A method and apparatus for constructing a medical knowledge graph

Abstract

This application discloses a method and apparatus for constructing a medical knowledge graph. The method includes: obtaining a first correspondence between a first disease and a first symptom based on medical text; obtaining a second correspondence between a second disease and a second symptom based on medical record data; determining a third symptom based on a first hierarchical relationship between the first and second symptoms; the third symptom including the first symptom and part of the second symptom; obtaining a third correspondence between the first disease and the third symptom based on the second hierarchical relationship, the first correspondence, and the second correspondence; and constructing a medical knowledge graph based on the third correspondence. By fusing the first correspondence obtained from the medical text and the second correspondence obtained from the medical record data through the first hierarchical relationship between symptoms and the second hierarchical relationship between diseases, the constructed medical knowledge graph exhibits higher comprehensiveness and accuracy.

Description

Technical Field

[0001] This application relates to the field of big data technology, and in particular to a method and apparatus for constructing a medical knowledge graph. Background Technology

[0002] With the advancement of medicine, the current medical knowledge system is vast and rapidly updated, revealing intricate relationships between diseases, clinical manifestations, and treatment methods. Medical knowledge graphs can visually represent these connections and have various applications in the field of big data. However, most existing medical knowledge graphs are simply constructed based on medical knowledge, resulting in relatively low comprehensiveness and accuracy. Therefore, those skilled in the art urgently need a method to construct medical knowledge graphs with higher comprehensiveness and accuracy. Summary of the Invention

[0003] To address the aforementioned technical issues, this application provides a method and apparatus for constructing a medical knowledge graph, which is used to construct a more comprehensive and accurate medical knowledge graph.

[0004] To achieve the above objectives, the technical solutions provided in this application are as follows:

[0005] Firstly, this application provides a method for constructing a medical knowledge graph, including:

[0006] Based on medical texts, obtain the first correspondence between the first disease and the first symptom;

[0007] Based on medical record data, a second correspondence between the second disease and the second symptom is obtained;

[0008] The third symptom is determined based on the hierarchical relationship between the first and second symptoms; the third symptom includes the first symptom and part of the second symptom.

[0009] Based on the second-level relationship, the first correspondence relationship, and the second correspondence relationship between the first disease and the second disease, a third correspondence relationship between the first disease and the third symptom is obtained.

[0010] Based on the third correspondence, a medical knowledge graph is constructed.

[0011] In some possible embodiments, the method further includes:

[0012] The strength of the first association between the second disease and the second symptom was determined based on medical record data.

[0013] The second association strength between the first disease and the third symptom is obtained based on the first association strength.

[0014] Based on the third correspondence, a medical knowledge graph is constructed, including:

[0015] A medical knowledge graph is constructed based on the third correspondence and the second association strength.

[0016] In some possible embodiments, obtaining a second correspondence between a second disease and a second symptom based on medical record data includes:

[0017] Standardize the disease names and symptom names in medical record data;

[0018] Based on the processed medical record data, a second correspondence between the second disease and the second symptom is obtained.

[0019] In some possible embodiments, obtaining the first association strength between the second disease and the second symptom based on medical record data includes:

[0020] The frequency of the second symptom corresponding to the second disease in the medical record data was statistically analyzed to obtain the first association strength between the second disease and the second symptom.

[0021] In some possible embodiments, it also includes:

[0022] Obtain the target correspondence; the first correspondence does not contain the target correspondence, and the second correspondence does contain the target correspondence;

[0023] When a symptom in the target correspondence corresponds to a disease in the first correspondence, the target correspondence is deleted from the second correspondence.

[0024] In some possible embodiments, it also includes:

[0025] Obtain a fourth correspondence between the first disease or first symptom and the population category;

[0026] Based on the third correspondence, a medical knowledge graph is constructed, including:

[0027] A medical knowledge graph is constructed based on the third and fourth correspondence relationships.

[0028] In some possible embodiments, medical texts include medical textbooks, medical guidelines, and medical literature.

