CN112883736A - Medical entity relationship extraction method and device - Google Patents
Medical entity relationship extraction method and device Download PDFInfo
- Publication number
- CN112883736A CN112883736A CN202110197630.2A CN202110197630A CN112883736A CN 112883736 A CN112883736 A CN 112883736A CN 202110197630 A CN202110197630 A CN 202110197630A CN 112883736 A CN112883736 A CN 112883736A
- Authority
- CN
- China
- Prior art keywords
- entity
- medical
- extraction
- relation
- medical record
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000605 extraction Methods 0.000 title claims abstract description 109
- 230000007246 mechanism Effects 0.000 claims abstract description 33
- 238000000034 method Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims description 20
- 238000004364 calculation method Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 11
- 230000002452 interceptive effect Effects 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000004140 cleaning Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 6
- 238000012512 characterization method Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000003058 natural language processing Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 208000010839 B-cell chronic lymphocytic leukemia Diseases 0.000 description 4
- 208000031422 Lymphocytic Chronic B-Cell Leukemia Diseases 0.000 description 4
- 230000006978 adaptation Effects 0.000 description 4
- 208000032852 chronic lymphocytic leukemia Diseases 0.000 description 4
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 3
- 206010018910 Haemolysis Diseases 0.000 description 3
- 208000007502 anemia Diseases 0.000 description 3
- 230000008588 hemolysis Effects 0.000 description 3
- 239000007787 solid Substances 0.000 description 3
- 239000008280 blood Substances 0.000 description 2
- 210000004369 blood Anatomy 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
Landscapes
- Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The application discloses a medical entity relationship extraction method and device. The method comprises the steps of obtaining a medical electronic medical record; performing entity extraction on the medical electronic medical record based on a BERT part of a combined extraction model to obtain each entity corresponding to the medical electronic medical record; and performing relation prediction on each entity corresponding to the medical electronic medical record based on a multi-head selection mechanism part of a combined extraction model to obtain a relation classification result corresponding to each entity. Therefore, the problem that the accuracy of the entity relationship of the combined extraction is low due to the low judgment capability of the medical entity relationship in the related technology is solved.
Description
Technical Field
The application relates to the technical field of data processing, in particular to a medical entity relationship extraction method and device.
Background
Currently, medical records of most hospitals in China are recorded in natural language, and the unstructured medical records cannot be directly used by a machine and need to be converted into structured information through a Natural Language Processing (NLP) technology so as to be processed by the machine. With the development of medical informatization, key information is accurately and quickly extracted from massive Electronic Medical Records (EMR), and a structured model conforming to medical specifications is constructed, so that the method becomes a key step for secondary use of EMR data. The case history structuring is mainly based on an information extraction technology in NLP, and relates to entity extraction, relationship extraction, entity standardization and the like. Wherein the extraction of entities and relationships is a key step of the structuring of medical records.
Entity and Relationship Extraction (ERE) is one of the key tasks of information Extraction. ERE is a cascading task that is divided into two subtasks: entity extraction and relationship extraction. Common treatment methods fall into two broad categories: the first type is a pipeline mode, entities are extracted firstly, and then relationship classification is carried out, so that error accumulation exists, the internal relation between two tasks of the entities and the relationship is neglected, and the problems of entity redundancy calculation and the like exist at the same time; the second type is a joint extraction mode, and entities and relations share the same network code, so that error propagation is relieved. In summary, the mainstream approach to ERE today is joint extraction, which can generally translate into a multitask learning or structured prediction problem. Moreover, the ERE task based on joint extraction can be divided mainly from three perspectives: 1) different labeling frameworks; 2) different extraction sequences; 3) whether to share the coding layer.
In recent years, the pre-trained language model has become an indispensable "killer" in the NLP field. For example, BERTs from Google in 2018, once released, placed the leaders of many NLP leaderboards. Therefore, it has become a first step to perform finetune using the pretrained language model such as BERT as the basic encoder of ERE task. For medical entity relationship joint extraction, since the BERT is language model pre-training based on general corpus, the problem of field adaptation exists when the pre-training model BERT is directly applied to medical texts. The existing medical entity relation technology does not perform combined extraction based on a Chinese medical pre-training model with field adaptation.
