A disease insurance relationship graph construction method, device and equipment

Abstract

The application discloses a disease insurance relationship graph construction method, device and equipment. Specifically, in the execution of the disease insurance relationship graph construction method provided by the application, first, the disease hierarchy based on medical coding division can be obtained. Then, the medical graph is constructed according to the disease name and the drug name. Then, the disease hierarchy and the medical graph are integrated to obtain the first integrated graph according to the corresponding relationship between the disease name contained in the first entity layer of the medical graph and the disease hierarchy. Then, the underwriting and claim graph is constructed according to the insurance name and the disease name. Finally, the first integrated graph and the underwriting and claim graph are integrated to obtain the second integrated graph according to the corresponding relationship between the disease name contained in the second entity layer of the underwriting and claim graph and the first integrated graph. The application constructs the medical graph and the relationship graph of insurance underwriting and claim based on medical coding, and realizes the combination of the medical graph and the related insurance scene.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of disease insurance relationship graph construction, in particular to a disease insurance relationship graph construction method, device and equipment. BACKGROUND

[0002] With the continuous development of knowledge graph technology, more and more people apply knowledge graph technology to related fields. Including retrieval, recommendation, question and answer, reasoning and other NLP related fields. The advantage of knowledge graph is to solve the problem of explainability of deep learning, and to infer the answers required by users by constructing the underlying graph relationship library. At the same time, more and more people in the medical field begin to build and use knowledge graphs because the explainability of medical knowledge is stronger than that in other fields. However, the existing medical graph rarely combines with the related insurance scene to show the relationship between the two, so that a lot of manpower and time is needed to determine whether the customer meets the conditions of underwriting and claims when underwriting and claims.

[0003] Therefore, how to combine medical graph and related insurance scene to display is a technical problem that needs to be solved by those skilled in the art. SUMMARY

[0004] Based on the above problems, the present application provides a disease insurance relationship graph construction method, device and equipment to improve the ease of use of icons in the interface of the terminal device for different users. The present application embodiment discloses the following technical scheme:

[0005] A disease insurance relationship graph construction method, the method comprises:

[0006] Obtaining a disease hierarchy based on medical coding division;

[0007] Constructing a medical graph according to the disease name and the drug name, the medical graph comprising a first entity layer, the first entity layer comprising the disease name and the drug name;

[0008] Integrating the disease hierarchy and the medical graph according to the corresponding relationship between the disease name in the first entity layer and the disease hierarchy to obtain a first integrated graph;

[0009] Constructing an underwriting and claims graph according to the insurance name and the disease name, the underwriting and claims graph comprising a second entity layer, the second entity layer comprising the disease name and the insurance name;

[0010] Integrating the first integrated graph and the underwriting and claims graph according to the corresponding relationship between the disease name in the second entity layer and the first integrated graph to obtain a second integrated graph.

[0011] In a possible implementation, the constructing the underwriting and claim graph according to the insurance name and the disease name comprises:

[0012] obtaining an underwriting and claim rule between the insurance name and the disease name;

[0013] constructing a first triple combination of the insurance name, the disease name and the underwriting and claim rule as the underwriting and claim graph.

[0014] In a possible implementation, the disease hierarchy comprises five disease hierarchies.

[0015] In a possible implementation, the constructing the medical graph according to the disease name and the drug name comprises:

[0016] inputting each disease name and each drug name into a small sample model for training to obtain a corresponding relationship between each disease name and each drug name with a matching rate less than a first threshold;

[0017] generating a second triple combination comprising each disease name, each drug name and the corresponding relationship between each disease name and each drug name with the matching rate less than the first threshold;

[0018] obtaining the second triple combination with the corresponding relationship between each disease name and each drug name being partially labeled as a third triple combination;

[0019] inputting the third triple combination into a relationship extraction model performing semi-supervised learning for training to obtain a corresponding relationship between each disease name and each drug name with a matching rate greater than a second threshold;

[0020] generating a fourth triple combination comprising each disease name, each drug name and the corresponding relationship between each disease name and each drug name with the matching rate greater than the second threshold as the medical graph.