[0029] Secondly, this application provides an apparatus for constructing a medical knowledge graph, comprising:

[0030] The first acquisition module is used to obtain the first correspondence between the first disease and the first symptom based on medical text;

[0031] The second acquisition module is used to obtain a second correspondence between a second disease and a second symptom based on medical record data;

[0032] The determination module is used to determine a third symptom based on a first hierarchical relationship between the first symptom and the second symptom; the third symptom includes the first symptom and a portion of the second symptom.

[0033] The third acquisition module is used to obtain the third correspondence between the first disease and the third symptom based on the second hierarchical relationship, the first correspondence relationship and the second correspondence relationship between the first disease and the second disease;

[0034] The module is used to construct a medical knowledge graph based on the third correspondence.

[0035] In some possible embodiments, the apparatus for constructing a medical knowledge graph further includes:

[0036] The association strength acquisition module is used to obtain the first association strength between the second disease and the second symptom based on medical record data, and to obtain the second association strength between the first disease and the third symptom based on the first association strength.

[0037] The building module is used to construct a medical knowledge graph based on the third correspondence and the second association strength.

[0038] In some possible embodiments, the second obtaining module is used to standardize the disease names and symptom names in the medical record data; and to obtain a second correspondence between the second disease and the second symptom based on the processed medical record data.

[0039] In some possible embodiments, the association strength acquisition module is used to statistically analyze the frequency of occurrence of the second symptom among the symptoms corresponding to the second disease in the medical record data, obtain the first association strength between the second disease and the second symptom, and obtain the second association strength between the first disease and the third symptom based on the first association strength.

[0040] In some possible embodiments, it also includes:

[0041] The correspondence deletion module is used to obtain the target correspondence; the first correspondence does not contain the target correspondence, and the second correspondence contains the target correspondence; when the symptom in the target correspondence corresponds to the disease in the first correspondence, the target correspondence is deleted from the second correspondence.

[0042] In some possible embodiments, it further includes: a fourth obtaining module, used to obtain a fourth correspondence between the first disease or first symptom and the population category;

[0043] The module is used to construct a medical knowledge graph based on the third and fourth correspondence relationships.

[0044] In some possible embodiments, medical texts include medical textbooks, medical guidelines, and medical literature.

[0045] Thirdly, this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method for constructing a medical knowledge graph.

[0046] Fourthly, this application provides a computer-readable storage medium storing a computer program thereon, characterized in that the computer program, when executed by a processor, implements the steps of the above-described method for constructing a medical knowledge graph.

[0047] As can be seen from the above technical solution, this application has the following beneficial effects:

[0048] This application provides a method for constructing a medical knowledge graph, including: obtaining a first correspondence between a first disease and a first symptom based on medical text; obtaining a second correspondence between a second disease and a second symptom based on medical record data; determining a third symptom based on a first hierarchical relationship between the first symptom and the second symptom; the third symptom including the first symptom and part of the second symptom; obtaining a third correspondence between the first disease and the third symptom based on the second hierarchical relationship, the first correspondence, and the second correspondence; and constructing a medical knowledge graph based on the third correspondence.

[0049] Therefore, the medical knowledge graph construction method provided in this application integrates the first correspondence obtained from medical text and the second correspondence obtained from medical record data through the first-level relationship between symptoms and the second-level relationship between diseases, resulting in a more comprehensive and accurate medical knowledge graph. Specifically, on the one hand, since the disease and most symptom data in the medical knowledge graph constructed in this application come from medical text, the accuracy of the medical knowledge graph is high. On the other hand, since the medical knowledge graph in this application also supplements some symptom data from medical data and the correspondence between symptoms and diseases, the comprehensiveness of the medical knowledge graph in this application is high. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 A flowchart illustrating a method for constructing a medical knowledge graph, as provided in an embodiment of this application;

[0052] Figure 2A schematic diagram of a medical knowledge graph construction device provided in an embodiment of this application;

[0053] Figure 3 This is a schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0054] To help better understand the solutions provided in the embodiments of this application, before introducing the methods provided in the embodiments of this application, we will first introduce the application scenarios of the solutions in the embodiments of this application.