In recent years, a joint extraction model based on a multi-head selection mechanism is widely used for entity relationship extraction, and entity extraction is performed first, and then relationship classification is performed. The multi-head selection mechanism is to construct a relation classifier aiming at the entity pair, namely, each entity pair only selects the last character of the current entity segment to carry out relation prediction. In fact, how to extract text features to construct a relational classifier is a key point for improving the performance of the model. The existing technical method does not consider how to fuse the BERT pre-training model with a multi-head selection mechanism so as to strengthen the discrimination capability of the model relation.
In the process of implementing the embodiment of the present application, the inventor finds that the related art has at least the following problems:
in the related technology, the judgment capability of the medical entity relationship is low, so that the accuracy of the entity relationship of the combined extraction is low.
Disclosure of Invention
The main objective of the present application is to provide a method and an apparatus for extracting a medical entity relationship, so as to solve the problem in the related art that the accuracy of a jointly extracted entity relationship is low due to low discrimination capability of the medical entity relationship.
In order to achieve the above object, in a first aspect, an embodiment of the present application provides a medical entity relationship extraction method, including:
acquiring a medical electronic medical record;
performing entity extraction on the medical electronic medical record based on a BERT part of a combined extraction model to obtain each entity corresponding to the medical electronic medical record;
and performing relation prediction on each entity corresponding to the medical electronic medical record based on a multi-head selection mechanism part of a combined extraction model to obtain a relation classification result corresponding to each entity.
Optionally, the performing, by the multi-head selection mechanism part based on the joint extraction model, relationship prediction on each entity corresponding to the medical electronic medical record includes:
obtaining subject representations and object representations corresponding to each entity corresponding to the medical electronic medical record;
performing Biaffine interactive calculation on the subject token representation and the object representation to obtain a relation classification matrix of NxNxC, wherein N is the sequence length, and C is the total number of relation categories;
and performing softmax calculation on the relation classification matrix to obtain a relation classification result corresponding to each entity.
Optionally, the entity extraction of the medical electronic medical record based on the BERT part of the joint extraction model includes:
and extracting the entity by adopting a pointer network when extracting the entity.
Optionally, the method further includes:
cleaning based on medical electronic medical record big data to obtain pre-trained corpora;
performing vocabulary word segmentation on the corpus according to a medical field dictionary to obtain a training corpus;
and pre-training the language model of the training corpus based on the covering language model to obtain a BERT part of the combined extraction model.
Optionally, the method further includes:
and inserting a relation classifier based on a multi-head selection mechanism at the tail part of the BERT part to obtain a multi-head selection mechanism part of the combined extraction model.
In a second aspect, the present application further provides a medical entity relationship extraction apparatus, the apparatus comprising:
the acquisition module is used for acquiring the medical electronic medical record;
the entity extraction module is used for performing entity extraction on the medical electronic medical record based on a BERT part of the combined extraction model to obtain each entity corresponding to the medical electronic medical record;
and the relation extraction module is used for carrying out relation prediction on each entity corresponding to the medical electronic medical record based on the multi-head selection mechanism part of the joint extraction model to obtain a relation classification result corresponding to each entity.
Optionally, the relationship extracting module is configured to:
obtaining subject representations and object representations corresponding to each entity corresponding to the medical electronic medical record;
performing Biaffine interactive calculation on the subject token representation and the object representation to obtain a relation classification matrix of NxNxC, wherein N is the sequence length, and C is the total number of relation categories;
and performing softmax calculation on the relation classification matrix to obtain a relation classification result corresponding to each entity.
Optionally, the entity extraction module is configured to:
and extracting the entity by adopting a pointer network when extracting the entity.