[0021] In a possible implementation, the medical graph further comprises a first relationship layer:

[0022] the first relationship layer comprises the corresponding relationship between the disease name and the drug name in the first entity layer.

[0023] In a possible implementation, the underwriting and claim graph comprises a second relationship layer:

[0024] the second relationship layer comprises the corresponding relationship between the disease name and the drug name in the second entity layer.

[0025] A device for constructing a disease insurance relationship graph, the device comprising:

[0026] The first obtaining module is configured to obtain a disease hierarchy based on medical coding division;

[0027] The first constructing module is configured to construct a medical graph based on the disease name and the drug name; the medical graph comprises a first entity layer; the first entity layer comprises the disease name and the drug name;

[0028] The first integrating module is configured to integrate the disease hierarchy and the medical graph according to the correspondence between the disease name in the first entity layer and the disease hierarchy, to obtain a first integrated graph.

[0029] The second constructing module is configured to construct an insurance underwriting and claim graph based on the insurance name and the disease name; the insurance underwriting and claim graph comprises a second entity layer; the second entity layer comprises the disease name and the insurance name.

[0030] The second integrating module is configured to integrate the first integrated graph and the insurance underwriting and claim graph according to the correspondence between the disease name in the second entity layer and the first integrated graph, to obtain a second integrated graph.

[0031] In a possible implementation, the apparatus further comprises:

[0032] The second obtaining module is configured to obtain insurance underwriting and claim rules between the insurance name and the disease name.

[0033] The second constructing module is configured to construct a first triadic relation group combination of the insurance name, the disease name and the insurance underwriting and claim rules as the insurance underwriting and claim graph.

[0034] An electronic device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the disease insurance relationship graph construction method as described above when executing the computer program.

[0035] In a possible implementation, the computer readable storage medium stores instructions, and when the instructions run on the terminal device, the terminal device executes the disease insurance relationship graph construction method as described above.

[0036] Compared with the prior art, the application has the following beneficial effects: the application discloses a disease insurance relationship graph construction method, device and equipment. Specifically, when the disease insurance relationship graph construction method provided by the application is executed, a disease hierarchy based on medical coding division can be obtained first. Then, a medical graph is constructed according to disease names and drug names, the medical graph includes a first entity layer, and the first entity layer includes the disease names and the drug names. Then, the disease hierarchy is integrated with the medical graph to obtain a first integrated graph according to the correspondence between the disease names in the first entity layer and the disease hierarchy. Then, an underwriting and claim graph is constructed according to insurance names and disease names, the underwriting and claim graph includes a second entity layer, and the second entity layer includes the disease names and the insurance names. Finally, the first integrated graph is integrated with the underwriting and claim graph to obtain a second integrated graph according to the correspondence between the disease names in the second entity layer and the first integrated graph. The application constructs the medical graph and the relationship graph of insurance underwriting and claim based on medical coding, and realizes the combination of the medical graph and the related insurance scene, so that a large amount of manpower and time can be saved to determine whether the customer meets the underwriting and claim conditions when underwriting and claim is performed. BRIEF DESCRIPTION OF DRAWINGS

[0037] To make the technical solution in the embodiments or the prior art clearer, the drawings needed in the embodiments or the prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0038] Figure 1 A flowchart of a disease insurance relationship graph construction method provided by an embodiment of the application;

[0039] Figure 2 A structural schematic diagram of a disease insurance relationship graph construction device provided by an embodiment of the application. DETAILED DESCRIPTION

[0040] To make the technical solution in the embodiments or the prior art clearer, the drawings needed in the embodiments or the prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0041] To make the technical solution in the embodiments or the prior art clearer, the drawings needed in the embodiments or the prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0042] Knowledge Graph, also known as knowledge domain visualization or knowledge field mapping map in the library and information field, is a series of various different graphs that show the development process and structural relationship of knowledge. It uses visualization technology to describe knowledge resources and their carriers, and excavates, analyzes, constructs, draws and displays knowledge and their mutual relationships. Knowledge graph is a present theory that combines the theories and methods of applied mathematics, graphics, information visualization technology, information science and other disciplines with citation analysis, co-occurrence analysis and other methods of bibliometrics, and uses visualized graph to show the core structure, development history, frontier field and overall knowledge architecture of a discipline to achieve the purpose of multidisciplinary integration.

