Medical knowledge intelligent consulting system based on knowledge graph

Abstract

This application relates to the field of medical knowledge consulting technology and discloses a knowledge graph-based intelligent medical knowledge consulting system. The system includes a consultation acquisition module, a consultation interpretation module, and a consultation output module, which are sequentially communicatively connected. Through the collaborative design of the consultation acquisition, interpretation, and output modules, this application achieves efficient integration and intelligent consultation of multi-source medical data, thereby improving the efficiency of genetic variation analysis, interpretation accuracy, and prediction accuracy, enabling natural language interaction and visual analysis, and meeting the needs of medical data and its security regulatory requirements.

Description

Technical Field

[0001] The present application relates to the field of medical knowledge consulting technology, and specifically to a medical knowledge intelligent consulting system based on knowledge graphs. Background Art

[0002] With the popularization of gene sequencing technology, the demand for genetic counseling is growing rapidly. The traditional model relying on expert experience is difficult to meet clinical needs. Clinical genetic counseling faces the following problems:

[0003] The interpretation of genetic data is highly dependent on manual labor, resulting in inefficient consultation responses; insufficient integration of multi-source databases for genetic knowledge leads to severe information fragmentation; the lack of a natural language interactive interface makes it difficult for users to use and inconvenient to use; the assessment accuracy of genetic risks is limited and the degree of personalization is insufficient; there is a lack of cross-platform integration capabilities, and data silos are common.

[0004] The Chinese invention patent with authorization announcement number CN116246701B discloses a data analysis device, medium and equipment based on phenotypic terms and variant genes, but the invention performs poorly in medical knowledge consultation, especially genetic knowledge consultation.

[0005] In summary, there is an urgent need for a new technical solution for intelligent medical knowledge consultation based on knowledge graphs. Summary of the Invention

[0006] The purpose of this application is to provide a medical knowledge intelligent consulting system based on knowledge graph to solve the technical problems raised in the above background technology.

[0007] To achieve the above-mentioned purpose, the present application discloses the following technical solutions: a medical knowledge intelligent consultation system based on a knowledge graph, the system comprising a consultation collection module, a consultation interpretation module, and a consultation output module that are sequentially communicatively connected;

[0008] The consultation collection module is configured to: process the collected consultation conditions and corresponding consultation needs of the consultant to generate a corresponding consultation feature set; wherein the consultation conditions are the consultant's medical-related data, the consultation needs are the content of the consultation desired based on the consultation conditions, and the consultation feature set includes condition features and demand features;

[0009] The consultation interpretation module is configured to: receive the consultation feature set, retrieve the consultation feature set using a medical knowledge graph, and generate corresponding interpretation data; wherein the medical knowledge graph is used to perform retrieval based on medical knowledge, disease and phenotype relationships and the associations therebetween in combination with the consultation feature set;

[0010] The consultation output module is configured to: receive the interpretation data, process the interpretation data using an interpretation engine, generate and output corresponding visual answers; wherein, the interpretation engine is determined based on the medical knowledge graph and a preset atlas, the atlas is used to interpret medical knowledge, and the visual answers are used to meet the consultation needs.

[0011] Preferably, the generation of the consultation feature set specifically includes:

[0012] Using artificial intelligence technology to process the consultation conditions and the corresponding consultation needs; extracting features from the medical-related data provided by the consultant for the consultation conditions and defining them as the condition features; extracting features of the content provided by the consultant based on the consultation conditions and defining them as the demand features for the consultation needs corresponding to the consultation conditions;

[0013] The condition feature and the demand feature are combined to generate the consultation feature set.

[0014] Preferably, the medical knowledge graph specifically includes:

[0015] The existing medical knowledge base is integrated, and the medical knowledge, diseases and phenotype relationships in the integrated medical knowledge base are used as entities, and the associations between the medical knowledge, diseases and phenotype relationships in the integrated medical knowledge base are used as edges corresponding to the entities to generate the medical knowledge graph.