In a third aspect, the present application further provides a computer device, including: a memory and a processor;
the memory is used for storing a computer program;
the processor is configured to execute a computer program stored in the memory;
the computer program is used for executing the medical entity relationship extraction method.
In a fourth aspect, the present application also provides a computer-readable storage medium storing computer code which, when executed, performs the medical entity relationship extraction method as described above.
In the medical entity relationship extraction method provided by the embodiment of the application, a medical electronic medical record is acquired; performing entity extraction on the medical electronic medical record based on a BERT part of a combined extraction model to obtain each entity corresponding to the medical electronic medical record; and performing relation prediction on each entity corresponding to the medical electronic medical record based on a multi-head selection mechanism part of a combined extraction model to obtain a relation classification result corresponding to each entity. Therefore, the method of fusing the medical BERT and the multi-head selection mechanism applies the Chinese medical BERT to the medical entity relationship so as to better obtain the medical text language representation and perform field adaptation; meanwhile, a multi-head selection mechanism is fused with the BERT, so that the medical BERT and a feedforward neural network are interacted to construct a double affine mechanism, the relationship prediction characterization capability is better enhanced, the purpose of jointly extracting the medical entity relationship which can not fuse the pre-training model in the medical field and the multi-head selection mechanism is realized, and the problem of low accuracy of the jointly extracted entity relationship caused by low medical entity relationship discrimination capability in the related technology is solved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, serve to provide a further understanding of the application and to enable other features, objects, and advantages of the application to be more apparent. The drawings and their description illustrate the embodiments of the invention and do not limit it. In the drawings:
fig. 1 is a flowchart of a medical entity relationship extraction method according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a medical pre-training language model according to an embodiment of the present application;
FIG. 3 is a diagram of a medical entity relationship triplet according to an embodiment of the present application;
FIG. 4 is a diagram illustrating a method for labeling relationship triples of medical entities according to an embodiment of the present application;
FIG. 5 is a schematic diagram of a medical entity relationship extraction model fusing a domain pre-training model and a multi-head selection mechanism according to an embodiment of the present application;
fig. 6 is a schematic structural diagram of a medical entity relationship extraction device according to an embodiment of the present application.
Detailed Description
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only partial embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used may be interchanged under appropriate circumstances such that embodiments of the application described herein may be used. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
An embodiment of the present application provides a medical entity relationship extraction method, fig. 1 is a flowchart of the medical entity relationship extraction method provided in the embodiment of the present application, and as shown in fig. 1, the medical entity relationship extraction method provided in the embodiment of the present application includes the following steps 100 to 300:
100, acquiring the medical electronic medical record.
The medical electronic medical record can be input by a user, can be stored in a local storage, and can also be stored in a cloud server.
And 200, performing entity extraction on the medical electronic medical record based on a BERT part of the combined extraction model to obtain each entity corresponding to the file to be extracted.
The BERT part of the combined extraction model is constructed and mainly based on the open source BERT of Google for continuous pre-training, as shown in figure 2, the corpus of the continuous pre-training is cleaned and obtained based on medical electronic medical record big data, meanwhile, a medical field dictionary is introduced for word segmentation to construct the corpus, and the medical BERT is obtained finally by adopting a covering language model in a pre-training mode target. The BERT part construction method of the combined extraction model comprises the following steps:
cleaning based on medical electronic medical record big data to obtain pre-trained corpora;
performing vocabulary word segmentation on the corpus according to a medical field dictionary to obtain a training corpus;
and pre-training the language model of the training corpus based on the covering language model to obtain a BERT part of the combined extraction model.
Specifically, the existing medical entity relationship extraction is usually converted into subject (subject), predicate (predicate), and object (object) to construct a triple relationship, and there are 2 triplets (chronic lymphocytic leukemia, complications, hemolysis) and (chronic lymphocytic leukemia, complications, secondary anemia) in the sample example shown in fig. 3. Therefore, the BERT part of the combined extraction model only needs to perform entity extraction to obtain each entity corresponding to the file to be extracted.