[0043] With the continuous development of knowledge graph technology, more and more people apply knowledge graph technology to related fields. Including retrieval, recommendation, question and answer, reasoning and other NLP related fields. The advantage of knowledge graph is to solve the problem of explainability of deep learning, and to infer the answers required by users by constructing the underlying graph relationship database. At the same time, more and more people start to build and use knowledge graph in the medical field because the explainability of medical knowledge is stronger than that in other fields. However, the existing medical graph rarely combines with the related insurance scene to show the relationship between the two, so that a lot of manpower and time is needed to determine whether the customer meets the conditions of insurance and claim when performing insurance and claim.

[0044] In order to solve this problem, the application provides a disease insurance relationship graph construction method, device and equipment. First, the disease hierarchy based on medical coding division is obtained in the application. Then, according to the disease name and the drug name, a medical graph is constructed, and the medical graph includes a first entity layer, and the first entity layer includes the disease name and the drug name. Next, according to the corresponding relationship between the disease name in the first entity layer and the disease hierarchy, the disease hierarchy and the medical graph are integrated to obtain a first integrated graph. After obtaining the first integrated graph, an insurance and claim graph is constructed according to the insurance name and the disease name, and the insurance and claim graph includes a second entity layer, and the second entity layer includes the disease name and the insurance name. Finally, according to the corresponding relationship between the disease name in the second entity layer and the first integrated graph, the first integrated graph and the insurance and claim graph are integrated to obtain a second integrated graph. The application constructs the medical graph and the relationship graph of insurance and claim based on medical coding, realizes the combination of the medical graph and the related insurance scene, so as to save a lot of manpower and time to determine whether the customer meets the conditions of insurance and claim when performing insurance and claim.

[0045] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of the present application.

[0046] Referring to Figure 1 The figure is a flowchart of a disease insurance relationship graph construction method provided by an embodiment of the present application, as shown in Figure 1 The disease insurance relationship graph construction method can include steps S101-S105.

[0047] S101: Obtain a disease hierarchy based on medical coding division.

[0048] In order to construct a disease insurance relationship graph, the disease insurance relationship graph construction system can first obtain a disease hierarchy based on medical coding division of diseases.

[0049] In one possible implementation, the medical coding can be international ICD-10 coding. International ICD-10 coding is international Classification of Diseases (ICD), which is a system that classifies diseases according to certain characteristics of diseases, classifies diseases according to rules, and represents them by coding.

[0050] In one possible implementation, the disease hierarchy includes five disease hierarchies. For example, Table 1 is a disease hierarchy of a disease insurance relationship graph:

[0051] Table 1 Disease hierarchy

[0052]

[0053] Wherein, A00-B99 is a disease module, and some infectious diseases and parasitic diseases are specific names of the disease module. The disease module is the first level in the disease hierarchy. A00-A09 is a disease title, and intestinal infectious diseases are specific names of the disease title. The disease title is the second level in the disease hierarchy. A01 is the first kind of disease, and typhoid and paratyphoid are specific names of the first kind of disease. The first kind of disease is the third level in the disease hierarchy. A01.000 is the second kind of disease, and typhoid is a specific name of the second kind of disease. The second kind of disease is the fourth level in the disease hierarchy. A01.000X006 is the third kind of disease, and typhoid is a specific name of the third kind of disease. The third kind of disease is the fifth level in the disease hierarchy.

[0054] S102: Construct a medical graph according to the disease name and the drug name, the medical graph comprising a first entity layer, the first entity layer comprising the disease name and the drug name.