[0055] With the advancement of medicine, the current medical knowledge system is vast and rapidly updated, revealing intricate relationships between diseases, clinical manifestations, and treatment methods. Medical knowledge graphs can visually represent these connections and have various applications in the field of big data. However, most existing medical knowledge graphs are simply constructed based on medical knowledge, resulting in relatively low comprehensiveness and accuracy. Therefore, those skilled in the art urgently need a method to construct medical knowledge graphs with higher comprehensiveness and accuracy.

[0056] To address the aforementioned technical problems, embodiments of this application provide a method for constructing a medical knowledge graph, comprising: obtaining a first correspondence between a first disease and a first symptom based on medical text; obtaining a second correspondence between a second disease and a second symptom based on medical record data; determining a third symptom based on a first hierarchical relationship between the first and second symptoms; the third symptom including the first symptom and part of the second symptom; obtaining a third correspondence between the first disease and the third symptom based on the second hierarchical relationship, the first hierarchical relationship, the first correspondence, and the second correspondence; and constructing a medical knowledge graph based on the third correspondence. Therefore, the method for constructing a medical knowledge graph provided in this application integrates the first correspondence obtained from medical text and the second correspondence obtained from medical record data through the first hierarchical relationship between symptoms and the second hierarchical relationship between diseases, resulting in a more comprehensive and accurate medical knowledge graph.

[0057] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the embodiments of this application will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0058] See Figure 1 The figure is a flowchart of a method for constructing a medical knowledge graph provided in an embodiment of this application.

[0059] like Figure 1 As shown in the embodiments of this application, the method for constructing a medical knowledge graph includes:

[0060] S101: Obtain the first correspondence between the first disease and the first symptom based on medical texts.

[0061] It should be noted that the medical text in this application embodiment may include medical textbooks, medical guidelines, medical literature, and drug instructions, etc. In practical applications, this application can first determine the ontology scope of the knowledge graph according to requirements, and then obtain the correspondence between the medical text and the ontology scope. In this application embodiment, the ontology scope of the knowledge graph can be diseases and symptoms, etc. This application can obtain multiple disease names to form a first disease and multiple symptom names to form a first symptom from published standard datasets (such as ICD10) and clinically used terms (such as symptoms and test report names). Then, for the medical text, the first correspondence between the first disease and the first symptom is obtained through manual extraction or extraction using natural language processing technology.

[0062] S102: Obtain a second correspondence between the second disease and the second symptom based on medical record data.

[0063] It should be noted that the medical record data provided in this application embodiment originates from the structured content of medical record records. After obtaining the medical record data, this application can process and standardize the data. The purpose of data processing and standardization is to align disease names or symptom names, avoiding inaccurate statistical results caused by multiple words having the same meaning. The standard for aligning disease names or symptom names is consistent with the constructed knowledge graph based on medical knowledge. This application can use the set of multiple disease names obtained after aligning disease names as the second disease, and the set of multiple symptom names obtained after aligning symptom names as the second symptom. Then, through the medical record data, a second correspondence between the second disease and the second symptom can be obtained.

[0064] In this embodiment of the application, in order to ensure that the relation names in the statistical results belong to the specified category, it is necessary to filter the words. The filtering criteria can be disease names or symptom names in the knowledge base or a dictionary that defines the range of values ​​for the statistical results.

[0065] As one possible implementation, this application embodiment can compare the first correspondence and the second correspondence to obtain a comparison result. Then, based on the comparison result, the second correspondence can be filtered to delete some irrelevant symptom correspondences. It should be noted that when a target correspondence exists in the second correspondence but does not exist in the first correspondence, this application embodiment can identify the target correspondence through preset logical judgment rules or artificial intelligence models. When the target correspondence is identified as an irrelevant symptom correspondence, this application embodiment can delete the target correspondence from the second correspondence.

[0066] As another possible implementation, embodiments of this application can also assist in identifying target correspondences by judging the correspondence between symptoms in the target correspondence and other diseases. Specifically, embodiments of this application can obtain target correspondences; a first correspondence does not contain a target correspondence, and a second correspondence contains a target correspondence; when a symptom in the target correspondence corresponds to a disease in the first correspondence, the target correspondence is deleted from the second correspondence. In practical applications, when a symptom in the target correspondence corresponds to a disease in the first correspondence, the target correspondence can also be reconfirmed through logical judgment rules, artificial intelligence models, or manual verification. When the target correspondence is again identified as a correspondence of irrelevant symptoms, embodiments of this application can delete the target correspondence from the second correspondence.