In addition, optionally, the performing entity extraction on the file to be extracted based on the BERT part of the joint extraction model includes:
and extracting the entity by adopting a pointer network when extracting the entity.
Specifically, when the entity extraction is performed, a pointer network can be used for extraction, so that the problem of nested entities can be solved.
And 300, performing relation prediction on each entity corresponding to the file to be extracted based on a multi-head selection mechanism part of the joint extraction model to obtain a relation classification result corresponding to each entity.
The multi-head selection mechanism is to construct a relation classifier for the entity pairs, that is, each entity pair only selects the last character of the current entity segment for relation prediction.
Specifically, when a multiple-head selection mechanism is adopted, three medical entities, namely chronic lymphocytic leukemia, hemolysis and secondary anemia, are extracted, and then the relationship prediction is performed based on the last character of each entity segment: as shown in fig. 4, the relationship prediction is performed only on the relationship prediction between "disease" in the solid "chronic lymphocytic leukemia" and "blood" in the solid "hemolysis", or on the relationship prediction between "blood" in the solid "secondary anemia".
Optionally, step 300 specifically includes:
obtaining a subject representation and an object representation corresponding to each entity corresponding to the file to be extracted;
performing Biaffine interactive calculation on the subject token representation and the object representation to obtain a relation classification matrix of NxNxC, wherein N is the sequence length, and C is the total number of relation categories;
and performing softmax calculation on the relation classification matrix to obtain a relation classification result corresponding to each entity.
In this embodiment, an embodiment of the present application provides a method for extracting a medical entity relationship, further including:
and inserting a relation classifier based on a multi-head selection mechanism at the tail part of the BERT part to obtain a multi-head selection mechanism part of the combined extraction model.
Specifically, in order to better fuse the medical BERT with the multi-head selection mechanism, the embodiment of the present application constructs a double affine mechanism (Biaffine Classifier) model, as shown in fig. 5, that is, the last two layers of BERT are respectively fed into two feed-forward neural networks, one neural network represents a subject subljet characterization, the other neural network represents an object characterization, and then the subject sublet characterization and the object characterization are subjected to Biaffine interactive computation (Biaffine computation is nonlinear computation) to obtain a relation classification matrix of nxnxnxnxxc, where N is a sequence length and C is a total number of relation classes. And finally, performing softmax calculation on the relation classification matrix to realize the relation classification.
In the medical entity relationship extraction method provided by the embodiment of the application, a file to be extracted is obtained; performing entity extraction on the file to be extracted based on a BERT part of a combined extraction model to obtain each entity corresponding to the file to be extracted; and performing relation prediction on each entity corresponding to the file to be extracted based on a multi-head selection mechanism part of the combined extraction model to obtain a relation classification result corresponding to each entity. Therefore, the method of fusing the medical BERT and the multi-head selection mechanism applies the Chinese medical BERT to the medical entity relationship so as to better obtain the medical text language representation and perform field adaptation; meanwhile, a multi-head selection mechanism is fused with the BERT, so that the medical BERT and a feedforward neural network are interacted to construct a double affine mechanism, the relationship prediction characterization capability is better enhanced, the purpose of jointly extracting the medical entity relationship which can not fuse the pre-training model in the medical field and the multi-head selection mechanism is realized, and the problem of low accuracy of the jointly extracted entity relationship caused by low medical entity relationship discrimination capability in the related technology is solved.
Based on the same technical concept, the present application further provides a medical entity relationship extraction device, fig. 6 is a schematic structural diagram of the medical entity relationship extraction device provided in the embodiment of the present application, and as shown in fig. 6, the device includes:
the acquisition module 10 is used for acquiring a file to be extracted;
an entity extraction module 20, configured to perform entity extraction on the file to be extracted based on a BERT portion of a joint extraction model to obtain each entity corresponding to the file to be extracted;
and the relation extraction module 30 is configured to perform relation prediction on each entity corresponding to the file to be extracted based on a multi-head selection mechanism part of the joint extraction model, so as to obtain a relation classification result corresponding to each entity.