[0055] After obtaining the disease hierarchy of the disease divided based on the medical coding, the disease insurance relationship graph construction system can construct a medical graph according to the disease name and the drug name, and the medical graph comprises a first entity layer, and the first entity layer comprises the name of the disease and the name of the drug. For example, the medical graph

[0056] In some possible implementation manners, the constructing the medical graph according to the disease name and the drug name can comprise A1-A5.

[0057] A1: input each disease name and each drug name into a small sample model for training to obtain a corresponding relationship between each disease name and each drug name with a matching rate less than a first threshold.

[0058] When constructing the medical graph according to the disease name and the drug name, first, each disease name and each drug name can be input into a small sample model for training to obtain a training result, and the training result is that the matching rate of the corresponding relationship between the disease name and the drug name is less than the first threshold.

[0059] In some possible implementation manners, the first threshold can be, but is not limited to, 50%. The matching rate refers to the probability of matching between the corresponding relationship between the disease name and the drug name and the disease name and the drug name. For example, the disease name of a disease is cold, and the name of the drug is indomethacin. At this time, the corresponding relationship between cold and indomethacin is that they correspond to each other. At this time, the matching rate of the corresponding relationship is less than 50%, because cold can be caused by inflammation, and indomethacin can eliminate inflammation but cannot completely treat cold.

[0060] In some possible implementation manners, the small sample model is a model constructed based on small sample learning. Since small sample learning requires less data to train the model, it eliminates the high cost related to data collection and labeling. The small amount of training data means that the dimensionality of the training data set is low, which can significantly reduce the computing cost and improve the computing time.

[0061] A2: generate a second triple combination comprising each disease name, each drug name, and a corresponding relationship between each disease name and each drug name with a matching rate less than the first threshold.

[0062] After obtaining the corresponding relationship between each disease name and each drug name with a matching rate less than the first threshold, a second triple combination including each disease name, each drug name, and the corresponding relationship between each disease name and each drug name with the matching rate less than the first threshold needs to be generated. For example, the second triple can be: cold-matches-indomethacin. The second triple combination is a set of all second triples.

[0063] In some possible implementation manners, a triple refers to a set in the form of ((x, y), z), where x and y are two elements, and z is a mapping relationship between x and y. A triple combination is a set of triples.

[0064] A3: Obtain a second triple combination in which the corresponding relationship between each disease name and each drug name is partially labeled as a third triple combination.

[0065] After generating the second triple combination including each disease name, each drug name, and the corresponding relationship between each disease name and each drug name with the matching rate less than the first threshold, in order to improve the matching rate of the corresponding relationship between each disease name and each drug name, a part of the corresponding relationship between the disease name and the drug name needs to be labeled, and the third triple combination is obtained and input into the relationship extraction model for semi-supervised learning for training.

[0066] A4: Input the third triple combination into the relationship extraction model for semi-supervised learning for training, to obtain the corresponding relationship between each disease name and each drug name with a matching rate greater than a second threshold.

[0067] After obtaining the second triple combination in which the corresponding relationship between each disease name and each drug name is partially labeled as the third triple combination, the third triple combination needs to be input into the relationship extraction model for semi-supervised learning for training, to improve the matching rate of the corresponding relationship between each disease name and each drug name, so that the matching rate is greater than the second threshold.

[0068] In a possible implementation manner, the second threshold can be, but is not limited to, 98%.

[0069] In a possible implementation manner, the relationship extraction model for semi-supervised learning can be a relationship extraction model with a semi-supervised learning function. The relationship extraction model can be, but is not limited to, a relationship extraction pipeline model PURE.

[0070] A5: Generate a fourth triple combination including each disease name, each drug name, and the corresponding relationship between each disease name and each drug name with the matching rate greater than the second threshold as a medical atlas.