[0067] For example, a patient suffers from multiple diseases and is diagnosed with "hypertension and gout," with symptoms including "elevated blood pressure" and "finger joint pain." In the second correspondence, the symptoms corresponding to hypertension may include "elevated blood pressure" and "finger joint pain," and the symptoms corresponding to gout may also include "elevated blood pressure" and "finger joint pain." In the first correspondence, the symptoms corresponding to hypertension only include "elevated blood pressure," so the correspondence between "hypertension" and "finger joint pain" is the target correspondence. In the first correspondence, the disease "finger joint pain" in the target correspondence appears in the correspondence for the disease "gout," therefore, this application can delete the correspondence between "hypertension" and "finger joint pain" in the second correspondence.

[0068] As another possible implementation, embodiments of this application can delete the symptom correspondences matching preset phrases in the second correspondence based on preset phrases in the corpus. For example, in diagnosis, words such as "consultation" and "follow-up" may appear. These words do not belong to the disease category to be counted and need to be filtered out. Therefore, this application can set irrelevant words such as "consultation" and "follow-up" that are common in medical data as preset phrases, and delete the symptom correspondences matching preset phrases in the second correspondence based on the preset phrases. Of course, this application can also have a technician verify the parts in the second correspondence that differ from the first correspondence, and delete some incorrect correspondences in the second correspondence.

[0069] S103: Determine the third symptom based on the first hierarchical relationship between the first symptom and the second symptom; the third symptom includes the first symptom and part of the second symptom.

[0070] It should be noted that the first hierarchical relationship between the first symptom and the second symptom is the parent-child relationship between each symptom within the first and second symptom. After determining this first hierarchical relationship, this application can add statistical data from the second symptom that belong to the first symptom to the first symptom. Sub-symptoms from the second symptom that do not belong to the first symptom, together with the first symptom, constitute the third symptom.

[0071] As one possible implementation, this application can also obtain a first association strength between a second disease and a second symptom based on medical record data. Specifically, this application can statistically analyze the frequency of occurrence of the second symptom among the symptoms corresponding to the second disease in the standardized medical record data to obtain a first association strength between the second disease and the second symptom. During the statistical analysis, the statistical results can be processed according to knowledge hierarchy. Symptoms or diseases with an inclusion relationship in the knowledge hierarchy are called parent entities and child entities. The parent entity contains the child entity. For example, "hypertension" contains "stage 1 hypertension". In this embodiment, the statistical count of the parent entity is the sum of the statistical counts of the current entity and each child entity in the original data. Then, the association strength between the parent entity and other entities is generated based on the statistical data. As an example, the statistical result for "hypertension" is the sum of the statistical counts of terms such as "hypertension", "stage 1 hypertension", "stage 2 hypertension", and "stage 3 hypertension" in the original medical record.

[0072] In this embodiment, the first association strength can indicate the probability of a second symptom occurring when a second disease occurs. To obtain the first association strength, this application can also choose to merge the statistical results by person. Specifically, this involves taking the union of the statistical results from each of the same person's previous medical visits, but the frequency of each person's statistical result is only counted once. For example, in symptom statistics, if a person's first diagnosis is "hypertension" with symptoms of "elevated blood pressure" and "dizziness," and their second diagnosis is also "hypertension" with symptoms of "elevated blood pressure" and "palpitations," then when counting the symptoms of this person's "hypertension," the symptoms would be "elevated blood pressure," "dizziness," and "palpitations," but the frequency of each symptom would only be counted once. This helps avoid the duplicate counting of certain symptoms of a disease due to multiple medical visits by the same person.

[0073] S104: Based on the second-level relationship, first correspondence relationship and second correspondence relationship between the first disease and the second disease, obtain the third correspondence relationship between the first disease and the third symptom.

[0074] It should be noted that the inclusion-being-included relationship between the first disease and the second disease in this application embodiment is a second-level relationship. Based on this second-level relationship, this application embodiment can add the data corresponding to the second disease to the data corresponding to the first disease, thus fusing the first and second correspondences to obtain a third correspondence. For example, if the first disease is "hypertension" and the second disease is "hypertensive disease," which are essentially the same disease, this application can add the data corresponding to "hypertensive disease" to the data corresponding to "hypertension." The disease "hypertension" can be further subdivided into "Stage 1 hypertension," "Stage 2 hypertension," and "Stage 3 hypertension." Therefore, "hypertension" is recorded as the parent of these hypertension subtypes, and the data corresponding to "Stage 1 hypertension," "Stage 2 hypertension," and "Stage 3 hypertension" can all be included in the data corresponding to "hypertension." For example, the association strength corresponding to "Stage 1 hypertension" can be included in the association strength corresponding to "hypertension."

[0075] This application embodiment can also obtain a second association strength between the first disease and the third symptom based on the first association strength. Specifically, the first association strength is the association strength between the second disease and the second symptom. This application embodiment can determine the second association strength based on the first hierarchical relationship between the second disease and the first disease, and the correspondence between the third symptom and the second symptom. The second association strength can indicate the probability of the third symptom occurring when the first disease occurs. This application embodiment can increase the association strength corresponding to the second symptom to the association strength corresponding to the third symptom based on the correspondence between the third symptom and the second symptom, thereby generating a second association strength between the first disease and the third symptom.

[0076] S105: Construct a medical knowledge graph based on the third correspondence.

[0077] It should be noted that this application can construct a medical knowledge graph based solely on the third correspondence, or it can construct a medical knowledge graph based on the third correspondence and the second association strength. The embodiments of this application are not limited here. Based on the medical knowledge graph constructed using the third association strength, the association strength between diseases and symptoms in the medical knowledge graph can be determined.

[0078] As one possible implementation, the entities provided in this application embodiment can be diseases or symptoms, or other medical entities; this application embodiment does not limit the scope of implementation. As an example, the types of entities provided in this application embodiment may also include surgeries, drugs, symptoms, tests, examinations, population categories, etc., and the types of entities can be added or removed as needed. When the entities in this application embodiment also include population categories, the method provided in this application embodiment further includes: obtaining a fourth correspondence between the first disease or the first symptom and the population category. Accordingly, this application embodiment can construct a medical knowledge graph based on the third correspondence and the fourth correspondence.

[0079] It should be noted that when the entities in the embodiments of this application include three or more entity types, the medical knowledge graph provided in the embodiments of this application will become a multidimensional complex graph network. In a multidimensional medical knowledge graph network, there may be a correspondence between any two entity types. For example, when the entity types in the embodiments of this application include diseases, symptoms, and population categories, the medical knowledge graph may include correspondences between diseases and symptoms, between diseases and population categories, and between symptoms and population categories.

[0080] It should be noted that the information recorded in textbooks, guidelines, and literature often cannot exhaustively cover all situations. For example, disease symptoms usually list common symptoms and some rare examples, but not all symptoms that have occurred clinically. Therefore, knowledge graph relationships based on medical knowledge are often not comprehensive. However, using only the second correspondence from medical record data to construct a medical knowledge graph can lead to inaccuracies in some correspondences. The inaccuracy may be due to the format of writing medical records, which makes it impossible to achieve a one-to-one correspondence between entities. For example, a patient may have multiple diseases, so the doctor writes multiple diagnoses, but the description of symptoms in the medical record does not distinguish between diagnoses, resulting in a chaotic correspondence between diagnoses and symptoms in the second correspondence extracted from the medical record information.

