Intelligent document retrieval system based on medical knowledge graph

Abstract

The invention is applicable to the technical field of intelligent retrieval, and provides an intelligent document retrieval system based on a medical knowledge graph, which comprises a knowledge graph construction module, a login module, a retrieval module, a storage module and a processing module, the processing module comprises an input unit, an early diagnosis unit, a pushing unit and a registration unit; the input end of the input unit is electrically connected with the retrieval module; the output end of the input unit is electrically connected with the early diagnosis unit; the input end of the pushing unit is electrically connected with the early diagnosis unit; the output end of the pushing unit is electrically connected with the registration unit, and the system constructs an infectious disease knowledge graph, so that a user can efficiently retrieve association between infectious disease knowledge and mining information; and data storage, retrieval, reasoning application, pushing and registration services are provided for constructing an infectious disease knowledge question-answering system, an infectious disease auxiliary diagnosis system, an infectious disease monitoring and early warning system and other applications.

Description

Technical Field

[0001] The present invention relates to the field of intelligent retrieval technology, and more specifically, to a document intelligent retrieval system based on medical knowledge graph. Background Art

[0002] As the medical field has developed to date, it has accumulated a huge amount of medical text data, including electronic medical records, medical books and documents, clinical practices, and Internet diagnoses.

[0003] These text data contain rich medical knowledge, but ordinary text forms cannot efficiently retrieve medical information and cannot mine the relationship between information. It is not conducive to building disease knowledge question-and-answer systems, disease auxiliary diagnosis systems, disease monitoring and early warning systems, etc., nor is it conducive to applications providing data storage, retrieval, reasoning and other services; at the same time, people have a large amount of useless information during the retrieval process, and cannot perform intelligent and fast retrieval, and there is no subsequent related push and registration function. Summary of the invention

[0004] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a document intelligent retrieval system based on a medical knowledge graph. By constructing an infectious disease knowledge graph, users can efficiently retrieve infectious disease knowledge and mine the relationship between information, and provide data storage, retrieval, reasoning application, push and registration services for the construction of infectious disease knowledge question and answer systems, infectious disease auxiliary diagnosis systems, infectious disease monitoring and early warning systems and other applications.

[0005] To achieve the above object, the present invention provides the following technical solutions: A document intelligent retrieval system based on a medical knowledge graph, comprising a knowledge graph construction module, a login module, a retrieval module, a storage module and a processing module; the knowledge graph construction module is connected to the Internet through communication; the storage module is respectively connected to the knowledge graph construction module and the retrieval module through communication; the input end of the retrieval module is electrically connected to the login module; the output end of the retrieval module is electrically connected to the processing module; the processing module comprises an input unit, a preliminary diagnosis unit, a push unit and a registration unit; the input end of the input unit is electrically connected to the retrieval module; the output end of the input unit is electrically connected to the preliminary diagnosis unit; the input end of the push unit is electrically connected to the preliminary diagnosis unit; the output end of the push unit is electrically connected to the registration unit.

[0006] The present invention is further configured as follows: the login unit includes an adult login unit and a child login unit; the adult login unit and the child login unit are both connected to the input end of the search module through communication.

[0007] The present invention is further configured as follows: the knowledge graph construction module includes a data sorting unit, an entity relationship extraction model construction unit and a knowledge graph construction unit.

[0008] The present invention is further configured as follows: the data sorting unit collects disease medical data from medical literature books, electronic medical records, Internet diagnosis and clinical practice through network retrieval and crawler technology, performs data management on the acquired medical disease data, and obtains text information description.

[0009] The present invention is further configured as follows: the storage module stores the knowledge graph network in the graph database neo4j, and uses the graph database neo4j and provides disease query and reasoning services through the retrieval module.

[0010] The present invention is further configured as follows: the entity relationship extraction model construction unit is based on Google's BERT Chinese pre-trained model according to text information, and combined with the entity relationship extraction model CasRel to perform entity recognition and relationship extraction modeling, and is obtained by training and debugging the model using the Chinese medical entity relationship extraction open data set CMeIE.

[0011] The present invention is further configured as follows: the knowledge graph construction unit extracts knowledge from each piece of text information through an entity relationship extraction model, obtains triples of subject, relationship and object, and forms a structure based on the triples.

[0012] The present invention is further configured as follows: the input unit inputs the patient's basic information and symptoms, which are processed by the preliminary diagnosis unit to form an electronic report guide sheet; the processing module is connected to the medical system, and is intelligently pushed through the push unit, and registration is performed through the registration unit according to demand.

[0013] The advantages of the present invention are: The present invention constructs an infectious disease knowledge graph, allowing users to efficiently retrieve infectious disease knowledge and mine the relationship between information, and provides data storage, retrieval, reasoning application, push and registration services for the construction of infectious disease knowledge question and answer systems, infectious disease auxiliary diagnosis systems, infectious disease monitoring and early warning systems and other applications. Patients can log in directly to obtain relevant knowledge through intelligent retrieval, and can get a preliminary diagnosis by entering basic information and symptoms. The system pushes the electronic report guide to the medical treatment system as needed, and can register as needed, which optimizes the treatment process and improves the efficiency of treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a flow chart of an intelligent literature retrieval system based on medical knowledge graph of the present invention.

[0015] Figure 2This is a composition diagram of the knowledge graph building module of the present invention. DETAILED DESCRIPTION

[0016] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application may be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0017] It should be noted that, unless otherwise specified, all technical and scientific terms used in this application have the same meanings as commonly understood by ordinary technicians in the technical field to which this application belongs.

[0018] In the present invention, unless otherwise specified, the directions used, such as "up" and "down", usually refer to the directions shown in the drawings, or to the vertical, perpendicular or gravity directions; similarly, for ease of understanding and description, "left" and "right" usually refer to the left and right shown in the drawings; "inside" and "outside" refer to the inside and outside relative to the outline of each component itself, but the above-mentioned directions are not used to limit the present invention.

[0019] For example, see Figure 1-2 , the present invention provides the following technical solutions: A document intelligent retrieval system based on a medical knowledge graph, specifically, includes a knowledge graph construction module, a login module, a retrieval module, a storage module and a processing module; the knowledge graph construction module is connected to the Internet through communication; the storage module is respectively connected to the knowledge graph construction module and the retrieval module through communication; the input end of the retrieval module is electrically connected to the login module; the output end of the retrieval module is electrically connected to the processing module; the processing module includes an input unit, a preliminary diagnosis unit, a push unit and a registration unit; the input end of the input unit is electrically connected to the retrieval module; the output end of the input unit is electrically connected to the preliminary diagnosis unit; the input end of the push unit is electrically connected to the preliminary diagnosis unit; the output end of the push unit is electrically connected to the registration unit.

[0020] Working principle of the first embodiment: By constructing an infectious disease knowledge graph, users can efficiently retrieve infectious disease knowledge and mine the relationship between information, and provide data storage, retrieval, reasoning application, push and registration services for building infectious disease knowledge question and answer systems, infectious disease auxiliary diagnosis systems, infectious disease monitoring and early warning systems and other applications. Patients can log in directly to obtain relevant knowledge through intelligent retrieval, and can get a preliminary diagnosis by entering basic information and symptoms. The system pushes the electronic report guide to the medical treatment system as needed, and can register as needed, which optimizes the treatment process and improves the efficiency of treatment.

[0021] For example 2, please refer to Figure 1-2, this embodiment 2 makes the following improvements on the basis of embodiment 1, specifically. The login unit includes an adult login unit and a child login unit; the adult login unit and the child login unit are both connected to the input end of the retrieval module through communication; the knowledge graph construction module includes a data sorting unit, an entity relationship extraction model construction unit and a knowledge graph construction unit; the data sorting unit collects disease medical data from medical literature books, electronic medical records, Internet diagnosis and clinical practice through network retrieval and crawler technology, performs data governance on the acquired medical disease data, and obtains text information description; the storage module stores the knowledge graph network in the graph database neo4j, uses the graph database neo4j and provides disease query inference through the retrieval module management service; the entity relationship extraction model construction unit is based on the text information based on Google's BERT Chinese pre-trained model, and combined with the entity relationship extraction model CasRel for entity recognition and relationship extraction modeling, and the Chinese medical entity relationship extraction open data set CMeIE training and debugging model is used to obtain the knowledge graph construction unit; the knowledge graph construction unit is to extract knowledge from each piece of text information through the entity relationship extraction model, and obtain the triples of subject, relationship and object, and form it according to the triples; the input unit is to enter the patient's basic information and symptoms, and form an electronic report guide after processing by the preliminary diagnosis unit. The processing module is connected to the medical system, and the push unit pushes intelligently, and the registration is carried out through the registration unit according to the needs.

[0022] Working principle of the second embodiment: Data governance is performed on the acquired infectious disease medical data, where rules are set for structured data, and the data is directly stored in the corpus after processing. For unstructured data, natural language processing technology is used to extract diseases, symptoms and correlations in the text and then store them in the corpus; the BERT Chinese pre-trained model BERT-wwm with full word coverage is used to encode medical Chinese text vectors, and the encoding end of the entity relationship extraction model CasRel uses the BERT encoding layer to obtain contextual semantic information to represent the characters / words. The encoded information is decoded through the head entity recognition layer of the decoding end to identify all possible main entities, and the main entity encoding vector and the BERT encoding layer encoding vector are obtained through the relationship and tail entity joint recognition layer of the decoding end. All set relationships are traversed to identify objects that may be related to the subject; knowledge is merged, ambiguity is eliminated, and heterogeneous problems are handled in multiple entity relationship descriptions.

[0023] Neo4j is an open source NoSQL graph database that provides native graph data storage, retrieval, and processing. It uses the Property graph model, which greatly improves and enriches the graph data model. It can easily represent connected data. Retrieving / traversing / navigating more connected data is very easy and fast. It can easily represent semi-structured data. Neo4j CQL query statement commands are in a human-readable format, which is very easy to learn. Using a simple and powerful data model does not require complex connections to retrieve connected / related data, because it is easy to retrieve its adjacent nodes or relationship details without connections or indexes.

[0024] Relational Triple Extraction (RTE), also known as entity-relation joint extraction, is a classic task in the field of information extraction. It aims to extract structured relation triplets (Subject, Relation, Object) from text to build a knowledge graph. Sometimes it is also collectively referred to as relation extraction (RE) with the relation classification (RC) task; the main difference between RC and RTE is that RC extracts the relationship expressed by entity pairs in sentences given entity pairs and input text; while RTE extracts all possible relation triplets contained in the text given only input text; information extraction is divided into two parts, one is named entity recognition, which identifies entities in the text, and the other is relation extraction, which constructs corresponding relationships for the identified entities. The two are the basic components for building triples.

[0025] Extracting the open dataset CMeIE training and debugging model is a method that uses entity and relationship extraction as an important subtask of information extraction, applies these technologies to the medical field, extracts unstructured and semi-structured medical texts to build medical knowledge graphs, and can serve downstream subtasks. Unstructured medical texts, such as every natural paragraph in medical textbooks, the topics under each disease in clinical practice, the chief complaint, current medical history, differential diagnosis, etc. in electronic medical record data, are all composed of Chinese natural language sentences or sentence sets. Entity relationship extraction is the process of finding medical entities from unstructured medical texts and determining entity-to-relationship facts.

[0026] Obviously, the above-described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0029] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0030] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A document intelligent retrieval system based on medical knowledge graph, characterized by: It includes knowledge graph construction module, login module, retrieval module, storage module and processing module; The knowledge graph construction module is connected to the Internet via communication; The storage module is respectively connected with the knowledge graph construction module and the retrieval module through communication; The input end of the retrieval module is electrically connected to the login module; the output end of the retrieval module is electrically connected to the processing module; The processing module includes an input unit, a preliminary diagnosis unit, a push unit and a registration unit; The input end of the input unit is electrically connected to the search module; the output end of the input unit is electrically connected to the early diagnosis unit; The push unit input end is electrically connected to the early diagnosis unit; the push unit output end is electrically connected to the registration unit.

2. According to claim 1, a document intelligent retrieval system based on medical knowledge graph is characterized by: The login unit includes an adult login unit and a child login unit; The adult login unit and the child login unit are both connected to the input end of the retrieval module through communication.

3. According to claim 2, a document intelligent retrieval system based on medical knowledge graph is characterized in that: The knowledge graph construction module includes a data sorting unit, an entity relationship extraction model construction unit and a knowledge graph construction unit.

4. According to claim 3, a document intelligent retrieval system based on medical knowledge graph is characterized in that: The data sorting unit collects disease medical data from medical literature books, electronic medical records, Internet diagnosis and clinical practice through network retrieval and crawler technology, performs data management on the acquired medical disease data, and obtains text information description.

5. According to claim 4, a document intelligent retrieval system based on medical knowledge graph is characterized in that: The storage module stores the knowledge graph network in the graph database neo4j, and uses the graph database neo4j and provides disease query and reasoning services through the retrieval module.

6. According to claim 5, a document intelligent retrieval system based on medical knowledge graph is characterized in that: The entity relationship extraction model construction unit is based on Google's BERT Chinese pre-trained model according to text information, and combines the entity relationship extraction model CasRel to perform entity recognition and relationship extraction modeling, and uses the Chinese medical entity relationship extraction open data set CMeIE to train and debug the model.

7. The intelligent document retrieval system based on medical knowledge graph according to claim 6 is characterized by: The knowledge graph construction unit extracts knowledge from each piece of text information through an entity relationship extraction model, obtains triples of subject, relationship and object, and forms a structure based on the triples.

8. The intelligent document retrieval system based on medical knowledge graph according to claim 7 is characterized by: The input unit inputs the patient's basic information and symptoms, which are processed by the preliminary diagnosis unit to form an electronic report guide. The processing module is connected to the medical system and is intelligently pushed through the push unit. Registration is performed through the registration unit according to demand.

Images