[0071] After the third triple combination is input into the relation extraction model for semi-supervised learning to train, so as to obtain the corresponding relationship between each disease name and each drug name with a matching rate greater than the second threshold, a fourth triple combination including the disease name, the drug name and the corresponding relationship between the disease name and the drug name with the matching rate greater than the second threshold is generated as a medical atlas.

[0072] In a possible implementation, the medical atlas further includes a first relationship layer.

[0073] The first relationship layer includes the corresponding relationship between the disease name and the drug name in the first entity layer.

[0074] S103: integrating the disease hierarchy with the medical atlas according to the corresponding relationship between the disease name in the first entity layer and the disease hierarchy to obtain a first integrated atlas.

[0075] After the medical atlas is constructed according to the disease name and the drug name, the disease insurance relationship atlas construction system further needs to integrate the disease hierarchy with the medical atlas according to the corresponding relationship between the disease name in the first entity layer and the disease hierarchy to obtain a first integrated atlas.

[0076] In a possible implementation, the first integrated atlas includes a disease module, a disease title, a disease first category, a disease second category, a disease third category, a drug name, and a corresponding relationship between the disease name and the disease hierarchy.

[0077] S104: constructing an underwriting and claims atlas according to the insurance name and the disease name, the underwriting and claims atlas including a second entity layer, the second entity layer including the disease name and the insurance name.

[0078] After the disease hierarchy is integrated with the medical atlas according to the corresponding relationship between the disease name in the first entity layer and the disease hierarchy to obtain a first integrated atlas, an underwriting and claims atlas needs to be constructed according to the insurance name and the disease name to make the first integrated atlas combined with the scenario corresponding to the insurance, so as to form a complete disease insurance relationship atlas. The underwriting and claims atlas includes a second entity layer, and the second entity layer includes the disease name and the insurance name.

[0079] In a possible implementation, the underwriting and claims atlas constructed according to the insurance name and the disease name includes B1-B2.

[0080] B1: obtaining an underwriting and claims rule between the insurance name and the disease name.

[0081] To construct the underwriting and claim graph based on the insurance name and the disease name, it is necessary to first obtain the underwriting and claim rules between the insurance name and the disease name.

[0082] In some possible implementation manners, the underwriting and claim rules can include, but are not limited to, the claim settlement time of the insurance, the taboo of purchasing the insurance, the claim settlement taboo of the insurance, the claim settlement amount of the insurance, and the like. For example, the taboo of purchasing the million medical insurance is that the disease of lung nodule cannot be obtained.

[0083] B2: Construct a first three-tuple combination of the insurance name, the disease name, and the underwriting and claim rule as the underwriting and claim graph.

[0084] After obtaining the underwriting and claim rules between the insurance name and the disease name, it is necessary to construct a combination of the first three-tuple of the insurance name, the disease name, and the underwriting and claim rule as the underwriting and claim graph.

[0085] In some possible implementation manners, the underwriting and claim graph includes a second relationship layer. The second relationship layer includes the corresponding relationship between the disease name and the drug name in the second entity layer.

[0086] S105: Integrate the first integrated graph and the underwriting and claim graph according to the corresponding relationship between the disease name in the second entity layer and the first integrated graph to obtain a second integrated graph.

[0087] After constructing the underwriting and claim graph based on the insurance name and the disease name, the disease insurance relationship graph construction system also needs to integrate the first integrated graph and the underwriting and claim graph according to the corresponding relationship between the disease name in the second entity layer and the first integrated graph to obtain a second integrated graph, so that a large amount of manpower and time is saved to determine whether the customer meets the underwriting and claim condition when performing underwriting and claim.

[0088] Based on the content of S101-S105, the disease hierarchy based on medical coding division is obtained, the medical graph is constructed according to the disease name and the drug name, and then the disease hierarchy and the medical graph are integrated according to the corresponding relationship between the disease name in the first entity layer in the medical graph and the disease hierarchy to obtain a first integrated graph. Then, the underwriting and claim graph is constructed based on the insurance name and the disease name, and finally, the first integrated graph and the underwriting and claim graph are integrated according to the corresponding relationship between the disease name in the second entity layer in the underwriting and claim graph and the first integrated graph to obtain a second integrated graph. The medical graph and the underwriting and claim relationship graph are constructed based on the medical coding in the present application, and the medical graph and the related insurance scene are combined, so that a large amount of manpower and time is saved to determine whether the customer meets the underwriting and claim condition when performing underwriting and claim.