[0016] Preferably, the generation of the interpretation data specifically includes:

[0017] The conditional features and the demand features in the consultation feature set are mapped to the corresponding entities, and the explanation data is generated based on the edge where the mapped entity is located and each of the entities located on the edge.

[0018] Preferably, the interpretation engine specifically includes:

[0019] Obtaining an interpretation standard corresponding to the medical knowledge graph, wherein the interpretation standard is used to perform a standardized evaluation corresponding to the consultation feature set based on the medical knowledge graph;

[0020] Matching the conditional features based on the interpretation data and the interpretation criteria, the matching being used to match data in the interpretation data that satisfies a preset matching relationship with the conditional features, with the interpretation criteria as a constraint, defining the matched data as evidence data, and matching the corresponding atlas based on the evidence data;

[0021] The interpretation criteria, the evidence data, and the atlas are combined to determine the interpretation engine.

[0022] Preferably, the generation of the visual answer specifically includes:

[0023] The interpretation engine is processed using artificial intelligence technology; with respect to the interpretation standard and the evidence data, the evidence data is interpreted using the interpretation standard; with respect to the interpretation standard, the evidence data and the atlas, while interpreting the evidence data using the interpretation standard, the atlas is used for visualization assistance to generate the visual answer.

[0024] Preferably, the generation of the consultation feature set further includes:

[0025] Constructing a demand feature prediction model using the historical consultation feature set, wherein the demand feature prediction model is used to generate predicted demand features based on the consultation feature set and the historical consultation feature set;

[0026] Run the demand feature prediction model, and generate the predicted demand feature based on the condition feature in the consultation feature set and in combination with the historical condition feature in the historical consultation feature set; wherein, the combination with the historical condition feature in the historical consultation feature set specifically comprises: analyzing the condition similarity or condition correlation between the condition feature and the historical condition feature, and when the condition similarity meets a preset condition similarity threshold or the condition correlation meets a preset condition correlation threshold, using the historical demand feature corresponding to the historical condition feature as the predicted demand feature corresponding to the condition feature; wherein, the condition similarity is used to characterize the similarity between the condition feature and the historical condition feature, and the condition correlation is used to characterize the correlation between the condition feature and the historical condition feature;

[0027] The condition features, the demand features and the predicted demand features are combined to generate a predicted consultation feature set, which is used to verify the consultation compliance of the consultant.

[0028] Preferably, the system further comprises a security protection module, and the consultation collection module, the security protection module, the consultation interpretation module and the consultation output module are communicatively connected in sequence;

[0029] The security protection module is configured to: receive the predicted consulting feature set, analyze the demand similarity or demand correlation between the demand feature and the predicted demand feature, and when the demand similarity does not meet any one or more of the preset demand similarity thresholds or the demand correlation does not meet the preset demand correlation thresholds, determine that the consultant's consulting compliance is non-compliant, stop the operation of the consulting interpretation module and the consulting output module, and output the complete operation process of the demand feature prediction model for auditing.

[0030] Preferably, the update of the medical knowledge graph includes version update and customization update;

[0031] The version update means that when the medical knowledge base is updated, the medical knowledge graph is updated accordingly;

[0032] The customized update is updated based on the extended consultation conditions provided by the consultant, and the extended consultation conditions at least include environmental conditions and family conditions.

[0033] Preferably, the interpretation engine further comprises:

[0034] When there is no one or more of the requirement similarity not meeting the preset requirement similarity threshold or the requirement relevance not meeting the preset requirement relevance threshold, the consultant's consulting compliance is determined to be compliant;

[0035] After determining that the consultant's consultation compliance is compliant, a risk assessment is performed based on the interpretation standard and the evidence data, and the results of the risk assessment are combined with the interpretation standard, the evidence data and the atlas to determine a new interpretation engine; wherein the risk assessment is performed based on the interpretation standard.

[0036] Beneficial effects: The medical knowledge intelligent consultation system based on knowledge graph of this application realizes efficient integration and intelligent consultation of multi-source medical data through the collaborative design of consultation collection, interpretation and output modules; the collection module generates a consultation feature set including condition and demand features, and combines artificial intelligence technology with the prediction model constructed by historical data to improve the accuracy of demand understanding and the efficiency of compliance review; the interpretation module realizes the association retrieval of knowledge, disease and phenotype relationship based on the medical knowledge graph, generates explanation data through entity mapping and edge relationship analysis, and solves the problem of information fragmentation; the output module uses the explanation engine to integrate evidence data and atlas, generate visual answers, and lower the usage threshold; the security protection module ensures data security; the knowledge graph supports version and customization updates to ensure the timeliness of knowledge; thereby improving the efficiency of genetic variation analysis, the accuracy of interpretation and the accuracy of prediction, realizing natural language interaction and visual analysis, and meeting the needs of medical data and its safety regulations. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1A schematic diagram of the process of intelligent medical knowledge consultation based on knowledge graph provided in an embodiment of the present application;

[0039] Figure 2 A structural block diagram of the medical knowledge intelligent consulting system based on knowledge graph provided in an embodiment of the present application. DETAILED DESCRIPTION

[0040] The following is a clear and complete description of the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0041] In this document, the term "comprising" is intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising..." does not preclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0042] With the popularization of gene sequencing technology, the demand for genetic counseling is growing rapidly. The traditional model relying on expert experience is difficult to meet clinical needs, and there is an urgent need to develop intelligent auxiliary systems to improve work efficiency and accuracy.

[0043] In response to the above requirements, this embodiment discloses Figure 1 The flowchart of medical knowledge intelligent consultation is shown in FIG. Figure 2 As shown, this embodiment discloses a medical knowledge intelligent consultation system based on a knowledge graph, which includes a consultation collection module, a consultation interpretation module, and a consultation output module that are sequentially communicatively connected;

[0044] The consultation collection module is configured to: process the collected consultation conditions and corresponding consultation needs of the consultant and generate a corresponding consultation feature set; wherein the consultation conditions are the consultant's medical-related data, the consultation needs are the content of the consultation based on the consultation conditions, and the consultation feature set includes condition features and demand features;

[0045] The consultation interpretation module is configured to: receive a consultation feature set, retrieve the consultation feature set using a medical knowledge graph, and generate corresponding interpretation data; wherein the medical knowledge graph is used to retrieve the consultation feature set based on medical knowledge, disease and phenotype relationships and the associations therebetween;

[0046] The consultation output module is configured to: receive interpretation data, process the interpretation data using an interpretation engine, generate corresponding visual answers and output them; wherein the interpretation engine is determined based on a medical knowledge graph and a preset atlas, the atlas is used to interpret medical knowledge, and the visual answers are used to meet consultation needs.

[0047] It should be noted that the consultant in this embodiment can be either a doctor or a patient.

[0048] Through the above, efficient processing of multi-source medical data and intelligent consultation are achieved. The acquisition module generates standardized feature sets to improve data processing efficiency. The interpretation module retrieves medical knowledge based on knowledge graph associations to solve the problem of information fragmentation. The output module generates visual answers through the interpretation engine, lowering the user threshold and realizing an automated intelligent consultation process from data acquisition to answer output, improving consultation efficiency and accuracy.

[0049] Utilizing the medical knowledge intelligent consultation system based on knowledge graph, the consultation feature set is generated through existing artificial intelligence technologies, for example, natural language processing technology to process consultation conditions and needs, respectively extracting conditional features in medical-related data and demand features of consultation content, and combining them to form a multi-dimensional consultation feature set.

[0050] Specifically, the generation of the consultation feature set includes:

[0051] Utilize artificial intelligence technology to process consultation conditions and their corresponding consultation needs; extract features from the medical-related data provided by the consultant for the consultation conditions and define them as conditional features; extract features of the content provided by the consultant based on the consultation conditions and define them as demand features for the consultation needs corresponding to the consultation conditions;

[0052] Combine conditional features and demand features to generate a consulting feature set.

[0053] Through the above, based on the integration of medical-related data and consultation demand characteristics, the problem of information fragmentation in traditional consultation is solved and the integrity of feature extraction is improved; the multi-dimensional feature combination realizes the accurate representation of consultation intentions, provides structured input for the semantic matching of knowledge graphs, improves the accuracy and efficiency of medical consultation, and avoids misdiagnosis or consultation deviation caused by feature omissions.

[0054] Utilizing the medical knowledge intelligent consulting system based on knowledge graph, the medical knowledge graph integrates the existing medical knowledge base, takes medical knowledge, diseases, and phenotypes as entities, and the relationships between the three as edges to construct a structured medical knowledge graph.

[0055] Specifically, the medical knowledge graph includes:

[0056] Integrate the existing medical knowledge base, and use the medical knowledge, diseases and phenotype relationships in the integrated medical knowledge base as entities, and the associations between the medical knowledge, diseases and phenotype relationships in the integrated medical knowledge base as the edges corresponding to the entities to generate a medical knowledge graph.

[0057] It should be noted that the medical knowledge base in this embodiment can be, but is not limited to, authoritative databases such as ClinVar, OMIM, HGMD, and gnomAD.

[0058] Through the above, the medical knowledge graph method integrates multi-source medical knowledge bases, structures the scattered medical knowledge, diseases and phenotype relationships, forms an entity- and edge-related knowledge network, solves the information fragmentation problem of traditional knowledge bases, improves the efficiency of knowledge association queries, provides structured support for semantic understanding and knowledge reasoning in medical consultation, and improves the accuracy and response speed of the consultation system.

[0059] Based on the medical knowledge graph, explanatory data is generated by mapping the conditional features and demand features in the consultation feature set into entities in the knowledge graph, and generating structured data for explaining the consultation results based on the edges where the entities are located and the associated entities on the edges.

[0060] Specifically, the generation of explanation data includes:

[0061] The conditional features and requirement features in the consultation feature set are mapped to corresponding entities, and explanation data is generated based on the edge where the mapped entity is located and each entity located on the edge.

[0062] In a simple example, the consultation conditions are "Condition A, Condition B". "Condition A, Condition B" are mapped to entity a and entity b in the knowledge graph, and along the edge where entity a and entity b are located, explanatory data including risk factors and inspection suggestions are generated.

[0063] Through the above, the generation of explanatory data transforms consultation features into traceable knowledge association paths through entity and edge mapping of the knowledge graph, solves the problem of ambiguous explanations in traditional consultation systems, and improves the interpretability of consultation results; structured explanatory data supports the deduction of associations from symptoms, diseases, and suggestions, helps users understand the knowledge basis of medical advice, and improves the credibility and practicality of the intelligent consultation system.

[0064] Based on the consultation feature set, the interpretation engine obtains the interpretation standards corresponding to the medical knowledge graph, matches the evidence data and the corresponding atlas in the conditional features based on the interpretation data and standards, and combines them to form an interpretation engine with standardized evaluation capabilities.

[0065] Specifically, the interpretation engine includes:

[0066] Obtaining interpretation criteria corresponding to the medical knowledge graph, where the interpretation criteria are used to perform standardized evaluation corresponding to the consultation feature set based on the medical knowledge graph;

[0067] Matching is performed in the conditional features based on the interpretation data and the interpretation criteria. The matching is used to match data in the interpretation data that meets a preset matching relationship with the conditional features, using the interpretation criteria as a constraint. The matched data is defined as evidence data, and the corresponding atlas is matched based on the evidence data.

[0068] Combine interpretation criteria, evidence data, and atlas to determine the interpretation engine.

[0069] It should be noted that the interpretation criteria in this embodiment can be, but are not limited to, the 28 evidence criteria evaluations automatically implemented based on the ACMG guidelines.

[0070] Through the above, the interpretation engine solves the problem of lack of standardization in traditional consultation interpretations through a matching mechanism between interpretation standards and evidence data, thereby improving the evidence support rate of medical recommendations; combined with the visual presentation of the atlas, it realizes the three-dimensional interpretation of standards, evidence, and illustrations, improves users' understanding and trust in genetic counseling recommendations, and provides structured interpretation support for complex medical decisions.

[0071] Based on the interpretation engine, the generation of visual answers uses existing artificial intelligence technology to parse the interpretation standards and evidence data in the interpretation engine, and combines them with the atlas for visual presentation to form a visual consultation answer that combines pictures and text.

[0072] Specifically, the generation of visual answers includes:

[0073] Use artificial intelligence technology to process the interpretation engine; for interpretation standards and evidence data, use the interpretation standards to interpret the evidence data; for interpretation standards, evidence data and atlases, while using the interpretation standards to interpret the evidence data, use the atlases for visualization assistance to generate visual answers.

[0074] In a simple example, the consultation condition is "both husband and wife are carriers of gene C". The interpretation standard of "gene C inheritance law" in the interpretation engine is called to analyze the evidence data of "husband and wife carry gene C → probability of fetus getting sick", match the "gene C inheritance pattern diagram" and "prenatal genetic testing process animation" to generate a visual answer.

[0075] Through the above, the method of generating visual answers solves the problem of obscure professional terms in traditional medical consultation by integrating text explanations and visual materials, and improves the efficiency of understanding genetic consultation content; through the auxiliary presentation of genetic laws and testing processes through graphics and animations, it lowers the understanding threshold for non-professional users, improves the readability of consultation results and the efficiency of doctor-patient communication, and provides intuitive decision-making support for complex genetic consultation.

[0076] Based on the generation of the consulting feature set, this embodiment further utilizes historical consulting data to construct a demand feature prediction model, generates predicted demand features by analyzing the similarity or correlation between conditional features and historical conditional features, and combines them to form a predicted consulting feature set for verifying consulting compliance.

[0077] Specifically, the generation of the consulting feature set also includes:

[0078] A demand feature prediction model is constructed using the historical consultation feature set. The demand feature prediction model is used to generate predicted demand features based on the consultation feature set and the historical consultation feature set.

[0079] Running a demand feature prediction model to generate predicted demand features based on the conditional features in the consulting feature set and in combination with the historical conditional features in the historical consulting feature set; wherein combining the historical conditional features in the historical consulting feature set specifically comprises: analyzing the conditional similarity or conditional correlation between the conditional features and the historical conditional features, and when the conditional similarity satisfies a preset conditional similarity threshold or the conditional correlation satisfies a preset conditional correlation threshold, using the historical demand feature corresponding to the historical conditional feature as the predicted demand feature corresponding to the conditional feature; wherein the conditional similarity is used to characterize the similarity between the conditional feature and the historical conditional feature, and the conditional correlation is used to characterize the correlation between the conditional feature and the historical conditional feature;

[0080] The condition features, demand features and predicted demand features are combined to generate a predicted consulting feature set, which is used to verify the consulting compliance of the consultant.

[0081] In a simple example, the consultation condition is "Condition E and Condition F". The similarity between the current condition features (Condition E, Condition F) and the condition features in the historical consultation data is analyzed, and the condition correlation with "Condition D" is 85% (in this embodiment, the condition similarity threshold is 80%). The "demand d" corresponding to "Condition D" in the historical demand features is extracted as the predicted demand feature, and a predicted consultation feature set containing the predicted feature is generated.

[0082] Through the above, the generation of predictive consultation feature sets solves the problem of missing demand features in traditional consultations through similarity analysis of historical data, thereby improving consultation integrity; automatically supplements potential demand features, realizes pre-verification of consultation compliance, reduces the risk of misdiagnosis due to missing information, and provides an intelligent demand supplement and risk warning mechanism for genetic counseling.

[0083] Based on the generation of the predictive consultation feature set, this embodiment further sets up a security protection module to verify the compliance of the consultation in real time by analyzing the similarity or correlation between the demand characteristics and the predicted demand characteristics. In the event of non-compliance, the subsequent process is stopped and the audit record is output.

[0084] Specifically, the system further includes a security protection module, and the consultation collection module, the security protection module, the consultation interpretation module and the consultation output module are sequentially communicatively connected;

[0085] The security protection module is configured to: receive a set of predicted consulting features, analyze the demand similarity or demand correlation between the demand features and the predicted demand features, and when the demand similarity does not meet any one or more of the preset demand similarity thresholds or the demand correlation does not meet the preset demand correlation thresholds, determine that the consultant's consulting compliance is non-compliant, stop the operation of the consulting interpretation module and the consulting output module, and output the complete operation process of the demand feature prediction model for auditing.

[0086] In a simple example, the consultation condition is "premarital genetic counseling." The condition features provided by the consultant include "family genetic history investigation for both parties." Historical data predicts the demand feature as "genetic testing recommendation," but the actual demand feature is "genetic risk assessment process consultation." The security protection module analyzes the two needs and finds a 40% correlation (in this example, the demand correlation threshold is 50%). The module determines that the consultation is non-compliant, immediately terminates the interpretation and output modules, and generates an audit record containing the derivation logic for the family medical history and genetic testing recommendation.

[0087] Through the above, the security protection module solves the problems of information omissions or improper requests that may exist in traditional consulting systems through a demand feature verification mechanism, thereby improving the efficiency of consulting compliance verification; automatically terminates the process and outputs audit records when non-compliance occurs, realizing security control of the entire genetic counseling process, protecting patient privacy and the accuracy of medical advice, and improving the reliability and traceability of the system.

[0088] Based on the construction of the medical knowledge graph, this embodiment further divides the update of the knowledge graph into version update and customized update. The version update is synchronized with the medical knowledge base, and the customized update is dynamically updated based on the expanded consulting conditions such as environmental conditions and family conditions provided by the consultant.

[0089] Specifically, updates to the medical knowledge graph include version updates and customization updates;

[0090] Version update means that when the medical knowledge base is updated, the medical knowledge graph will be updated accordingly;

[0091] Customized updates are updates based on the extended consultation conditions provided by the consultant, which at least include environmental conditions and family conditions.

[0092] In this embodiment, when the version update is based on the medical association updating the screening guidelines for a certain type of genetic disease, the associated edges (such as "applicable population range") of the corresponding entity (such as "screening process") in the knowledge graph are automatically updated.

[0093] In this embodiment, the customized update is based on increasing the weight of the association edge between environmental factors and reproductive health in the knowledge graph when the consultant provides environmental conditions; when family conditions are provided, a new association path between family history and genetic counseling recommendations is added.

[0094] Through the above, the knowledge graph update mechanism ensures the timeliness of medical knowledge through version updates and improves the efficiency of knowledge base synchronization; through customized updates, it realizes dynamic response to the consultant's personalized conditions, improves the pertinence of consultation results, provides real-time and personalized knowledge support for genetic counseling, and improves the adaptability and accuracy of the system.

[0095] Based on the compliance verification of consulting by the security protection module, this embodiment further adds a risk assessment function to compliance consulting, quantifies risks based on interpretation standards and evidence data, and integrates the assessment results into the interpretation engine to form a more comprehensive decision support system.

[0096] Specifically, the interpretation engine also includes:

[0097] When there is no requirement similarity that does not meet the preset requirement similarity threshold or requirement relevance that does not meet the preset requirement relevance threshold, the consultant's consulting compliance is determined to be compliant;

[0098] After the consultant's consulting compliance is determined to be compliant, a risk assessment is performed based on the interpretation standards and evidence data, and the results of the risk assessment are combined with the interpretation standards, evidence data and atlas to determine a new interpretation engine; wherein, the risk assessment is performed based on the interpretation standards.

[0099] It should be noted that the risk assessment in this embodiment can adopt the existing polygenic risk scoring algorithm.

[0100] Through the above, risk assessment solves the problem of vague risk description in traditional consultation by introducing quantitative risk analysis in compliance consultation, thereby improving the accuracy of risk assessment in genetic consultation; combining risk results with visual atlases provides consultants with more intuitive decision-making basis, reduces decision-making bias caused by information asymmetry, and improves the scientificity and operability of genetic consultation.

[0101] In summary, the medical knowledge intelligent consultation system based on knowledge graph in this embodiment realizes efficient integration and intelligent consultation of multi-source medical data through the collaborative design of consultation collection, interpretation and output modules; the collection module generates a consultation feature set including condition and demand features, and combines artificial intelligence technology with the prediction model constructed by historical data to improve the accuracy of demand understanding and the efficiency of compliance review; the interpretation module realizes the association retrieval of knowledge, disease and phenotype relationships based on the medical knowledge graph, generates explanation data through entity mapping and edge relationship analysis, and solves the problem of information fragmentation; the output module uses the explanation engine to integrate evidence data and atlas, generates visual answers, and lowers the usage threshold; the security protection module ensures data security; the knowledge graph supports version and customization updates to ensure the timeliness of knowledge; thereby improving the efficiency of genetic variation analysis, the accuracy of interpretation and the accuracy of prediction, realizing natural language interaction and visual analysis, and meeting the needs of medical data and its safety regulations.

[0102] In the embodiments provided herein, it should be understood that the embodiments described herein can be implemented using hardware, software, firmware, middleware, code, or any appropriate combination thereof. For hardware implementation, the processor may be implemented in one or more of the following: an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field-programmable gate array (FPGA), a processor, a controller, a microcontroller, a microprocessor, or other electronic units designed to implement the functionality described herein, or any combination thereof. For software implementation, some or all of the processes of the embodiments may be performed by a computer program instructing the relevant hardware. During implementation, the program may be stored in a computer-readable storage medium or transmitted as one or more instructions or codes on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein communication media includes any medium that facilitates the transmission of a computer program from one location to another. The storage medium may be any available medium that can be accessed by a computer. Computer-readable storage media may include, but are not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing the desired program code in the form of instructions or data structures and accessible by a computer.

[0103] Finally, it should be noted that the above is only a preferred embodiment of the present application and is not intended to limit the present application. Although the present application 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 make equivalent replacements for some of the technical features therein. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A medical knowledge intelligent consulting system based on knowledge graph, characterized by: The system comprises a consultation collection module, a consultation interpretation module, a consultation output module and a security protection module which are communicatively connected in sequence; The consultation collection module is configured to: process the collected consultation conditions of the consultant and their corresponding consultation needs, and generate a corresponding consultation feature set; wherein, the consultation conditions are the medical-related data of the consultant, and the consultation needs are the content that the consultant wants to consult based on the consultation conditions, and the consultation feature set includes conditional features and demand features; the generation of the consultation feature set includes: using the historical consultation feature set to construct a demand feature prediction model, and the demand feature prediction model is used to generate predicted demand features based on the consultation feature set and the historical consultation feature set; running the demand feature prediction model, based on the conditional features in the consultation feature set, combined with the historical conditional features in the historical consultation feature set, to generate the predicted demand features; wherein, the result The historical condition features in the historical consultation feature set are specifically: analyzing the condition similarity or condition correlation between the condition feature and the historical condition feature, and when the condition similarity meets a preset condition similarity threshold or the condition correlation meets a preset condition correlation threshold, the historical demand feature corresponding to the historical condition feature is used as the predicted demand feature corresponding to the condition feature; wherein the condition similarity is used to characterize the similarity between the condition feature and the historical condition feature, and the condition correlation is used to characterize the correlation between the condition feature and the historical condition feature; combining the condition feature, the demand feature and the predicted demand feature to generate a predicted consultation feature set, and the predicted consultation feature set is used to verify the consultant's consultation compliance; The consultation interpretation module is configured to: receive the consultation feature set, retrieve the consultation feature set using a medical knowledge graph, and generate corresponding interpretation data; wherein the medical knowledge graph is used to perform retrieval based on medical knowledge, disease and phenotype relationships and the associations therebetween in combination with the consultation feature set; The consultation output module is configured to: receive the interpretation data, process the interpretation data using an interpretation engine, generate and output a corresponding visual answer; wherein the interpretation engine is determined based on the medical knowledge graph and a preset atlas, the atlas is used to interpret medical knowledge, and the visual answer is used to meet the consultation needs; The security protection module is configured to: receive the predicted consulting feature set, analyze the demand similarity or demand correlation between the demand feature and the predicted demand feature, and when the demand similarity does not meet any one or more of the preset demand similarity thresholds or the demand correlation does not meet the preset demand correlation thresholds, determine that the consultant's consulting compliance is non-compliant, stop the operation of the consulting interpretation module and the consulting output module, and output the complete operation process of the demand feature prediction model for auditing.

2. The medical knowledge intelligent consulting system based on knowledge graph according to claim 1 is characterized in that: The generation of the consultation feature set specifically includes: Using artificial intelligence technology to process the consultation conditions and the corresponding consultation needs; extracting features from the medical-related data provided by the consultant for the consultation conditions and defining them as the condition features; extracting features of the content provided by the consultant based on the consultation conditions and defining them as the demand features for the consultation needs corresponding to the consultation conditions; The condition feature and the demand feature are combined to generate the consultation feature set.

3. The medical knowledge intelligent consulting system based on knowledge graph according to claim 1 is characterized in that: The medical knowledge graph specifically includes: The existing medical knowledge base is integrated, and the medical knowledge, diseases and phenotype relationships in the integrated medical knowledge base are used as entities, and the associations between the medical knowledge, diseases and phenotype relationships in the integrated medical knowledge base are used as edges corresponding to the entities to generate the medical knowledge graph.

4. The medical knowledge intelligent consulting system based on knowledge graph according to claim 3 is characterized in that: The generation of the interpretation data specifically includes: The conditional features and the demand features in the consultation feature set are mapped to the corresponding entities, and the explanation data is generated based on the edge where the mapped entity is located and each of the entities located on the edge.

5. The medical knowledge intelligent consulting system based on knowledge graph according to claim 2 is characterized in that: The interpretation engine specifically includes: Obtaining an interpretation standard corresponding to the medical knowledge graph, wherein the interpretation standard is used to perform a standardized evaluation corresponding to the consultation feature set based on the medical knowledge graph; Matching the conditional features based on the interpretation data and the interpretation criteria, the matching being used to match data in the interpretation data that satisfies a preset matching relationship with the conditional features, with the interpretation criteria as a constraint, defining the matched data as evidence data, and matching the corresponding atlas based on the evidence data; The interpretation criteria, the evidence data, and the atlas are combined to determine the interpretation engine.

6. The medical knowledge intelligent consulting system based on knowledge graph according to claim 5 is characterized in that: The generation of the visual answer specifically includes: The interpretation engine is processed using artificial intelligence technology; with respect to the interpretation standard and the evidence data, the evidence data is interpreted using the interpretation standard; with respect to the interpretation standard, the evidence data and the atlas, while interpreting the evidence data using the interpretation standard, the atlas is used for visualization assistance to generate the visual answer.

7. The medical knowledge intelligent consulting system based on knowledge graph according to claim 3 is characterized in that: The update of the medical knowledge graph includes version update and customization update; The version update means that when the medical knowledge base is updated, the medical knowledge graph is updated accordingly; The customized update is updated based on the extended consultation conditions provided by the consultant, and the extended consultation conditions at least include environmental conditions and family conditions.

8. The medical knowledge intelligent consulting system based on knowledge graph according to claim 5 is characterized in that: The interpretation engine further includes: When there is no one or more of the requirement similarity not meeting the preset requirement similarity threshold or the requirement relevance not meeting the preset requirement relevance threshold, the consultant's consulting compliance is determined to be compliant; After determining that the consultant's consultation compliance is compliant, a risk assessment is performed based on the interpretation standard and the evidence data, and the results of the risk assessment are combined with the interpretation standard, the evidence data and the atlas to determine a new interpretation engine; wherein the risk assessment is performed based on the interpretation standard.

Images