[0081] The method for constructing a medical knowledge graph provided in this application integrates a first correspondence obtained from medical text and a second correspondence obtained from medical record data through a first-level relationship between symptoms and a second-level relationship between diseases, resulting in a more comprehensive and accurate medical knowledge graph. Specifically, on the one hand, because the disease and most symptom data in the medical knowledge graph constructed in this application reference medical text, the accuracy of the medical knowledge graph is high. On the other hand, because the medical knowledge graph in this application also supplements some symptom data from medical data and the correspondence between symptoms and diseases, the comprehensiveness of the medical knowledge graph is high, and the medical knowledge graph provided in this application can assist in symptom diagnosis. Due to the high comprehensiveness of the medical knowledge graph in this application embodiment, using the medical knowledge graph provided in this application embodiment can improve the accuracy of determining the type of disease a patient suffers from based on their symptoms. Moreover, this application verifies the second correspondence extracted from medical record data through the first correspondence, thereby eliminating erroneous correspondences in the second correspondence, resulting in a higher accuracy and comprehensiveness of the medical knowledge graph obtained in this application embodiment.

[0082] Based on the medical knowledge graph construction method provided in the above embodiments, this application also provides a medical knowledge graph construction apparatus.

[0083] like Figure 2 As shown, the medical knowledge graph construction apparatus provided in this application embodiment includes:

[0084] The first acquisition module 100 is used to obtain a first correspondence between a first disease and a first symptom based on medical text;

[0085] The second acquisition module 200 is used to obtain a second correspondence between a second disease and a second symptom based on medical record data;

[0086] The determination module 300 is used to determine a third symptom based on a first hierarchical relationship between the first symptom and the second symptom; the third symptom includes the first symptom and part of the second symptom.

[0087] The third obtaining module 400 is used to obtain a third correspondence between the first disease and the third symptom based on the second hierarchical relationship, the first correspondence relationship and the second correspondence relationship between the first disease and the second disease;

[0088] Module 500 is used to construct a medical knowledge graph based on the third correspondence.

[0089] In some possible embodiments, the medical knowledge graph construction apparatus further includes: an association strength acquisition module, configured to obtain a first association strength between a second disease and a second symptom based on medical record data, and to obtain a second association strength between a first disease and a third symptom based on the first association strength; and a construction module, configured to construct a medical knowledge graph based on a third correspondence and the second association strength.

[0090] In some possible embodiments, the second obtaining module is used to standardize the disease names and symptom names in the medical record data; and to obtain a second correspondence between the second disease and the second symptom based on the processed medical record data.

[0091] In some possible embodiments, the association strength acquisition module is used to statistically analyze the frequency of occurrence of the second symptom among the symptoms corresponding to the second disease in the medical record data, obtain the first association strength between the second disease and the second symptom, and obtain the second association strength between the first disease and the third symptom based on the first association strength.

[0092] In some possible embodiments, the system further includes: a correspondence deletion module for obtaining a target correspondence; a first correspondence does not contain a target correspondence, and a second correspondence contains a target correspondence; when a symptom in the target correspondence corresponds to a disease in the first correspondence, the target correspondence is deleted from the second correspondence.

[0093] In some possible embodiments, it further includes: a fourth obtaining module for obtaining a fourth correspondence between the first disease or first symptom and the population category; and a construction module for constructing a medical knowledge graph based on the third and fourth correspondences.

[0094] In some possible embodiments, medical texts include medical textbooks, medical guidelines, and medical literature.

[0095] The medical knowledge graph construction apparatus provided in this application integrates a first correspondence obtained from medical text and a second correspondence obtained from medical record data through a first-level relationship between symptoms and a second-level relationship between diseases, resulting in a more comprehensive and accurate medical knowledge graph. Specifically, on the one hand, because the disease and most symptom data in the medical knowledge graph constructed in this application are referenced from medical text, the accuracy of the medical knowledge graph constructed in this application is high. On the other hand, because the medical knowledge graph in this application also supplements some symptom data from medical data and the correspondence between symptoms and diseases, the comprehensiveness of the medical knowledge graph in this application is high. Moreover, this application verifies the second correspondence extracted from medical record data through the first correspondence, thereby eliminating erroneous correspondences in the second correspondence, resulting in a higher accuracy and comprehensiveness of the medical knowledge graph obtained in this application embodiment.

[0096] Based on the medical knowledge graph construction method and apparatus provided in the above embodiments, this application also provides an electronic device, such as... Figure 3 As shown, the electronic device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method for constructing a medical knowledge graph.

[0097] According to the medical knowledge graph construction method and apparatus provided in the above embodiments, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described medical knowledge graph construction method.