[0089] Referring to Figure 2FIG. 1 is a structural schematic diagram of a disease insurance relationship graph construction device according to an embodiment of the present application. As shown in the figure, the disease insurance relationship graph construction device includes: Figure 2

[0090] A first obtaining module 201 is configured to obtain a disease hierarchy based on medical coding.

[0091] In a possible implementation, the medical coding can be international ICD-10 coding. The international ICD-10 coding is an international Classification of Diseases (ICD), which is a system that classifies diseases according to certain characteristics of diseases and classifies diseases according to rules and uses coding methods to represent the system.

[0092] In a possible implementation, the disease hierarchy includes five disease hierarchies. For example, Table 1 is a disease hierarchy of a disease insurance relationship graph:

[0093] Table 1 Disease hierarchy

[0094]

[0095] Wherein, A00-B99 is a disease module, and some infectious diseases and parasitic diseases are specific names of the disease module. The disease module is the first level in the disease hierarchy. A00-A09 is a disease title, and intestinal infectious diseases are specific names of the disease title. The disease title is the second level in the disease hierarchy. A01 is a first disease category, and typhoid and paratyphoid are specific names of the first disease category. The first disease category is the third level in the disease hierarchy. A01.000 is a second disease category, and typhoid is a specific name of the second disease category. The second disease category is the fourth level in the disease hierarchy. A01.000X006 is a third disease category, and typhoid is a specific name of the third disease category. The third disease category is the fifth level in the disease hierarchy.

[0096] A first construction module 202 is configured to construct a medical graph according to a disease name and a drug name, wherein the medical graph includes a first entity layer, and the first entity layer includes the disease name and the drug name.

[0097] A first integration module 203 is configured to integrate the disease hierarchy and the medical graph according to a corresponding relationship between the disease name in the first entity layer and the disease hierarchy to obtain a first integrated graph.

[0098] A second construction module 204 is configured to construct an insurance underwriting and claims graph according to an insurance name and the disease name, wherein the insurance underwriting and claims graph includes a second entity layer, and the second entity layer includes the disease name and the insurance name. ​

[0099] The second integration module 205 is configured to integrate the first integration graph and the underwriting and claim settlement graph according to the correspondence between the disease name in the second entity layer and the first integration graph to obtain a second integration graph.

[0100] In a possible implementation, the apparatus further includes:

[0101] The second acquisition module is configured to acquire underwriting and claim settlement rules between the insurance name and the disease name.

[0102] In some possible implementations, the underwriting and claim settlement rules can include, but are not limited to, claim settlement time of insurance, contraindication of purchasing insurance, claim settlement contraindication of insurance, claim settlement amount of insurance, and the like. For example, the contraindication of purchasing a million medical insurance is that the disease of lung nodule cannot be obtained.

[0103] The second construction module is configured to construct a first triadic relation group combination of the insurance name, the disease name, and the underwriting and claim settlement rules as the underwriting and claim settlement graph.

[0104] In a possible implementation, the apparatus further includes:

[0105] The first training module is configured to input each disease name and each drug name into a small sample model for training to obtain a first training result, the first training result including a correspondence between each disease name and each drug name with a matching rate less than a first threshold.

[0106] In some possible implementations, the first threshold can be, but is not limited to, 50%. The matching rate refers to the probability of the correspondence between the disease name and the drug name and the matching between the disease name and the drug name. For example, the disease name of a disease is cold, and the name of a drug is indomethacin. At this time, the correspondence between cold and indomethacin is that they correspond to each other. Therefore, the matching rate of the correspondence is less than 50% because cold can be caused by inflammation, and indomethacin can eliminate inflammation but cannot completely treat cold.

[0107] In some possible implementations, the small sample model is a model constructed based on small sample learning. Since small sample learning requires less data to train the model, it eliminates the high cost related to data collection and labeling. The small amount of training data means that the dimensionality in the training data set is low, which can significantly reduce the computing cost and improve the computing time.

[0108] In some possible implementations, a triad refers to a set of the form ((x, y), z) (that is, a triad is a pair whose first projection is also a pair), which is often abbreviated as (x, y, z). The triad combination is a set of triads.

[0109] The third obtaining module is configured to obtain a first training result of the first training module.

[0110] The second training module is configured to input the third triple combination into a relation extraction model performing semi-supervised learning to perform training, so as to obtain a second training result, the second training result including a corresponding relation between each disease name and each drug name with a matching rate greater than a second threshold.

[0111] In a possible implementation, the second threshold can be, but is not limited to, 98%.

[0112] In a possible implementation, the relation extraction model performing semi-supervised learning can be a relation extraction model with a semi-supervised learning function. The relation extraction model can be, but is not limited to, a relation extraction pipeline (pipeline) model PURE.

[0113] The fourth obtaining module is configured to obtain a second training result of the second training module.

[0114] The first generating module is configured to generate a second triple combination including each disease name, each drug name, and a corresponding relation between each disease name and each drug name with a matching rate less than a first threshold.

[0115] The fifth obtaining module is configured to obtain, as a third triple combination, a second triple combination in which a corresponding relation between each disease name and each drug name is partially labeled.

[0116] The sixth obtaining module is configured to input the third triple combination into a relation extraction model performing semi-supervised learning to perform training, so as to obtain a corresponding relation between each disease name and each drug name with a matching rate greater than a second threshold.

[0117] The second generating module is configured to generate, as a medical graph, a fourth triple combination including each disease name, each drug name, and a corresponding relation between each disease name and each drug name with a matching rate greater than a second threshold.

[0118] In addition, the embodiments of the present application further provide a device for constructing a disease insurance relation graph, the device comprising a memory and a processor, the memory being configured to store programs or codes, and the processor being configured to run the programs or codes stored in the memory to implement the disease insurance relation graph construction method described above.

[0119] In addition, the embodiments of the present application further provide a computer readable storage medium, characterized in that the computer readable storage medium stores codes, and when the codes are run, a device running the codes implements the disease insurance relation graph construction method described above.

[0120] The embodiment of the application provides a kind of construction device of disease insurance relationship graph, first can utilize first acquisition module 201 to obtain disease hierarchy based on medical coding division.Then first construction module 202 constructs medical atlas according to disease name and drug name.Then first integration module 203 integrates disease hierarchy and medical atlas according to the corresponding relationship of disease name contained in first entity layer in medical atlas and disease hierarchy, and obtains first integrated atlas.Second construction module 204 constructs again according to the corresponding relationship of disease name contained in second entity layer in insurance name and disease name, and obtains second integrated atlas.Finally, second integration module 205 integrates first integrated atlas and insurance underwriting and claim atlas according to the corresponding relationship of disease name contained in second entity layer in insurance underwriting and claim atlas and first integrated atlas.The application constructs medical atlas and the relationship atlas of insurance underwriting and claim based on medical coding, realizes the combination of medical atlas and related insurance scene.

[0121] It should be noted that each of the embodiments in the specification adopts a progressive manner, and the same and similar parts between each embodiment can be referred to each other, and each embodiment focuses on the difference from other embodiments. Part or all of them can be selected to achieve the purpose of the embodiment scheme according to the actual needs. Those skilled in the art can understand and implement without creating labor.

[0122] The above description of the disclosed embodiments enables those skilled in the art to implement or use the application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the application. Therefore, the application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a disease insurance relationship graph, characterized in that, The method includes: Obtain the disease hierarchy based on medical codes; A medical atlas is constructed based on disease names and drug names. The medical atlas includes a first entity layer, which includes the disease name and the drug name. Based on the correspondence between disease names and disease levels in the first entity layer, the disease levels are integrated with the medical atlas to obtain a first integrated atlas; An underwriting and claims mapping is constructed based on the insurance name and the disease name. The underwriting and claims mapping includes a second entity layer, which includes the disease name and the insurance name. Based on the correspondence between the disease names in the second entity layer and the first integrated map, the first integrated map and the underwriting and claims map are integrated to obtain the second integrated map; The construction of a medical atlas based on disease names and drug names includes: The disease name and drug name are input into the small sample model for training, and the correspondence between the disease name and drug name with a matching rate of less than the first threshold is obtained. Generate a second triplet combination that includes each disease name, each drug name, and the correspondence between each disease name and each drug name where the matching rate is less than the first threshold. The second triplet combination, which is partially labeled with the correspondence between each disease name and each drug name, is used as the third triplet combination. The third triplet combination is input into a semi-supervised learning relation extraction model for training to obtain the correspondence between each disease name and each drug name with a matching rate greater than the second threshold. A fourth triplet combination is generated, which includes each disease name, each drug name, and the correspondence between each disease name and each drug name with a matching rate greater than the second threshold, as a medical atlas.

2. The method according to claim 1, characterized in that, The construction of the underwriting and claims mapping based on the insurance name and the disease name includes: Obtain the underwriting and claims rules between the insurance name and the disease name; The first ternary combination of the insurance name, the disease name, and the underwriting and claims rules is constructed as the underwriting and claims map.

3. The method according to claim 1, characterized in that, The disease hierarchy includes five disease levels.

4. The method according to claim 1, characterized in that, The medical atlas also includes a first relation layer: The first relationship layer includes the correspondence between the disease name and the drug name in the first entity layer.

5. The method according to claim 1, characterized in that, The underwriting and claims mapping includes a second relationship layer: The second relationship layer includes the correspondence between the disease name and the drug name in the second entity layer.

6. A device for constructing a disease insurance relationship map, characterized in that, The device includes: The first acquisition module is used to acquire the disease level based on medical codes; The first construction module is used to construct a medical atlas based on disease names and drug names; the medical atlas includes a first entity layer; the first entity layer includes the disease name and the drug name; The first integration module is used to integrate the disease level with the medical atlas according to the correspondence between the disease name and the disease level in the first entity layer to obtain a first integrated atlas; The second construction module is used to construct an underwriting and claims graph based on the insurance name and the disease name; the underwriting and claims graph includes a second entity layer; the second entity layer includes the disease name and the insurance name; The second integration module is used to integrate the first integration map and the underwriting and claims map according to the correspondence between the disease names in the second entity layer and the first integration map to obtain the second integration map; The construction of a medical atlas based on disease names and drug names includes: The disease name and drug name are input into the small sample model for training, and the correspondence between the disease name and drug name with a matching rate of less than the first threshold is obtained. Generate a second triplet combination that includes each disease name, each drug name, and the correspondence between each disease name and each drug name where the matching rate is less than the first threshold. The second triplet combination, which is partially labeled with the correspondence between each disease name and each drug name, is used as the third triplet combination. The third triplet combination is input into a semi-supervised learning relation extraction model for training to obtain the correspondence between each disease name and each drug name with a matching rate greater than the second threshold. A fourth triplet combination is generated, which includes each disease name, each drug name, and the correspondence between each disease name and each drug name with a matching rate greater than the second threshold, as a medical atlas.

7. The apparatus according to claim 6, characterized in that, The device further includes: The second acquisition module is used to acquire the underwriting and claims rules between the insurance name and the disease name; The second construction module is used to construct a first ternary relationship group of the insurance name, the disease name and the underwriting and claims rules as the underwriting and claims map.

8. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method for constructing a disease insurance relationship graph as described in any one of claims 1-5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a terminal device, cause the terminal device to perform the method for constructing a disease insurance relationship graph as described in any one of claims 1-5.

Images