The medical entity relationship extraction device of this embodiment is used for a medical entity relationship extraction method, and therefore, a specific implementation manner of the device can be seen in the foregoing embodiment section of the medical entity relationship extraction method, and the specific implementation manner thereof may refer to descriptions of corresponding respective section embodiments, and is not described herein again.
Optionally, the relationship extracting module 30 is configured to:
obtaining a subject representation and an object representation corresponding to each entity corresponding to the file to be extracted;
performing Biaffine interactive calculation on the subject token representation and the object representation to obtain a relation classification matrix of NxNxC, wherein N is the sequence length, and C is the total number of relation categories;
and performing softmax calculation on the relation classification matrix to obtain a relation classification result corresponding to each entity.
Optionally, the entity extraction module 20 is configured to:
and extracting the entity by adopting a pointer network when extracting the entity.
Based on the same technical concept, the present application also provides a computer device, comprising: a memory and a processor;
the memory is used for storing a computer program;
the processor is configured to execute a computer program stored in the memory;
the computer program is used for executing the medical entity relationship extraction method.
Based on the same technical concept, the present application also provides a computer-readable storage medium storing computer code, and when the computer code is executed, the method for processing retrieval matching of medical semantics as described above is executed.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working process of the computer-readable storage medium described above may refer to the corresponding process in the foregoing method embodiments, and is not described herein again.
It will be apparent to those skilled in the art that the modules or steps of the present invention described above may be implemented by a general purpose computing device, they may be centralized on a single computing device or distributed across a network of multiple computing devices, and they may alternatively be implemented by program code executable by a computing device, such that they may be stored in a storage device and executed by a computing device, or fabricated separately as individual integrated circuit modules, or fabricated as a single integrated circuit module from multiple modules or steps. Thus, the present invention is not limited to any specific combination of hardware and software.
The computer program referred to in the present application may be stored in a computer-readable storage medium, which may include: any physical device capable of carrying computer program code, virtual device, flash disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only computer Memory (ROM), Random Access computer Memory (RAM), electrical carrier wave signal, telecommunications signal, and other software distribution media, and the like.
It will be apparent to those skilled in the art that the modules or steps of the present invention described above may be implemented by a general purpose computing device, they may be centralized on a single computing device or distributed across a network of multiple computing devices, and they may alternatively be implemented by program code executable by a computing device, such that they may be stored in a storage device and executed by a computing device, or fabricated separately as individual integrated circuit modules, or fabricated as a single integrated circuit module from multiple modules or steps. Thus, the present invention is not limited to any specific combination of hardware and software.
The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (10)
1. A medical entity relationship extraction method, the method comprising:
acquiring a medical electronic medical record;
performing entity extraction on the medical electronic medical record based on a BERT part of a combined extraction model to obtain each entity corresponding to the medical electronic medical record;
and performing relation prediction on each entity corresponding to the medical electronic medical record based on a multi-head selection mechanism part of a combined extraction model to obtain a relation classification result corresponding to each entity.
2. The medical entity relationship extraction method according to claim 1, wherein the performing of the relationship prediction on each entity corresponding to the medical electronic medical record by the multi-head selection mechanism part based on the joint extraction model comprises:
obtaining subject representations and object representations corresponding to each entity corresponding to the medical electronic medical record;
performing Biaffine interactive calculation on the subject token representation and the object representation to obtain a relation classification matrix of NxNxC, wherein N is the sequence length, and C is the total number of relation categories;
and performing softmax calculation on the relation classification matrix to obtain a relation classification result corresponding to each entity.
3. The medical entity relationship extraction method of claim 1, wherein the entity extraction of the medical electronic medical record based on the BERT part of the joint extraction model comprises:
and extracting the entity by adopting a pointer network when extracting the entity.
