data for retrieval and analysis.

[0038] Step S1: Obtain external input feedback and business rules, and based on the feedback and the business rules, verify and clean the structured data of the diagnosis and treatment cases to generate verified structured data.

[0039] Specifically, business rules can be some fixed verification standards, for example, checking whether the data is complete, ensuring that all required fields are filled in and there are no omissions; checking data consistency, ensuring that the data format and unit are correct, such as date format and measurement unit.

[0040] Since some inspection rules may vary according to different time, region, regulations, etc., feedback can be some flexible verification standards. For example, data range standards, check whether the data falls within a reasonable range, such as blood pressure values ​​should not exceed the normal physiological range of the human body. Compliance standards, check whether the data complies with relevant laws, regulations and industry standards, such as privacy protection regulations.

[0041] Before verifying and cleaning the structured data of diagnosis and treatment cases, it is necessary to convert the feedback and the business rules into data verification rules that can be understood and executed by computers, that is, logical expressions or conditional statements that can be executed by computers, parse the data verification rules, abstract the logical relationship and conditional judgment, and write verification scripts or logic codes based on this, and use the written verification logic or scripts to verify the structured data of diagnosis and treatment cases one by one. For example, clearly define the fields that need to be verified in the structured data of diagnosis and treatment cases, such as basic patient information (name, gender, age), diagnosis information (disease name, diagnosis time), treatment information (treatment plan, medication record), and examination results. Secondly, set the verification standards to verify the integrity, format, range, logical relationship, and data consistency of the fields. Thirdly, integrate the internal data quality standards of the hospital (such as data integrity, accuracy, and timeliness requirements) into the verification rules, and consider the requirements of laws and regulations, such as confidentiality of medical information and data desensitization, to ensure that the verification rules do not violate relevant laws and regulations.

[0042] During the verification process, the system will automatically check whether the data complies with the defined verification rules, eliminate the data that fails the verification, and retain the data that passes the verification, thereby obtaining the verified structured data.

[0043] Step S2: Identify and extract multiple entities from the verified structured data, determine the logical relationship between the multiple entities, and build a diagnosis and treatment case knowledge base.

[0044] Specifically, identifying and extracting multiple entities from the verified structured data includes: converting each character of the structured data into a word vector and a word vector to which each character belongs, and concatenating the word vector and the word vector to obtain a concatenated vector. Performing feature extraction on the concatenated vector, determining a concatenated feature vector, and obtaining the probability that each character corresponds to a different label based on the concatenated feature vector. Determining a label sequence of the structured data based on the probability that each character corresponds to a different label. Determining entities in the structured data based on the label sequence of the structured data.

[0045] Specifically, Figure 2 As shown, determining the logical relationship between multiple entities includes: extracting entity attributes of entities in structured data through an information extraction algorithm, and mapping the entities to data models of corresponding types according to a preset data model and the extracted entity attributes. Among them, the preset data models include: disease-drug data model, diagnosis-treatment method data model, etc. For example, for the sentence "cold needs to take cold medicine", the entities that can be extracted are: "cold", "cold medicine", and the entity attributes are: "disease = cold" and "drug = cold medicine". Thus, it is mapped to the corresponding disease-drug data model, and the relationship between the two is that cold medicine can treat colds. Based on the entity mapping, according to the type of model, the relationship between two entities can be directly obtained. These relationships can be directed (such as "patient→receive treatment→drug") or undirected (such as "drug-interaction-another drug"), depending on the actual application scenario and needs.

[0046] like Figure 3 As shown in the figure, the process of extracting entity attributes from entities in structured data through information extraction algorithms includes:

[0047] S201: Perform dependency analysis on the input entities based on the transfer dependency analysis to find the core words of the structured data.

[0048] S202: recursively search for all words that are parallel to the core word to form a predicate list.

[0049] S203: For each predicate, find its subject and object according to the dependency relationship to form an information tuple.

[0050] S204: For an information tuple without a subject, find a parent tuple that is parallel to its predicate and has a subject. If the parent tuple exists, set the subject of the information tuple as the subject of the parent tuple. If not found, do not process it.

[0051] S205: Return the extraction result and obtain the entity attributes of the entity.

[0052] Specifically, the above content performs dependency analysis on entities through transfer dependency analysis, thereby determining entity attributes of entities, and determining the relationship between entities based on the attributes of multiple entities, that is, determining the logical relationship between multiple entities.

[0053] It is not difficult to understand that this application realizes efficient recognition of entities in text by converting each character of structured data into word vectors and word vectors, and performing splicing and feature extraction, combining character-level and word-level information to improve the accuracy and robustness of entity recognition. At the same time, the label sequence of each character is determined based on the probability model, which further refines the recognition of entity boundaries and ensures the accuracy of entity extraction.