[0098] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that all or part of the steps in the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0099] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Regarding the methods disclosed in the embodiments, since they correspond to the systems disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the system section description.

[0100] It should also be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0101] The above description of the disclosed embodiments will enable those skilled in the art to make or use various modifications to these embodiments. It will be readily apparent to those skilled in the art that the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a medical knowledge graph, characterized in that, The method comprises the following steps: obtaining a first corresponding relationship between a first disease and a first symptom according to medical text; obtaining a second corresponding relationship between a second disease and a second symptom according to medical record data; determining a third symptom according to a first hierarchical relationship between the first symptom and the second symptom, wherein the third symptom comprises the first symptom and part of the second symptom, and the first hierarchical relationship is a parent-child relationship between each of the first symptom and the second symptom; obtaining a third corresponding relationship between the first disease and the third symptom according to a second hierarchical relationship between the first disease and the second disease, the first corresponding relationship and the second corresponding relationship, wherein the second hierarchical relationship is a containing and contained relationship between the first disease and the second disease; obtaining a first association strength between the second disease and the second symptom according to the medical record data, wherein the first association strength indicates a probability of the second symptom occurring when the second disease occurs; obtaining a second association strength between the first disease and the third symptom according to the first association strength, wherein the second association strength indicates a probability of the third symptom occurring when the first disease occurs; constructing a medical knowledge graph according to the third corresponding relationship and the second association strength; the method further comprises the following steps: counting a frequency of occurrence of the second symptom in symptoms corresponding to the second disease in the medical record data to obtain the first association strength between the second disease and the second symptom. 2.The method of claim 1, wherein, the method further comprises the following steps: standardizing disease names and symptom names in the medical record data; obtaining the second corresponding relationship between the second disease and the second symptom according to the processed medical record data. 3.The method of claim 1, wherein, The method further comprises the following steps: obtaining a target corresponding relationship; the first corresponding relationship does not contain the target corresponding relationship, and the second corresponding relationship contains the target corresponding relationship; when a symptom in the target corresponding relationship corresponds to a disease in the first corresponding relationship, deleting the target corresponding relationship in the second corresponding relationship. 4.The method of claim 1, wherein, The method further comprises the following steps: obtaining a fourth corresponding relationship between the first disease or the first symptom and a population category; the method further comprises the following steps: constructing the medical knowledge graph according to the third corresponding relationship and the fourth corresponding relationship. 5.The method of claim 1-4, wherein, The medical text comprises medical textbooks, medical guidelines and medical literature. 6.A device for constructing a medical knowledge graph, characterized in that, The method comprises the following steps: a first obtaining module, configured to obtain a first corresponding relationship between a first disease and a first symptom according to medical text; a second obtaining module, configured to obtain a second corresponding relationship between a second disease and a second symptom according to medical record data; a determining module, configured to determine a third symptom according to a first hierarchical relationship between the first symptom and the second symptom, wherein the third symptom comprises the first symptom and part of the second symptom, and the first hierarchical relationship is a parent-child relationship between each of the first symptom and the second symptom; a third obtaining module, configured to obtain a third correspondence relationship between the first disease and the third symptom according to a second hierarchical relationship between the first disease and the second disease, the first correspondence relationship and the second correspondence relationship, the second hierarchical relationship being a containing and contained relationship between the first disease and the second disease; an association strength obtaining module, configured to obtain a first association strength of the second disease and the second symptom according to the medical record data, the first association strength indicating a probability of the second symptom occurring when the second disease occurs; and obtain a second association strength of the first disease and the third symptom according to the first association strength, the second association strength indicating a probability of the third symptom occurring when the first disease occurs; and the obtaining of the first association strength of the second disease and the second symptom according to the medical record data comprises: counting a frequency of occurrence of the second symptom in symptoms corresponding to the second disease in the medical record data to obtain the first association strength of the second disease and the second symptom; a constructing module, configured to construct a medical knowledge graph according to the third correspondence relationship and the second association strength. 7.An electronic device comprising a memory and a processor, the memory storing a computer program, wherein, The processor executes the computer program to implement the steps of the method in any one of claims 1 to 4.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 4.

Images