4. The medical entity relationship extraction method of claim 1, further comprising:
cleaning based on medical electronic medical record big data to obtain pre-trained corpora;
performing vocabulary word segmentation on the corpus according to a medical field dictionary to obtain a training corpus;
and pre-training the language model of the training corpus based on the covering language model to obtain a BERT part of the combined extraction model.
5. The medical entity relationship extraction method of claim 1, further comprising:
and inserting a relation classifier based on a multi-head selection mechanism at the tail part of the BERT part to obtain a multi-head selection mechanism part of the combined extraction model.
6. A medical entity relationship extraction apparatus, the apparatus comprising:
the acquisition module is used for acquiring the medical electronic medical record;
the entity extraction module is used for performing entity extraction on the medical electronic medical record based on a BERT part of the combined extraction model to obtain each entity corresponding to the medical electronic medical record;
and the relation extraction module is used for carrying out relation prediction on each entity corresponding to the medical electronic medical record based on the multi-head selection mechanism part of the joint extraction model to obtain a relation classification result corresponding to each entity.
7. The medical entity relationship extraction device of claim 6, wherein the relationship extraction module is to:
obtaining subject representations and object representations corresponding to each entity corresponding to the medical electronic medical record;
performing Biaffine interactive calculation on the subject token representation and the object representation to obtain a relation classification matrix of NxNxC, wherein N is the sequence length, and C is the total number of relation categories;
and performing softmax calculation on the relation classification matrix to obtain a relation classification result corresponding to each entity.
8. The medical entity relationship extraction device of claim 6, wherein the entity extraction module is to:
and extracting the entity by adopting a pointer network when extracting the entity.
9. A computer device, comprising: a memory and a processor;
the memory is used for storing a computer program;
the processor is configured to execute a computer program stored in the memory;
the computer program for performing the medical entity relationship extraction method of any one of claims 1-5.
10. A computer readable storage medium storing computer code which, when executed, performs a medical entity relationship extraction method as claimed in any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110197630.2A CN112883736A (en) | 2021-02-22 | 2021-02-22 | Medical entity relationship extraction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110197630.2A CN112883736A (en) | 2021-02-22 | 2021-02-22 | Medical entity relationship extraction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112883736A true CN112883736A (en) | 2021-06-01 |
Family
ID=76056775
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110197630.2A Pending CN112883736A (en) | 2021-02-22 | 2021-02-22 | Medical entity relationship extraction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112883736A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023092719A1 (en) * | 2021-11-29 | 2023-06-01 | 中国科学院深圳先进技术研究院 | Information extraction method for medical record data, and terminal device and readable storage medium |
| CN117391092A (en) * | 2023-12-12 | 2024-01-12 | 中南大学 | Electronic medical record multi-mode medical semantic alignment method based on contrast learning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109243616A (en) * | 2018-06-29 | 2019-01-18 | 东华大学 | Mammary gland electronic health record joint Relation extraction and architectural system based on deep learning |
| CN110059320A (en) * | 2019-04-23 | 2019-07-26 | 腾讯科技（深圳）有限公司 | Entity relation extraction method, apparatus, computer equipment and storage medium |
| CN111274394A (en) * | 2020-01-16 | 2020-06-12 | 重庆邮电大学 | Method, device and equipment for extracting entity relationship and storage medium |
| CN111476023A (en) * | 2020-05-22 | 2020-07-31 | 北京明朝万达科技股份有限公司 | Method and device for identifying entity relationship |
| CN111488726A (en) * | 2020-03-31 | 2020-08-04 | 成都数之联科技有限公司 | Pointer network-based unstructured text extraction multi-task joint training method |