[0054] Furthermore, the present application conducts an in-depth analysis of the dependency relationships between entities through information extraction algorithms, especially transfer-based dependency analysis. It not only identifies the core words of the entities, but also constructs a complete predicate list and information tuples through recursive search and dependency analysis, thereby accurately extracting the attributes of the entities. It not only considers the direct attributes of the entities, but also solves the problem of missing subjects in subject-less sentences by finding parallel relationships and parent tuples, thereby ensuring the comprehensiveness and accuracy of the logical relationships between entities.

[0055] Furthermore, the data model preset in this application provides a clear framework for the mapping of entities and relationships. By mapping entities and their attributes to corresponding data models, standardization and normalization of relationships between entities are achieved, which not only simplifies the relationship determination process, but also improves the versatility and scalability of the knowledge base.

[0056] Step S3: Verify the diagnosis and treatment case knowledge base through a machine learning algorithm to generate a verified diagnosis and treatment case knowledge base.

[0057] Specifically, Figure 4 As shown in the figure, the verification of the diagnosis and treatment case knowledge base through machine learning algorithms includes:

[0058] Step S301: extract each structured data item from the diagnosis and treatment case knowledge base, wherein the structured data item includes the patient's diagnosis information, treatment records, and examination results.

[0059] Step S302: pre-process the data, including cleaning, conversion and feature extraction. Select a suitable machine learning algorithm according to the verification target and data type. In the verification of the diagnosis and treatment case knowledge base, the algorithm includes:

[0060] Classification algorithm: used to determine whether a data item belongs to a specific category, such as whether a diagnosis is correct or a treatment plan is reasonable. Classification algorithms include logistic regression, support vector machine (SVM), decision tree, and random forest.

[0061] Anomaly detection algorithm: It is used to identify abnormal data items that are significantly different from the majority of data. Anomaly detection algorithms include Isolation Forest and Local Outlier Factor (LOF).

[0062] Step S303: Use a portion of data with known labels as a training set to train the machine learning model. The labels are obtained by expert annotation, historical data comparison, or automatic rule generation. During the training process, the parameters and structure of the model are continuously adjusted to optimize the performance of the model.

[0063] Step S304: Input the remaining unlabeled data into the trained model and let the model predict or evaluate each data item. The model outputs a verification score or prediction label for each data item based on the knowledge it has learned.

[0064] Step S305: Determine whether each data item is normal data or abnormal data by comparing the verification score with the preset value. The setting of the preset value needs to be adjusted according to actual business needs and model performance.

[0065] Step S306: For data items judged to be abnormal, further processing is performed, including manual review, data correction or data elimination operations, to obtain a verified and processed diagnosis and treatment case knowledge base, in which the data items are more accurate and reliable.

[0066] Through the above steps, machine learning algorithms played an important role in verifying the diagnosis and treatment case knowledge base and improved the quality and reliability of the data.

[0067] Specifically, Figure 5 As shown, the method for determining the preset value in step S305 includes:

[0068] Step S3051: discretize the continuous floating-point eigenvalues ​​of the diagnosis and treatment data set into k integers, and construct a histogram with a width of k;

[0069] Step S3052: when traversing the diagnosis and treatment data, the statistics are accumulated in the histogram according to the discretized values ​​as indexes. After traversing the data once, the histogram accumulates the required statistics;

[0070] Step S3053: traverse and search for the optimal segmentation point according to the discrete values ​​of the histogram, and the optimal segmentation point is the preset value.

[0071] Step S4: Based on the verified diagnosis and treatment case knowledge base, relevant content of the diagnosis and treatment case is displayed on the client.

[0072] Specifically, the client is the interface or application that the user uses to interact with the system. Depending on the usage scenario, the client can take many forms, including:

[0073] Web client: An online platform or website accessed through a web browser that can be used by users without installing additional software and is suitable for multiple devices and operating systems.

[0074] Mobile application (App): An application designed for mobile devices such as smartphones or tablets to provide a more convenient and personalized user experience.

[0075] Desktop Software: Software programs installed on personal computers, usually with richer features and higher processing performance.

[0076] The relevant content of the diagnosis and treatment cases displayed on the client should be customized and displayed according to the needs and permissions of the user role (such as doctor, nurse, patient or family member). The relevant content includes:

[0077] Patient basic information: such as name, gender, age, contact information, and ID number (desensitized), to ensure that medical staff can accurately identify the patient.

[0078] Diagnostic information: including disease name, diagnosis time, and diagnosis basis, to help medical staff understand the patient's condition and diagnosis results.

[0079] Treatment records: Detailed records of the patient's treatment process, including treatment plan, medication (drug name, dosage, usage, medication time, etc.), treatment steps, and treatment effect evaluation, to provide treatment reference for medical personnel.

[0080] Examination results: Display the patient's various examination results, such as blood tests, imaging tests, and pathological examination results, to assist medical personnel in analyzing and diagnosing the condition.

[0081] Medical advice information: Medical advice given by doctors to patients, including dietary guidance, activity suggestions, and follow-up appointment arrangements, to help patients understand their own condition and care requirements.

[0082] Through a carefully designed client interface and rich content display, the user experience and satisfaction can be greatly improved, and the effective transmission and sharing of medical information can be promoted.

[0083] like Figure 6 As shown, a medical knowledge base verification system is used to execute the above-mentioned medical knowledge base verification method, and the system includes: a data acquisition and preprocessing module 1, a feedback and verification rule update module 2, a rule engine module 3, a system integration and coordination module 4, a data management and analysis module 5, a deployment and maintenance module 6, a machine learning module 7, and a user interface module 8.

[0084] The feedback and verification rule update module 2 collects feedback from medical personnel, verifies, corrects and updates the verification rules, implements a feedback loop mechanism, and optimizes the rule engine and machine learning model using user feedback; automatically adjusts and optimizes the rule engine and machine learning model by collecting feedback from medical professionals; and transmits the updated verification rules back to the rule engine module 3.

[0085] Specifically, the feedback and verification rule update module 2 focuses on building a closed-loop feedback system, which is responsible for collecting feedback from medical experts and using it as a benchmark to verify, correct, and iteratively update existing verification rules. This mechanism not only ensures that the rule engine and machine learning model can continue to absorb cutting-edge knowledge, but also achieves precise tuning of both through automated means. Whenever feedback is received from medical professionals, the module will process it immediately and seamlessly connect the optimized verification rules back to the rule engine module 3, thus forming a continuous cycle of improvement process aimed at maximizing the accuracy and adaptability of the system.

[0086] The rule engine module 3 defines and executes a series of business rules to detect logical errors and inconsistencies in the knowledge base; develops a configurable rule management interface that allows medical experts to define and update verification rules; transmits test results to the machine learning module 7 and feedback mechanism; evaluates and operates data according to predefined business rules; provides the ability to define, execute and manage rules through the use of tools; the rule engine reads and parses the rule file, and then applies it to the data to perform corresponding logical judgments and operations.

[0087] Specifically, the medical knowledge base automatic verification system uses the rule engine module 3 to conduct in-depth verification and cleaning of the structured data corresponding to the diagnosis and treatment case data to ensure the accuracy, consistency and integrity of the data. Define and execute business rules to detect logical errors and inconsistencies in the medical knowledge base; the rule engine module 3 includes a data verification submodule, a rule consistency submodule, a knowledge update submodule and a compliance check submodule, wherein the data verification submodule verifies the integrity, consistency, scope and logic of the data; the rule consistency submodule detects the consistency of the rules themselves; the knowledge update submodule monitors the update status of the data; and the compliance check submodule checks whether the data complies with laws, regulations and standards.

[0088] The machine learning module 7 applies a machine learning algorithm to analyze the medical knowledge base data, identify patterns and trends, and predict potential errors; uses anomaly detection and prediction models, including isolation forest or time series analysis, to identify anomalies and prediction errors; transmits the predicted potential errors to the feedback mechanism and user interface, verifies the diagnosis and treatment case knowledge base through a machine learning algorithm, and processes the abnormal data in the diagnosis and treatment case knowledge base based on the verification results to generate a verified diagnosis and treatment case knowledge base. The machine learning algorithm is trained based on the diagnosis and treatment case data set, and the corresponding verification score is obtained by inputting the medical data into the trained machine learning algorithm, and comparing it with the preset value to determine whether the medical data is normal data or abnormal data; the preset value is obtained by first discretizing the continuous floating-point eigenvalues ​​of the diagnosis and treatment data set into integers, constructing a histogram, traversing the diagnosis and treatment data, accumulating statistics in the histogram, and then traversing to find the optimal segmentation point.

[0089] Through automatic verification and continuous optimization, this system solves the problems of error detection and verification rule updating in the maintenance of the medical knowledge base, and improves the accuracy and reliability of the knowledge base.

[0090] It should be understood that the embodiments of the present invention can be implemented in hardware, software or a combination thereof. Software can be stored in a volatile or non-volatile storage device (regardless of whether the data is deletable or rewritable) like a ROM, in a memory like a RAM, a memory chip, a device or an integrated circuit, or in a storage medium on which data can be optically or magnetically recorded and from which data can be read by a machine (e.g., a computer), such as a compact disc (CD), a digital versatile disc (DVD), a disk, or a tape. Moreover, the embodiments of the present invention can be implemented in a computer or portable terminal having a controller and a memory, and the memory is an example of a non-temporary computer-readable storage medium suitable for storing one or more programs including commands for implementing the embodiments of the present invention. Therefore, the present invention includes a program having a code for implementing the device or method defined by the claims, and a machine-readable non-temporary storage medium storing the program. The program can be transferred electrically by a medium such as a communication signal transmitted via a wired or wireless connection, and the program and its equivalent are included in the present invention.

[0091] The device may receive a program from a program providing device connected via a cable or wirelessly and store the program. The program providing device may include a program containing commands for implementing the embodiments of the present invention, a memory for storing information required by the embodiments of the present invention, a communication module for communicating with the mobile device via a cable or wirelessly, and a controller for transmitting the program to the device automatically or upon request of the device.

[0092] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A medical knowledge base verification method, characterized in that: The method comprises: Obtaining external input feedback and business rules, and based on the feedback and the business rules, verifying and cleaning the structured data of the diagnosis and treatment cases to generate verified structured data; Identify and extract multiple entities from the verified structured data, determine the logical relationship between multiple entities, and build a diagnosis and treatment case knowledge base; The diagnosis and treatment case knowledge base is verified through a machine learning algorithm to generate a verified diagnosis and treatment case knowledge base.

2. The medical knowledge base verification method according to claim 1, characterized in that: The checking and cleaning of the structured data of the diagnosis and treatment case based on the feedback and the business rules includes: According to the feedback and business rules, data verification rules are formed; Convert the data verification rules into executable verification logic or script; Verifying the structured data of the diagnosis and treatment case using the verification logic or script; For structured data that fails verification, the error information is recorded and marked, corrected, or deleted as needed.

3. The medical knowledge base verification method according to claim 2, characterized in that: Converting the data verification rules into executable verification logic or scripts includes: Parse the data verification rules and abstract the parsed data verification rules into logical expressions or conditional statements that can be understood by computers; Write verification scripts or logic codes based on the abstracted logical expressions or conditional statements.

4. The medical knowledge base verification method according to claim 1, characterized in that: The identifying and extracting multiple entities from the verified structured data includes: Convert each character of the structured data into a word vector and a word vector to which each character belongs, and concatenate the word vector and the word vector to obtain a concatenated vector; Determine a splicing feature vector by performing feature extraction on the splicing vector, and obtain the probability that each character corresponds to a different label according to the splicing feature vector; Determining a label sequence of the structured data according to the probability that each character corresponds to different labels; An entity in the structured data is determined based on a tag sequence of the structured data.

5. The medical knowledge base verification method according to claim 1, characterized in that: Determining the logical relationship between the multiple entities includes: Using an information extraction algorithm to extract entity attributes of the entity to obtain entity attributes of the entity; According to the categories of the preset data model and the entity attributes of the entities, the entities are mapped to the data models of the corresponding categories to obtain the logical relationships between the entities.

6. The medical knowledge base verification method according to claim 1, characterized in that: The method for generating the verified diagnosis and treatment case knowledge base includes: Each structured data in the diagnosis and treatment case knowledge base is verified through a machine learning algorithm to obtain a corresponding verification score. A judgment is made based on the verification score and the preset value to determine whether the corresponding structured data is normal data or abnormal data. The abnormal data in the diagnosis and treatment case knowledge base is processed to generate a verified diagnosis and treatment case knowledge base.

7. The medical knowledge base verification method according to claim 6, characterized in that: The method for obtaining the preset value includes: First, the continuous floating-point eigenvalues ​​of the diagnosis and treatment case data are discretized into k integers, and a histogram with a width of k is constructed; Traverse the diagnosis and treatment case data and accumulate statistics in the histogram based on the discretized values ​​as indices; According to the discrete values ​​of the histogram, the optimal segmentation point is traversed and found, and the optimal segmentation point is the preset value.

8. A medical knowledge base verification system, characterized in that: The system includes the following modules: Module M1: Obtain feedback and business rules provided by medical professionals, and based on the feedback and the business rules, verify and clean the structured data corresponding to the diagnosis and treatment case data to generate verified structured data; Module M2: Identify and extract multiple entities from the verified structured data, determine the logical relationship between multiple entities, and build a diagnosis and treatment case knowledge base; Module M3: Verify the diagnosis and treatment case knowledge base through machine learning algorithms to generate a verified diagnosis and treatment case knowledge base.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the medical knowledge base verification method described in any one of claims 1 to 7 are implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by a processor, the steps of the medical knowledge base verification method described in any one of claims 1 to 7 are implemented.

Images