| CN111666427A (en) * | 2020-06-12 | 2020-09-15 | 长沙理工大学 | Entity relationship joint extraction method, device, equipment and medium |
| CN112364174A (en) * | 2020-10-21 | 2021-02-12 | 山东大学 | Patient medical record similarity evaluation method and system based on knowledge graph |
-
2021
- 2021-02-22 CN CN202110197630.2A patent/CN112883736A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109243616A (en) * | 2018-06-29 | 2019-01-18 | 东华大学 | Mammary gland electronic health record joint Relation extraction and architectural system based on deep learning |
| CN110059320A (en) * | 2019-04-23 | 2019-07-26 | 腾讯科技（深圳）有限公司 | Entity relation extraction method, apparatus, computer equipment and storage medium |
| CN111274394A (en) * | 2020-01-16 | 2020-06-12 | 重庆邮电大学 | Method, device and equipment for extracting entity relationship and storage medium |
| CN111488726A (en) * | 2020-03-31 | 2020-08-04 | 成都数之联科技有限公司 | Pointer network-based unstructured text extraction multi-task joint training method |
| CN111476023A (en) * | 2020-05-22 | 2020-07-31 | 北京明朝万达科技股份有限公司 | Method and device for identifying entity relationship |
| CN111666427A (en) * | 2020-06-12 | 2020-09-15 | 长沙理工大学 | Entity relationship joint extraction method, device, equipment and medium |
| CN112364174A (en) * | 2020-10-21 | 2021-02-12 | 山东大学 | Patient medical record similarity evaluation method and system based on knowledge graph |
Non-Patent Citations (1)
| Title |
|---|
| WEIPENG HUANG ET AL.: "BERT-Based Multi-Head Selection for Joint Entity-Relation Extraction", 《ARXIV:1908.05908V2 [CS.CL] 26 SEP 2019》, pages 1 - 12 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023092719A1 (en) * | 2021-11-29 | 2023-06-01 | 中国科学院深圳先进技术研究院 | Information extraction method for medical record data, and terminal device and readable storage medium |
| CN117391092A (en) * | 2023-12-12 | 2024-01-12 | 中南大学 | Electronic medical record multi-mode medical semantic alignment method based on contrast learning |
| CN117391092B (en) * | 2023-12-12 | 2024-03-08 | 中南大学 | Electronic medical record multi-mode medical semantic alignment method based on contrast learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10725836B2 (en) | Intent-based organisation of APIs | |
| US11521713B2 (en) | System and method for generating clinical trial protocol design document with selection of patient and investigator | |
| US9613185B2 (en) | Influence filtering in graphical models | |
| JP2018190188A (en) | Summary creating device, summary creating method and computer program | |
| RU2679988C1 (en) | Extracting information objects with the help of a classifier combination | |
| CN111401928B (en) | Method and device for determining semantic similarity of text based on graph data | |
| US20220100963A1 (en) | Event extraction from documents with co-reference | |
| CN112084789A (en) | Text processing method, device, equipment and storage medium | |
| Zhong et al. | A comprehensive survey on automatic knowledge graph construction | |
| US20220100772A1 (en) | Context-sensitive linking of entities to private databases | |
| CN113535986B (en) | Data fusion method and device applied to medical knowledge graph | |
| CN112883736A (en) | Medical entity relationship extraction method and device | |
| US10191921B1 (en) | System for expanding image search using attributes and associations | |
| CN111177375A (en) | Electronic document classification method and device | |
| CN115798661A (en) | Knowledge mining method and device in clinical medicine field | |
| Li et al. | [Retracted] Knowledge Graph Entity Similarity Calculation under Active Learning | |
| US20240012809A1 (en) | Artificial intelligence system for translation-less similarity analysis in multi-language contexts | |
| US20220100967A1 (en) | Lifecycle management for customized natural language processing | |
| CN110489740A (en) | Semantic analytic method and Related product | |
| CN113569581B (en) | Intention recognition method, device, equipment and storage medium | |
| CN113392220B (en) | Knowledge graph generation method and device, computer equipment and storage medium | |
| US20210303802A1 (en) | Program storage medium, information processing apparatus and method for encoding sentence | |
| US11347928B2 (en) | Detecting and processing sections spanning processed document partitions | |
| CN111767710B (en) | Indonesia emotion classification method, device, equipment and medium | |
| CN113807920A (en) | Artificial intelligence based product recommendation method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |