WO2022160614A1 - Method and apparatus for constructing medical entity relationship diagram, method and apparatus for medical order quality control, device, and medium - Google Patents

Method and apparatus for constructing medical entity relationship diagram, method and apparatus for medical order quality control, device, and medium Download PDF

Info

Publication number
WO2022160614A1
WO2022160614A1 PCT/CN2021/106769 CN2021106769W WO2022160614A1 WO 2022160614 A1 WO2022160614 A1 WO 2022160614A1 CN 2021106769 W CN2021106769 W CN 2021106769W WO 2022160614 A1 WO2022160614 A1 WO 2022160614A1
Authority
WO
WIPO (PCT)
Prior art keywords
medical
entity
entities
medical entity
drug
Prior art date
Application number
PCT/CN2021/106769
Other languages
French (fr)
Chinese (zh)
Inventor
施振辉
夏源
陆超
代小亚
黄海峰
Original Assignee
北京百度网讯科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 北京百度网讯科技有限公司 filed Critical 北京百度网讯科技有限公司
Publication of WO2022160614A1 publication Critical patent/WO2022160614A1/en

Links

Images

Classifications

    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/10ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/367Ontology
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00Handling natural language data
    • G06F40/20Natural language analysis
    • G06F40/279Recognition of textual entities

Definitions

  • the present application relates to the field of artificial intelligence technology, such as deep learning technology, such as a method and device for constructing a medical entity relationship graph, and a method and device, equipment, and medium for quality control of medical orders.
  • artificial intelligence technology such as deep learning technology
  • a method and device for constructing a medical entity relationship graph such as a method and device for constructing a medical entity relationship graph, and a method and device, equipment, and medium for quality control of medical orders.
  • the present application provides a method and device for constructing a medical entity relationship diagram, a method and device, equipment and medium for quality control of a doctor's order.
  • a method for building a medical entity relationship diagram including:
  • a medical entity relationship graph is constructed according to the at least two medical entities and the drug classification labels associated with the at least two medical entities.
  • a method for quality control of doctor's orders including:
  • the first medical entity refers to the medical entity whose entity type is symptom
  • the second medical entity refers to the medical entity whose entity type is medicine
  • the quality control information between the first medical entity and the second medical entity is determined; wherein, the entity relationship diagram is constructed by using the above method for constructing a medical entity relationship diagram.
  • Also provided is an apparatus for constructing a medical entity relationship diagram including:
  • the entity recognition module is set to perform entity recognition on medical-related data to obtain at least two medical entities
  • the label prediction module is configured to determine the drug classification labels associated with at least two medical entities
  • the relationship diagram building module is configured to construct a medical entity relationship diagram according to the at least two medical entities and the drug classification labels associated with the at least two medical entities.
  • a device for quality control of a doctor's order comprising:
  • the entity acquisition module is configured to acquire the first medical entity and the second medical entity from the medical order medical record data to be detected; wherein, the first medical entity refers to the medical entity whose entity type is symptom, and the second medical entity refers to the entity type is the medical entity of the drug;
  • the quality control module is configured to use an entity relationship diagram to determine the quality control information between the first medical entity and the second medical entity; wherein the entity relationship diagram is constructed by using the above method for constructing a medical entity relationship diagram.
  • an electronic device comprising:
  • the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can execute the above-mentioned method for constructing a medical entity relationship diagram or a method for quality control of a doctor's order.
  • a non-transitory computer-readable storage medium storing computer instructions is also provided, and the computer instructions are used to cause a computer to execute the above-mentioned method for constructing a medical entity relationship diagram or method for quality control of a doctor's order.
  • a computer program product is also provided, including a computer program.
  • the computer program is executed by a processor, the above-mentioned method for constructing a medical entity relationship diagram or a method for quality control of a doctor's order is implemented.
  • FIG. 1 is a schematic flowchart of a method for constructing a medical entity relationship diagram provided by an embodiment of the present application
  • FIG. 2 is a schematic flowchart of another method for constructing a medical entity relationship diagram provided by an embodiment of the present application
  • FIG. 3 is a schematic flowchart of a method for quality control of a doctor's order provided by an embodiment of the present application
  • FIG. 4 is a schematic flowchart of another method for quality control of a doctor's order provided by an embodiment of the present application.
  • FIG. 5 is a schematic structural diagram of an apparatus for constructing a medical entity relationship diagram provided by an embodiment of the present application.
  • FIG. 6 is a schematic structural diagram of a device for quality control of medical orders provided by an embodiment of the present application.
  • FIG. 7 is a block diagram of an electronic device for implementing a method for constructing a medical entity relationship diagram and a method for quality control of a doctor's order provided by an embodiment of the present application.
  • the quality control of doctor's orders is different from clinical auxiliary diagnosis, intelligent consultation or drug recommendation. It does not give recommendations or suggestions based on the information of medical records.
  • the abnormality of the doctor's prescription man-made abnormality or abnormality caused by the doctor's omission
  • the quality control of the doctor's order is more inclined to abnormal detection.
  • anomaly detection it is necessary to first learn the relationship between the patient's condition and the drug, as well as the relationship between the drug and the drug. Therefore, the embodiments of the present application learn the relationship between medical entities by constructing a knowledge graph.
  • FIG. 1 is a schematic flowchart of a method for constructing a medical entity relationship diagram provided by an embodiment of the present application. This embodiment can be applied to a situation where a medical entity relationship diagram is constructed by adding a label node, and then the medical entity relationship diagram is used for quality control of medical orders. .
  • the method can be performed by an apparatus for constructing a medical entity relationship diagram, which is implemented in software and/or hardware and integrated on an electronic device.
  • S101 Perform entity recognition on medical-related data to obtain at least two medical entities.
  • Medical-related data can optionally be text data, including at least one of the following: drug instructions, medical books, medical guides, and users' medical orders and medical records; entities refer to things in the real world, such as people, place names, companies, telephones, Animals, etc., so medical entities refer to things in the medical field, such as drug names, phrases that describe symptoms of diseases.
  • a natural language processing technology may be used to perform entity recognition on medical-related data to obtain at least two medical entities.
  • a drug classification label is introduced, and each drug classification label is pre-associated with at least one drug name, so it is only necessary to determine the drug classification label associated with a medical entity (such as an entity describing disease symptoms), That is, the relationship between the drug name and the symptoms can be established through the drug classification label, that is, the relationship between at least two medical entities is established.
  • a medical entity such as an entity describing disease symptoms
  • constructing a medical entity relationship graph according to the medical entity and the drug classification label includes: using the medical entity and the drug classification label as nodes, and using the relationship between the at least two medical entities (for example, the symptom and the The relationship between drugs), the relationship between medical entities and drug classification labels (such as the relationship between symptoms and labels or the relationship between drugs and labels), and the relationship between at least two drug classification labels are used as edges to construct a relationship graph of medical entities. Then use the medical entity relationship diagram to carry out the quality control of the doctor's order.
  • a medical entity relationship diagram is established by introducing a drug classification label, and the relationship between at least two medical entities is accurately determined; further, the medical order quality control is performed based on the medical entity relationship diagram, which provides a guarantee for the medical order quality control. Improve the accuracy of the quality control of doctor's orders.
  • FIG. 2 is a schematic flowchart of another method for constructing a medical entity relationship diagram provided by an embodiment of the present application. This embodiment is described on the basis of the foregoing embodiment. Referring to Figure 2, the method for constructing a medical entity relationship diagram is as follows:
  • the medical-related data includes at least one of the following: drug instructions, medical books, medical guides, and the user's medical order data;
  • the medical natural language processing model is based on a bi-directional long-short-term memory network (Bi-directional Long-Short Term Memory, Bi-LSTM) , Attention mechanism (Attention) and a deep network model constructed by Condition Random Field (CRF).
  • the deep network model based on Bi-LSTM+CRF, compared with the traditional neural network (Deep Neural Networks (DNN), Recurrent Neural Network (RNN)) framework, on the one hand, considers the words in the text data.
  • the order relationship with words is more in line with the basic assumption of natural language processing (word order affects the expression of semantics).
  • the problems of gradient explosion and gradient vanishing make model training more stable.
  • the introduction of an attention mechanism can precisely determine the medical entities that need to be focused on.
  • the patient's main complaint description in the medical record data "I have a cough, a headache, a stomachache and a cold today.”
  • the medical natural language processing module first divides this sentence into words to obtain "I, today, cough, headache, stomachache, and a little cold", Then perform entity recognition, and the obtained medical entities include "cough, headache, stomach ache, cold”.
  • entity type identification it can be known that "cough, headache, stomach ache” is a symptom-based medical entity, and "cold” belongs to a disease-based medical entity.
  • the prediction result is the output result of the label prediction model, including the predicted drug classification label and probability.
  • the medical entity is used as the input of the label prediction model, and according to the output of the label prediction model, the probability value that the medical entity is associated with a drug classification label is obtained, and then the relationship between the medical entity and the drug classification label is determined according to the probability value. For example, if the probability value is greater than the preset threshold, it is considered that there is a relationship between the two; if the probability value is less than the preset threshold, it is considered that there is no relationship between the two.
  • the label prediction model is a knowledge-enhanced semantic representation model (Enhanced Representation from kNowledge IntEgration, ERNIE);
  • the sample data for training the label prediction model is the entity description data determined from the drug instructions and the drugs corresponding to the entity description data Category labels.
  • construct the sample data in the form of a 2-tuple for example, construct a 2-tuple of ⁇ entities, label>
  • entities is the entity description data related to the drug in a drug instruction manual, such as "used to relieve mild to moderate pain such as Arthralgia, muscle pain, neuralgia, headache, migraine, toothache, dysmenorrhea, also used for the entity in "fever caused by common cold or influenza”
  • label is the drug classification label corresponding to the entity description data, such as antibiotics, etc. .
  • the label prediction model is trained based on the entity and drug classification labels in the item description, so using the label prediction model for prediction can improve the efficiency and accuracy of determining the drug classification labels associated with medical entities.
  • entity recognition is performed based on a medical natural language processing model to improve the recognition accuracy; and the drug classification label associated with the entity is predicted by the label prediction model, which can improve the efficiency and accuracy of determining the drug classification label associated with the medical entity .
  • FIG. 3 is a schematic flowchart of a method for quality control of a doctor's order provided by an embodiment of the present application. This embodiment is described on the basis of the foregoing embodiment. Referring to Figure 3, the method for quality control of doctor's orders is as follows:
  • the first medical entity refers to a medical entity whose entity type is symptom
  • the second medical entity refers to a medical entity whose entity type is medicine.
  • obtaining the first medical entity and the second medical entity from the medical order medical record data to be detected including:
  • word segmentation is performed on the medical order and medical record data to be detected, and entity recognition is performed on the word segmentation results to obtain at least two medical entities.
  • entity recognition is performed on the word segmentation results to obtain at least two medical entities.
  • the identification process please refer to the description of the above embodiment; Identifying; wherein, the entity type includes at least one of the following: symptoms, diseases, examinations, tests, surgery, and medicines; and obtaining the first medical entity and the second medical entity according to the entity type identification result.
  • the efficiency of acquiring the first medical entity and the second medical entity can be guaranteed.
  • the entity relationship diagram if it is determined that there is an association relationship between the first medical entity and the second medical entity, for example, there is a connection path from the first medical entity to the second medical entity in the entity relationship diagram, then it is determined that there is an association relationship between the two , there is no need for quality control reminder; if there is no connection path from the first medical entity to the second medical entity in the entity relationship diagram, it is determined that the two are not related, and it is considered that the doctor's prescription issued by the doctor is in the patient's electronic medical record. In the information, there is no disease support. It may be that the doctor prescribes the wrong medicine due to negligence. On the other hand, it may be that the doctor deliberately prescribes the medicine to cheat the insurance, thus triggering the quality control reminder.
  • FIG. 4 is a schematic flowchart of another method for quality control of doctor’s orders provided in the embodiment of the present application. This embodiment is described on the basis of the above-mentioned embodiment. Referring to FIG. 4 , the method for quality control of doctor’s orders is as follows:
  • the first medical entity refers to a medical entity whose entity type is symptom
  • the second medical entity refers to a medical entity whose entity type is medicine.
  • the preset traversal algorithm can optionally be a breadth first search algorithm (BFS), or can be other traversal algorithms, such as a depth traversal algorithm, which is not limited here.
  • BFS breadth first search algorithm
  • the connected paths are determined according to the steps of S403-S404.
  • S403. Determine a target connected path with the shortest path from at least one connected path, and calculate an association score of the target connected path.
  • connection path The shorter the connection path, the greater the possibility that the first medical entity and the second medical entity are associated, wherein the length of the connection path can be determined by the number of drug classification label nodes passed by the first medical entity to the second medical entity Sure. Therefore, a target connected path with the shortest path can be determined from at least one connected path, and an association score of the target connected path can be calculated.
  • the process of calculating the correlation score of the target connection path includes: determining the first medical entity The entity is associated with the first probability value of the drug classification label, and the second medical entity is associated with the second probability value of the drug classification label; the first probability value and the second probability value are weighted to obtain an association score.
  • P reminder (P drug-tag +P disease-tag )/N ⁇ , where N represents the number of nodes.
  • an alarm threshold may be preset, and when the correlation score is lower than the alarm threshold, a quality control reminder is triggered.
  • the accuracy of the quality control can be ensured.
  • FIG. 5 is a schematic structural diagram of an apparatus for constructing a medical entity relationship diagram provided by an embodiment of the present application.
  • the apparatus includes: an entity identification module 501, configured to perform entity identification on medical-related data, and obtain at least two
  • the label prediction module 502 is configured to determine the drug classification label associated with the medical entity
  • the relationship diagram construction module 503 is configured to construct a medical entity relationship diagram according to the medical entity and the drug classification label.
  • the relationship diagram building module is set as:
  • the entity identification module is set to:
  • the medical natural language processing model Based on the medical natural language processing model, word segmentation is performed on medical-related data, and entity recognition is performed on the word segmentation results to obtain at least two medical entities; wherein, the medical-related data includes at least one of the following: drug instructions, medical books, medical guidelines and The user's doctor's order and medical record data; the medical natural language processing model is a deep network model constructed based on a bidirectional long-term and short-term memory network, an attention mechanism and a conditional random field.
  • the label prediction module is set to:
  • the pre-trained label prediction model is used to predict the labels of medical entities, and the drug classification labels associated with the medical entities are determined according to the prediction results.
  • the label prediction model is a knowledge-enhanced semantic representation model
  • the sample data for training the label prediction model is entity description data and entity description data determined from drug instructions, medical books and medical guides The corresponding drug classification label.
  • the apparatus for constructing a medical entity relationship diagram provided by the embodiment of the present application can execute the method for constructing a medical entity relationship diagram provided by any embodiment of the present application, and has corresponding functional modules and effects of the execution method.
  • the apparatus for constructing a medical entity relationship diagram provided by the embodiment of the present application can execute the method for constructing a medical entity relationship diagram provided by any embodiment of the present application, and has corresponding functional modules and effects of the execution method.
  • FIG. 6 is a schematic structural diagram of a device for quality control of medical orders provided by an embodiment of the present application.
  • the device includes: an entity acquisition module 601 configured to acquire a first medical entity from medical order medical record data to be detected and a second medical entity; wherein, the first medical entity refers to a medical entity whose entity type is a symptom, and the second medical entity refers to a medical entity whose entity type is a drug; the quality control module 602 is configured to use an entity relationship diagram to determine the first medical entity. Quality control information between a medical entity and a second medical entity; wherein, the entity relationship diagram is constructed by using the above method for constructing a medical entity relationship diagram.
  • the quality control module includes:
  • the traversal unit is set to adopt a preset traversal algorithm to determine from the entity relationship graph at least one connected path with the first medical entity as the starting node and the second medical entity as the terminating node; the screening calculation unit is set to start from at least one path The shortest target connection path among the connection paths is determined, and the correlation score of the target connection path is calculated; the quality control unit is set to determine the quality control information between the first medical entity and the second medical entity according to the correlation score.
  • the screening calculation unit includes:
  • the probability value determination subunit is set to determine the first probability value of the associated drug classification label of the first medical entity, and the second probability value of the associated drug classification label of the second medical entity; the calculation subunit is set to determine the first probability value and the second probability value of the associated drug classification label; The second probability value is weighted to obtain an association score.
  • the entity acquisition module includes:
  • the entity recognition unit is set to perform word segmentation processing on the medical order medical record data to be detected based on the medical natural language processing model, and performs entity recognition on the word segmentation results to obtain at least two medical entities;
  • the type recognition unit is set to the entity type of the medical entity. performing identification; wherein, the entity type includes at least one of the following: symptoms, diseases, examinations, tests, operations and medicines;
  • the acquiring unit is configured to acquire the first medical entity and the second medical entity according to the entity type identification result.
  • the device for quality control of a doctor's order provided by the embodiment of the present application can execute the method for quality control of a doctor's order provided by any embodiment of the present application, and has functional modules and effects corresponding to the execution method.
  • the description in any method embodiment of this application can be made to the description in any method embodiment of this application.
  • the present application further provides an electronic device, a readable storage medium, and a computer program product.
  • FIG. 7 is a block diagram of an electronic device 700 for implementing a method for constructing a medical entity relationship diagram and a method for quality control of a doctor's order provided by an embodiment of the present application.
  • Electronic device 700 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers.
  • Electronic device 700 may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices.
  • the components shown herein, their connections and relationships, and their functions are by way of example only, and are not intended to limit implementations of the application described and/or claimed herein.
  • the electronic device 700 includes a computing unit 701, which can be loaded into a random access memory (Random Access Memory) according to a computer program stored in a read-only memory (Read-Only Memory, ROM) 702 or from a storage unit 708, A computer program in RAM) 703 to perform various appropriate actions and processes.
  • ROM Read-Only Memory
  • RAM random access memory
  • various programs and data required for the operation of the electronic device 700 can also be stored.
  • the computing unit 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704.
  • An Input/Output (I/O) interface 705 is also connected to the bus 704 .
  • Various components in the electronic device 700 are connected to the I/O interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a magnetic disk, an optical disk etc.; and a communication unit 709, such as a network card, modem, wireless communication transceiver, and the like.
  • the communication unit 709 allows the electronic device 700 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks.
  • Computing unit 701 may be a variety of general-purpose and/or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU), various dedicated artificial intelligence (Artificial Intelligence, AI) computing chips, various operating A computational unit, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc., for the algorithm of the machine learning model.
  • the computing unit 701 executes the methods and processes described above, such as a method for quality control of a doctor's order or a method for constructing a medical entity relationship diagram.
  • a method of quality control or constructing a medical entity relationship diagram may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 708 .
  • part or all of the computer program may be loaded and/or installed on electronic device 700 via ROM 702 and/or communication unit 709 .
  • the computing unit 701 may be configured in any other suitable manner (eg, by means of firmware) to perform a method of quality control of a doctor's order or a method of constructing a medical entity relationship diagram.
  • FPGAs Field Programmable Gate Arrays
  • ASICs Application Specific Integrated Circuits
  • ASSP Application Specific Standard Parts
  • SoC System on Chip
  • CPLD Complex Programming Logic Device
  • These various embodiments may include implementation in one or more computer programs executable and/or interpretable on a programmable system including at least one programmable processor that
  • the processor which may be a special purpose or general-purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device an output device.
  • Program code for implementing the methods of the present application may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer or other programmable data processing apparatus, such that the program code, when executed by the processor or controller, performs the functions/functions specified in the flowcharts and/or block diagrams. Action is implemented.
  • the program code may execute entirely on the machine, partly on the machine, partly on the machine and partly on a remote machine as a stand-alone software package or entirely on the remote machine or server.
  • a machine-readable medium may be a tangible medium that may contain or store the program for use by or in connection with the instruction execution system, apparatus or device.
  • the machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium.
  • Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or devices, or any suitable combination of the foregoing.
  • machine-readable storage media examples include one or more wire-based electrical connections, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (Erasable Programmable Read-Only Memory, EPROM or flash memory), fiber optics, Compact Disc Read-Only Memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing .
  • RAM random access memory
  • ROM read only memory
  • EPROM erasable programmable read only memory
  • CD-ROM Compact Disc Read-Only Memory
  • optical storage devices magnetic storage devices, or any suitable combination of the foregoing .
  • the systems and techniques described herein may be implemented on a computer having a display device (eg, a Cathode Ray Tube (CRT) or a liquid crystal display) configured to display information to the user (Liquid Crystal Display, LCD monitor); and a keyboard and pointing device (eg, mouse or trackball) through which a user can provide input to the computer.
  • a display device eg, a Cathode Ray Tube (CRT) or a liquid crystal display
  • LCD monitor Liquid Crystal Display
  • keyboard and pointing device eg, mouse or trackball
  • Other kinds of devices may also be configured to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (eg, visual feedback, auditory feedback, or tactile feedback); and may be in any form (including acoustic input, voice input, or tactile input) to receive input from the user.
  • the systems and techniques described herein may be implemented on a computing system that includes back-end components (eg, as a data server), or a computing system that includes middleware components (eg, an application server), or a computing system that includes front-end components (eg, a user's computer having a graphical user interface or web browser through which a user may interact with implementations of the systems and techniques described herein), or including such backend components, middleware components, Or any combination of front-end components in a computing system.
  • the components of the system may be interconnected by any form or medium of digital data communication (eg, a communication network). Examples of communication networks include: Local Area Network (LAN), Wide Area Network (WAN), blockchain network, and the Internet.
  • a computer system can include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server arises by computer programs running on the respective computers and having a client-server relationship to each other.
  • the server can be a cloud server, also known as a cloud computing server or a cloud host. It is a host product in the cloud computing service system to solve the problems existing in traditional physical host and virtual private server (Virtual Private Server, VPS) services. The management is difficult and the business expansion is weak.
  • the server can also be a server of a distributed system, or a server combined with a blockchain.
  • Steps can be reordered, added, or removed using the various forms of flow shown above.
  • steps described in the present application may be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in the present application can be achieved, no limitation is imposed herein.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Chemical & Material Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Medicinal Chemistry (AREA)
  • Epidemiology (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Primary Health Care (AREA)
  • Public Health (AREA)
  • Artificial Intelligence (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

Provided are a method and apparatus for constructing a medical entity relationship diagram, a method and apparatus for medical order quality control, a device, and a medium. The method for constructing a medical entity relationship diagram comprises: performing entity recognition on medical related data to obtain at least two medical entities (S101); determining drug classification labels associated with the at least two medical entities (S102); and constructing a medical entity relationship diagram according to the at least two medical entities and the drug classification labels associated with the at least two medical entities (S103).

Description

构建医学实体关系图的方法及装置、医嘱质控的方法及装置、设备、介质Method and device for constructing medical entity relationship diagram, method and device, device and medium for quality control of doctor's orders
本申请要求在2021年01月29日提交中国专利局、申请号为202110129851.6的中国专利申请的优先权,该申请的全部内容通过引用结合在本申请中。This application claims the priority of a Chinese patent application with application number 202110129851.6 filed with the China Patent Office on January 29, 2021, the entire contents of which are incorporated herein by reference.
技术领域technical field
本申请涉及人工智能技术领域,例如涉及深度学习技术,例如涉及一种构建医学实体关系图的方法及装置、医嘱质控的方法及装置、设备、介质。The present application relates to the field of artificial intelligence technology, such as deep learning technology, such as a method and device for constructing a medical entity relationship graph, and a method and device, equipment, and medium for quality control of medical orders.
背景技术Background technique
在临床上,医生根据患者的不同病情制定不同的医嘱,然而有时医嘱中用药的不合理会导致医疗损害,进而引发医疗纠纷案件。因此如何对医生制定的医嘱进行精准的用药合理性检测成为亟待解决的问题。In clinical practice, doctors make different medical orders according to different conditions of patients. However, sometimes the unreasonable medication in the medical orders will lead to medical damage, which will lead to medical disputes. Therefore, how to accurately test the rationality of drug use in the medical orders made by doctors has become an urgent problem to be solved.
发明内容SUMMARY OF THE INVENTION
本申请提供了一种构建医学实体关系图的方法及装置、医嘱质控的方法及装置、设备、介质。The present application provides a method and device for constructing a medical entity relationship diagram, a method and device, equipment and medium for quality control of a doctor's order.
提供了一种构建医学实体关系图的方法,包括:Provides a method for building a medical entity relationship diagram, including:
对医学相关数据进行实体识别,得到至少两个医学实体;Perform entity recognition on medical-related data to obtain at least two medical entities;
确定至少两个医学实体所关联的药物分类标签;Identifying drug classification labels associated with at least two medical entities;
根据至少两个医学实体和至少两个医学实体所关联的药物分类标签,构建医学实体关系图。A medical entity relationship graph is constructed according to the at least two medical entities and the drug classification labels associated with the at least two medical entities.
还提供了一种医嘱质控的方法,包括:A method for quality control of doctor's orders is also provided, including:
从待检测的医嘱病历数据中获取第一医学实体和第二医学实体;其中,第一医学实体是指实体类型为症状的医学实体,第二医学实体是指实体类型为药品的医学实体;Obtain the first medical entity and the second medical entity from the medical order medical record data to be detected; wherein, the first medical entity refers to the medical entity whose entity type is symptom, and the second medical entity refers to the medical entity whose entity type is medicine;
采用实体关系图,确定第一医学实体和第二医学实体之间的质控信息;其中,实体关系图采用上述构建医学实体关系图的方法构建。Using an entity relationship diagram, the quality control information between the first medical entity and the second medical entity is determined; wherein, the entity relationship diagram is constructed by using the above method for constructing a medical entity relationship diagram.
还提供了一种构建医学实体关系图的装置,包括:Also provided is an apparatus for constructing a medical entity relationship diagram, including:
实体识别模块,设置为对医学相关数据进行实体识别,得到至少两个医学实体;The entity recognition module is set to perform entity recognition on medical-related data to obtain at least two medical entities;
标签预测模块,设置为确定至少两个医学实体所关联的药物分类标签;The label prediction module is configured to determine the drug classification labels associated with at least two medical entities;
关系图构建模块,设置为根据至少两个医学实体和至少两个医学实体所关联的药物分类标签,构建医学实体关系图。The relationship diagram building module is configured to construct a medical entity relationship diagram according to the at least two medical entities and the drug classification labels associated with the at least two medical entities.
还提供了一种医嘱质控的装置,包括:Also provided is a device for quality control of a doctor's order, comprising:
实体获取模块,设置为从待检测的医嘱病历数据中获取第一医学实体和第二医学实体;其中,第一医学实体是指实体类型为症状的医学实体,第二医学实体是指实体类型为药品的医学实体;The entity acquisition module is configured to acquire the first medical entity and the second medical entity from the medical order medical record data to be detected; wherein, the first medical entity refers to the medical entity whose entity type is symptom, and the second medical entity refers to the entity type is the medical entity of the drug;
质控模块,设置为采用实体关系图,确定第一医学实体和第二医学实体之间的质控信息;其中,实体关系图采用上述构建医学实体关系图的方法构建。The quality control module is configured to use an entity relationship diagram to determine the quality control information between the first medical entity and the second medical entity; wherein the entity relationship diagram is constructed by using the above method for constructing a medical entity relationship diagram.
还提供了一种电子设备,包括:Also provided is an electronic device comprising:
至少一个处理器;以及at least one processor; and
与至少一个处理器通信连接的存储器;其中,a memory communicatively coupled to the at least one processor; wherein,
存储器存储有可被至少一个处理器执行的指令,指令被至少一个处理器执行,以使至少一个处理器能够执行上述的构建医学实体关系图的方法或医嘱质控的方法。The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can execute the above-mentioned method for constructing a medical entity relationship diagram or a method for quality control of a doctor's order.
还提供了一种存储有计算机指令的非瞬时计算机可读存储介质,计算机指令用于使计算机执行上述的构建医学实体关系图的方法或医嘱质控的方法。A non-transitory computer-readable storage medium storing computer instructions is also provided, and the computer instructions are used to cause a computer to execute the above-mentioned method for constructing a medical entity relationship diagram or method for quality control of a doctor's order.
还提供了一种计算机程序产品,包括计算机程序,计算机程序在被处理器执行时实现上述的构建医学实体关系图的方法或医嘱质控的方法。A computer program product is also provided, including a computer program. When the computer program is executed by a processor, the above-mentioned method for constructing a medical entity relationship diagram or a method for quality control of a doctor's order is implemented.
附图说明Description of drawings
图1是本申请实施例提供的一种构建医学实体关系图的方法的流程示意图;1 is a schematic flowchart of a method for constructing a medical entity relationship diagram provided by an embodiment of the present application;
图2是本申请实施例提供的另一种构建医学实体关系图的方法的流程示意图;2 is a schematic flowchart of another method for constructing a medical entity relationship diagram provided by an embodiment of the present application;
图3是本申请实施例提供的一种医嘱质控的方法的流程示意图;3 is a schematic flowchart of a method for quality control of a doctor's order provided by an embodiment of the present application;
图4是本申请实施例提供的另一种医嘱质控的方法的流程示意图;4 is a schematic flowchart of another method for quality control of a doctor's order provided by an embodiment of the present application;
图5是本申请实施例提供的一种构建医学实体关系图的装置的结构示意图;5 is a schematic structural diagram of an apparatus for constructing a medical entity relationship diagram provided by an embodiment of the present application;
图6是本申请实施例提供的一种医嘱质控的装置的结构示意图;6 is a schematic structural diagram of a device for quality control of medical orders provided by an embodiment of the present application;
图7是本申请实施例提供的一种用来实现构建医学实体关系图的方法和医嘱质控的方法的电子设备的框图。FIG. 7 is a block diagram of an electronic device for implementing a method for constructing a medical entity relationship diagram and a method for quality control of a doctor's order provided by an embodiment of the present application.
具体实施方式Detailed ways
以下结合附图对本申请的示范性实施例做出说明,其中包括本申请实施例的多种细节以助于理解,应当将它们认为仅仅是示范性的。以下的描述中省略了对公知功能和结构的描述。Exemplary embodiments of the present application are described below with reference to the accompanying drawings, which include various details of the embodiments of the present application to facilitate understanding, and should be considered as exemplary only. Descriptions of well-known functions and constructions are omitted in the following description.
本申请实施例中,医嘱质控不同于临床辅助诊断、智能问诊或者用药推荐,它不是根据病历的信息给出推荐或者建议,相反的,医嘱质控是对医生的医嘱处方进行检测,发现医嘱处方的异常(人为的异常或者是医生疏漏产生的异常),也即医嘱质控更偏向于异常检测。而要进行异常检测,需要先学习患者病症和药物的关系,以及药物和药物的关系。因此本申请实施例通过构建知识图谱的方式学习医学实体之间的关系。In the embodiment of the present application, the quality control of doctor's orders is different from clinical auxiliary diagnosis, intelligent consultation or drug recommendation. It does not give recommendations or suggestions based on the information of medical records. The abnormality of the doctor's prescription (man-made abnormality or abnormality caused by the doctor's omission), that is, the quality control of the doctor's order is more inclined to abnormal detection. To perform anomaly detection, it is necessary to first learn the relationship between the patient's condition and the drug, as well as the relationship between the drug and the drug. Therefore, the embodiments of the present application learn the relationship between medical entities by constructing a knowledge graph.
图1是本申请实施例提供的一种构建医学实体关系图的方法的流程示意图,本实施例可适用于通过添加标签节点构建医学实体关系图,进而利用医学实体关系图进行医嘱质控的情况。该方法可由一种构建医学实体关系图的装置来执行,该装置采用软件和/或硬件的方式实现,并集成在电子设备上。1 is a schematic flowchart of a method for constructing a medical entity relationship diagram provided by an embodiment of the present application. This embodiment can be applied to a situation where a medical entity relationship diagram is constructed by adding a label node, and then the medical entity relationship diagram is used for quality control of medical orders. . The method can be performed by an apparatus for constructing a medical entity relationship diagram, which is implemented in software and/or hardware and integrated on an electronic device.
参见图1,构建医学实体关系图的方法如下:Referring to Figure 1, the method for constructing a medical entity relationship diagram is as follows:
S101、对医学相关数据进行实体识别,得到至少两个医学实体。S101. Perform entity recognition on medical-related data to obtain at least two medical entities.
医学相关数据可选的为文本数据,包括如下至少一项:药物说明书、医学书籍、医学指南和用户的医嘱病历数据;实体指的是现实世界中的事物,比如人、地名、公司、电话、动物等,因此医学实体是指医学领域的事物,例如药品名称、描述疾病症状的词组。在一种可选的实施方式中,可利用自然语言处理技术对医学相关数据进行实体识别,得到至少两个医学实体。Medical-related data can optionally be text data, including at least one of the following: drug instructions, medical books, medical guides, and users' medical orders and medical records; entities refer to things in the real world, such as people, place names, companies, telephones, Animals, etc., so medical entities refer to things in the medical field, such as drug names, phrases that describe symptoms of diseases. In an optional implementation manner, a natural language processing technology may be used to perform entity recognition on medical-related data to obtain at least two medical entities.
S102、确定医学实体所关联的药物分类标签。S102. Determine the drug classification label associated with the medical entity.
由于医嘱病历数据本身存在用药合理性问题,基于存在错误的数据学习病症和药物的关系,会导致后续预测的误差较大;而在医学书籍或者医学指南中对于病症和治疗用药的关系描述,更偏向于该病症应该使用一类药物,而不是直接描述药物名称,也即缺少病症和药物的关系。基于此,本申请实施例中,引入了药物分类标签,每个药物分类标签都预先关联至少一种药物名称,因此只需确定医学实体(例如描述疾病症状的实体)所关联的药物分类标签,即可通过药物分类标签建立药物名称与症状之间的关系,也即是建立至少两个医学实体之间的关系。Due to the rationality of drug use in the data of medical orders and medical records, learning the relationship between diseases and drugs based on erroneous data will lead to large errors in subsequent predictions. The preference for the condition should use a class of drugs, rather than a direct description of the drug name, ie the lack of a relationship between the condition and the drug. Based on this, in the embodiment of this application, a drug classification label is introduced, and each drug classification label is pre-associated with at least one drug name, so it is only necessary to determine the drug classification label associated with a medical entity (such as an entity describing disease symptoms), That is, the relationship between the drug name and the symptoms can be established through the drug classification label, that is, the relationship between at least two medical entities is established.
S103、根据医学实体和药物分类标签,构建医学实体关系图。S103 , constructing a medical entity relationship diagram according to the medical entity and the drug classification label.
在一种可选的实施方式中,根据医学实体和药物分类标签,构建医学实体 关系图,包括:将医学实体和药物分类标签作为节点,将至少两个医学实体之间的关系(例如症状与药物的关系)、医学实体与药物分类标签之间的关系(例如症状与标签的关系或药物与标签的关系)和至少两个药物分类标签之间的关系作为边,构建医学实体的关系图。进而利用医学实体关系图进行医嘱质控。In an optional embodiment, constructing a medical entity relationship graph according to the medical entity and the drug classification label includes: using the medical entity and the drug classification label as nodes, and using the relationship between the at least two medical entities (for example, the symptom and the The relationship between drugs), the relationship between medical entities and drug classification labels (such as the relationship between symptoms and labels or the relationship between drugs and labels), and the relationship between at least two drug classification labels are used as edges to construct a relationship graph of medical entities. Then use the medical entity relationship diagram to carry out the quality control of the doctor's order.
本申请实施例中,通过引入药物分类标签建立医学实体关系图,精准的确定了至少两个医学实体之间的关系;进而基于医学实体关系图进行医嘱质控,为医嘱质控提供了保证,提升了医嘱质控的准确性。In the embodiment of the present application, a medical entity relationship diagram is established by introducing a drug classification label, and the relationship between at least two medical entities is accurately determined; further, the medical order quality control is performed based on the medical entity relationship diagram, which provides a guarantee for the medical order quality control. Improve the accuracy of the quality control of doctor's orders.
图2是本申请实施例提供的另一种构建医学实体关系图的方法的流程示意图,本实施例是在上述实施例的基础上进行说明。参见图2,构建医学实体关系图的方法如下:FIG. 2 is a schematic flowchart of another method for constructing a medical entity relationship diagram provided by an embodiment of the present application. This embodiment is described on the basis of the foregoing embodiment. Referring to Figure 2, the method for constructing a medical entity relationship diagram is as follows:
S201、基于医学自然语言处理模型,对医学相关数据进行分词处理,以及对分词结果进行实体识别,得到至少两个医学实体。S201. Based on the medical natural language processing model, perform word segmentation processing on medical-related data, and perform entity recognition on the word segmentation result to obtain at least two medical entities.
医学相关数据包括如下至少一项:药物说明书、医学书籍、医学指南和用户的医嘱病历数据;医学自然语言处理模型是基于双向长短期记忆网络(Bi-directional Long-Short Term Memory,Bi-LSTM)、注意力机制(Attention)和条件随机场(Condition Random Field,CRF)构建的深层网络模型。基于Bi-LSTM+CRF的深层网络模型,相比传统的神经网络(深度神经网络(Deep Neural Networks,DNN),循环神经网络(Recurrent Neural Network RNN))的框架,一方面考虑到了文本数据中词与词之间的顺序关系,更加符合自然语言处理的基本假设(语序影响语义的表达),另一方面,基于长短记忆单元(LSTM)的方法有效地解决了传统循环神经网络(RNN)存在的梯度爆炸(gradient explosion)和梯度弥散(gradient vanishing)的问题,使得模型训练更加稳定。同时,引入注意力机制可精确地确定需要重点关注的医学实体。The medical-related data includes at least one of the following: drug instructions, medical books, medical guides, and the user's medical order data; the medical natural language processing model is based on a bi-directional long-short-term memory network (Bi-directional Long-Short Term Memory, Bi-LSTM) , Attention mechanism (Attention) and a deep network model constructed by Condition Random Field (CRF). The deep network model based on Bi-LSTM+CRF, compared with the traditional neural network (Deep Neural Networks (DNN), Recurrent Neural Network (RNN)) framework, on the one hand, considers the words in the text data. The order relationship with words is more in line with the basic assumption of natural language processing (word order affects the expression of semantics). The problems of gradient explosion and gradient vanishing make model training more stable. At the same time, the introduction of an attention mechanism can precisely determine the medical entities that need to be focused on.
示例性的,病历数据中患者主诉描述:“我今天咳嗽头疼肚子疼有点感冒”,医学自然语言处理模块先将这句话分词得到“我、今天、咳嗽、头疼、肚子疼、有点感冒”,然后进行实体识别,得到的医学实体包括“咳嗽、头疼、肚子疼、感冒”。通过实体类型识别可知,“咳嗽、头疼、肚子疼”是症状类医学实体,“感冒”属于疾病类医学实体。Exemplarily, the patient's main complaint description in the medical record data: "I have a cough, a headache, a stomachache and a cold today." The medical natural language processing module first divides this sentence into words to obtain "I, today, cough, headache, stomachache, and a little cold", Then perform entity recognition, and the obtained medical entities include "cough, headache, stomach ache, cold". Through entity type identification, it can be known that "cough, headache, stomach ache" is a symptom-based medical entity, and "cold" belongs to a disease-based medical entity.
S202、利用预先训练的标签预测模型对医学实体进行标签预测,根据预测结果确定医学实体所关联的药物分类标签。S202. Use a pre-trained label prediction model to perform label prediction on the medical entity, and determine the drug classification label associated with the medical entity according to the prediction result.
预测结果即是标签预测模型的输出结果,包括预测出的药物分类标签和概率。可选的,将医学实体作为标签预测模型的输入,根据标签预测模型的输出,得到医学实体关联一药物分类标签的概率值,进而根据概率值确定医学实体与 药物分类标签的关系。例如概率值大于预设阈值,则认为两者存在关联关系;若概率值小于预设阈值,则认为两者之间不存在关联关系。The prediction result is the output result of the label prediction model, including the predicted drug classification label and probability. Optionally, the medical entity is used as the input of the label prediction model, and according to the output of the label prediction model, the probability value that the medical entity is associated with a drug classification label is obtained, and then the relationship between the medical entity and the drug classification label is determined according to the probability value. For example, if the probability value is greater than the preset threshold, it is considered that there is a relationship between the two; if the probability value is less than the preset threshold, it is considered that there is no relationship between the two.
本申请实施例中,标签预测模型为知识增强的语义表示模型(Enhanced Representation from kNowledge IntEgration,ERNIE);训练标签预测模型的样本数据为从药物说明书中确定的实体描述数据和实体描述数据对应的药物分类标签。示例性的,构建二元组形式的样本数据,例如构建<entities,label>的二元组,entities是一药品说明书中该药物相关的实体描述数据,比如“用于缓解轻至中度疼痛如关节痛、肌肉痛、神经痛、头痛、偏头痛、牙痛、痛经,也用于普通感冒或流行性感冒引起的发热”中的实体,label是该实体描述数据对应的药物分类标签,比如抗生素等。In the embodiment of this application, the label prediction model is a knowledge-enhanced semantic representation model (Enhanced Representation from kNowledge IntEgration, ERNIE); the sample data for training the label prediction model is the entity description data determined from the drug instructions and the drugs corresponding to the entity description data Category labels. Exemplarily, construct the sample data in the form of a 2-tuple, for example, construct a 2-tuple of <entities, label>, entities is the entity description data related to the drug in a drug instruction manual, such as "used to relieve mild to moderate pain such as Arthralgia, muscle pain, neuralgia, headache, migraine, toothache, dysmenorrhea, also used for the entity in "fever caused by common cold or influenza", label is the drug classification label corresponding to the entity description data, such as antibiotics, etc. .
标签预测模型是根据物品说明书中的实体和药物分类标签训练的,因此利用标签预测模型进行预测可提升确定医学实体所关联的药物分类标签的效率和准确性。The label prediction model is trained based on the entity and drug classification labels in the item description, so using the label prediction model for prediction can improve the efficiency and accuracy of determining the drug classification labels associated with medical entities.
S203、根据医学实体和药物分类标签,构建医学实体关系图。S203 , constructing a medical entity relationship diagram according to the medical entity and the drug classification label.
本申请实施例中,基于医学自然语言处理模型进行实体识别,提升识别准确性;而且通过标签预测模型预测实体关联的药物分类标签,可提升确定医学实体所关联的药物分类标签的效率和准确性。In the embodiment of the present application, entity recognition is performed based on a medical natural language processing model to improve the recognition accuracy; and the drug classification label associated with the entity is predicted by the label prediction model, which can improve the efficiency and accuracy of determining the drug classification label associated with the medical entity .
图3是本申请实施例提供的一种医嘱质控的方法的流程示意图,本实施例是在上述实施例的基础上进行说明。参见图3,医嘱质控的方法如下:FIG. 3 is a schematic flowchart of a method for quality control of a doctor's order provided by an embodiment of the present application. This embodiment is described on the basis of the foregoing embodiment. Referring to Figure 3, the method for quality control of doctor's orders is as follows:
S301、从待检测的医嘱病历数据中获取第一医学实体和第二医学实体。S301. Acquire a first medical entity and a second medical entity from the medical order medical record data to be detected.
第一医学实体是指实体类型为症状的医学实体,第二医学实体是指实体类型为药品的医学实体。The first medical entity refers to a medical entity whose entity type is symptom, and the second medical entity refers to a medical entity whose entity type is medicine.
在一种可选的实施方式中,从待检测的医嘱病历数据中获取第一医学实体和第二医学实体,包括:In an optional implementation manner, obtaining the first medical entity and the second medical entity from the medical order medical record data to be detected, including:
基于医学自然语言处理模型,对待检测的医嘱病历数据进行分词处理,以及对分词结果进行实体识别,得到至少两个医学实体,识别过程可参见上述实施例的描述;进而对医学实体的实体类型进行识别;其中,实体类型包括如下至少一项:症状、疾病、检查、检验、手术和药品;根据实体类型识别结果,获取第一医学实体和第二医学实体。Based on the medical natural language processing model, word segmentation is performed on the medical order and medical record data to be detected, and entity recognition is performed on the word segmentation results to obtain at least two medical entities. For the identification process, please refer to the description of the above embodiment; Identifying; wherein, the entity type includes at least one of the following: symptoms, diseases, examinations, tests, surgery, and medicines; and obtaining the first medical entity and the second medical entity according to the entity type identification result.
通过对至少两个医学实体进行实例类型识别,可保证获取第一医学实体和第二医学实体的效率。By performing instance type identification on at least two medical entities, the efficiency of acquiring the first medical entity and the second medical entity can be guaranteed.
S302、采用实体关系图,确定第一医学实体和第二医学实体之间的质控信 息。S302. Using an entity relationship diagram, determine the quality control information between the first medical entity and the second medical entity.
实体关系图的构建可参见上述实施例,在此不再赘述。根据实体关系图,如果确定第一医学实体和第二医学实体之间存在关联关系,例如在实体关系图中存在由第一医学实体到第二医学实体的连通路径,则确定两者存在关联关系,不需要进行质控提醒;若实体关系图中不存在由第一医学实体到第二医学实体的连通路径,则确定两者没有关联关系,则认为医生开具的医嘱处方,在患者的电子病历信息中,没有病症支撑,可能是医生因为疏忽开错了药,另一方面可能是医生故意开药进行骗保的行为,因此触发质控提醒。For the construction of the entity relationship diagram, reference may be made to the foregoing embodiments, and details are not described herein again. According to the entity relationship diagram, if it is determined that there is an association relationship between the first medical entity and the second medical entity, for example, there is a connection path from the first medical entity to the second medical entity in the entity relationship diagram, then it is determined that there is an association relationship between the two , there is no need for quality control reminder; if there is no connection path from the first medical entity to the second medical entity in the entity relationship diagram, it is determined that the two are not related, and it is considered that the doctor's prescription issued by the doctor is in the patient's electronic medical record. In the information, there is no disease support. It may be that the doctor prescribes the wrong medicine due to negligence. On the other hand, it may be that the doctor deliberately prescribes the medicine to cheat the insurance, thus triggering the quality control reminder.
本申请实施例中,由于实体关系图中至少两个实体通过药物分类标签建立联系,因此只需判断两个实体间是否存在连通路径即可确定两者之间是否关联,进而确定是否触发质控提醒,由此可提升医嘱质控的效率和准确性。In the embodiment of the present application, since at least two entities in the entity relationship diagram are connected through drug classification labels, it is only necessary to determine whether there is a connection path between the two entities to determine whether the two entities are related, and then determine whether to trigger the quality control Reminder, which can improve the efficiency and accuracy of the quality control of doctor's orders.
图4是本申请实施例提供的另一种医嘱质控的方法的流程示意图,本实施例是在上述实施例的基础上进行说明,参见图4,医嘱质控的方法如下:FIG. 4 is a schematic flowchart of another method for quality control of doctor’s orders provided in the embodiment of the present application. This embodiment is described on the basis of the above-mentioned embodiment. Referring to FIG. 4 , the method for quality control of doctor’s orders is as follows:
S401、从待检测的医嘱病历数据中获取第一医学实体和第二医学实体。S401. Acquire a first medical entity and a second medical entity from the medical order medical record data to be detected.
第一医学实体是指实体类型为症状的医学实体,第二医学实体是指实体类型为药品的医学实体。The first medical entity refers to a medical entity whose entity type is symptom, and the second medical entity refers to a medical entity whose entity type is medicine.
S402、采用预设遍历算法,从实体关系图中确定以第一医学实体为起始节点,以第二医学实体为终止节点的至少一条连通路径。S402. Using a preset traversal algorithm, determine from the entity relationship graph at least one connection path with the first medical entity as the starting node and the second medical entity as the ending node.
预设遍历算法可选的为广度优先遍历算法(Breadth First Search,BFS),也可以为其他遍历算法,例如深度遍历算法,在此不做限定。本申请实施例中,若通过广度优先遍历算法确定存在多条以第一医学实体为起始节点,以第二医学实体为终止节点的连通路径,为了提升医嘱质控的准确性,需要对多条连通路径按照S403-S404的步骤做判断。The preset traversal algorithm can optionally be a breadth first search algorithm (BFS), or can be other traversal algorithms, such as a depth traversal algorithm, which is not limited here. In the embodiment of the present application, if it is determined by the breadth-first traversal algorithm that there are multiple connection paths with the first medical entity as the starting node and the second medical entity as the ending node, in order to improve the accuracy of the quality control of the doctor's order, it is necessary to The connected paths are determined according to the steps of S403-S404.
S403、从至少一条连通路径中确定路径最短的目标连通路径,并计算目标连通路径的关联得分。S403. Determine a target connected path with the shortest path from at least one connected path, and calculate an association score of the target connected path.
连通路径越短,则第一医学实体和第二医学实体相关联的可能性越大,其中,连通路径的长短可通过第一医学实体到第二医学实体所经过的药物分类标签节点的个数确定。因此可从至少一条连通路径中确定路径最短的目标连通路径,并计算目标连通路径的关联得分。The shorter the connection path, the greater the possibility that the first medical entity and the second medical entity are associated, wherein the length of the connection path can be determined by the number of drug classification label nodes passed by the first medical entity to the second medical entity Sure. Therefore, a target connected path with the shortest path can be determined from at least one connected path, and an association score of the target connected path can be calculated.
在一种可选的实施方式中,若目标连通路径是由第一医学实体、第二医学实体和至少一个药物分类标签组成的,则计算目标连通路径的关联得分的过程包括:确定第一医学实体关联药物分类标签的第一概率值,以及第二医学实体关联药物分类标签的第二概率值;对第一概率值和第二概率值进行加权处理, 得到关联得分。通过将第一医学实体和第二医学实体关联的标签的概率值进行加权处理,可以提升计算关联得分的效率。In an optional embodiment, if the target connection path is composed of a first medical entity, a second medical entity and at least one drug classification label, the process of calculating the correlation score of the target connection path includes: determining the first medical entity The entity is associated with the first probability value of the drug classification label, and the second medical entity is associated with the second probability value of the drug classification label; the first probability value and the second probability value are weighted to obtain an association score. By weighting the probability values of the tags associated with the first medical entity and the second medical entity, the efficiency of calculating the association score can be improved.
示例性的,假设连通路径由V drug(药物节点)、V tag(标签节点)和V disease(病症节点)三个节点组成,其中,P drug-tag代表节点V drug和V tag之间的边权重,边权重P drug-tag可选的是节点V drug关联节点V tag的概率,同样的,P disease-tag可以是病症节点关联节点V tag的概率,α(α>=1)代表路径长度的衰减因子(α>1时,路径越长,最终得分衰减的越快),最终的结果P reminder基于路径加权计算。这里举例的是3个节点的路径,构成两条边,实际应用可以考虑更多的节点构成的边。计算公式如下:P reminder=(P drug-tag+P disease-tag)/N α,其中,N表示节点的个数。 Exemplarily, it is assumed that the connected path consists of three nodes: V drug (drug node), V tag (tag node) and V disease (disease node), wherein P drug-tag represents the edge between nodes V drug and V tag Weight, edge weight P drug-tag can optionally be the probability of the node V drug associated with the node V tag , similarly, P disease-tag can be the probability of the disease node associated with the node V tag , α (α>=1) represents the path length (when α>1, the longer the path, the faster the final score decays), the final result P reminder is calculated based on the path weight. Here is an example of a path with three nodes, which constitutes two edges. In practical applications, edges composed of more nodes can be considered. The calculation formula is as follows: P reminder =(P drug-tag +P disease-tag )/N α , where N represents the number of nodes.
S404、根据关联得分,确定第一医学实体和第二医学实体之间的质控信息。S404. Determine the quality control information between the first medical entity and the second medical entity according to the correlation score.
本申请实施例中,为了保证质控的准确性,可预先设定一个报警阈值,在关联得分低于报警阈值时,则触发质控提醒。In the embodiment of the present application, in order to ensure the accuracy of quality control, an alarm threshold may be preset, and when the correlation score is lower than the alarm threshold, a quality control reminder is triggered.
本申请实施例中,通过计算连通路径的关联得分,评价构成连通路径的第一医学实体和第二医学实体的关联关系,可保证质控的准确性。In the embodiment of the present application, by calculating the association score of the connection path and evaluating the association relationship between the first medical entity and the second medical entity constituting the connection path, the accuracy of the quality control can be ensured.
图5是本申请实施例提供的一种构建医学实体关系图的装置的结构示意图,如图5所示,该装置包括:实体识别模块501,设置为对医学相关数据进行实体识别,得到至少两个医学实体;标签预测模块502,设置为确定医学实体所关联的药物分类标签;关系图构建模块503,设置为根据医学实体和药物分类标签,构建医学实体关系图。FIG. 5 is a schematic structural diagram of an apparatus for constructing a medical entity relationship diagram provided by an embodiment of the present application. As shown in FIG. 5 , the apparatus includes: an entity identification module 501, configured to perform entity identification on medical-related data, and obtain at least two The label prediction module 502 is configured to determine the drug classification label associated with the medical entity; the relationship diagram construction module 503 is configured to construct a medical entity relationship diagram according to the medical entity and the drug classification label.
在上述实施例的基础上,可选的,关系图构建模块设置为:On the basis of the above embodiment, optionally, the relationship diagram building module is set as:
将医学实体和药物分类标签作为节点,将至少两个医学实体之间的关系、医学实体与药物分类标签之间的关系和至少两个药物分类标签之间的关系作为边,构建医学实体的关系图。Use medical entities and drug classification labels as nodes, and use the relationship between at least two medical entities, the relationship between medical entities and drug classification labels, and the relationship between at least two drug classification labels as edges to construct the relationship between medical entities picture.
在上述实施例的基础上,可选的,实体识别模块设置为:On the basis of the above embodiment, optionally, the entity identification module is set to:
基于医学自然语言处理模型,对医学相关数据进行分词处理,以及对分词结果进行实体识别,得到至少两个医学实体;其中,医学相关数据包括如下至少一项:药物说明书、医学书籍、医学指南和用户的医嘱病历数据;医学自然语言处理模型是基于双向长短期记忆网络、注意力机制和条件随机场构建的深层网络模型。Based on the medical natural language processing model, word segmentation is performed on medical-related data, and entity recognition is performed on the word segmentation results to obtain at least two medical entities; wherein, the medical-related data includes at least one of the following: drug instructions, medical books, medical guidelines and The user's doctor's order and medical record data; the medical natural language processing model is a deep network model constructed based on a bidirectional long-term and short-term memory network, an attention mechanism and a conditional random field.
在上述实施例的基础上,可选的,标签预测模块设置为:On the basis of the above embodiment, optionally, the label prediction module is set to:
利用预先训练的标签预测模型对医学实体进行标签预测,根据预测结果确 定医学实体所关联的药物分类标签。The pre-trained label prediction model is used to predict the labels of medical entities, and the drug classification labels associated with the medical entities are determined according to the prediction results.
在上述实施例的基础上,可选的,标签预测模型为知识增强的语义表示模型;训练标签预测模型的样本数据为从药物说明书、医学书籍和医学指南中确定的实体描述数据和实体描述数据对应的药物分类标签。On the basis of the above embodiment, optionally, the label prediction model is a knowledge-enhanced semantic representation model; the sample data for training the label prediction model is entity description data and entity description data determined from drug instructions, medical books and medical guides The corresponding drug classification label.
本申请实施例提供的构建医学实体关系图的装置可执行本申请任意实施例提供的构建医学实体关系图的方法,具备执行方法相应的功能模块和效果。本实施例中未详尽描述的内容可以参考本申请任意方法实施例中的描述。The apparatus for constructing a medical entity relationship diagram provided by the embodiment of the present application can execute the method for constructing a medical entity relationship diagram provided by any embodiment of the present application, and has corresponding functional modules and effects of the execution method. For the content not described in detail in this embodiment, reference may be made to the description in any method embodiment of this application.
图6是本申请实施例提供的一种医嘱质控的装置的结构示意图,如图6所示,该装置包括:实体获取模块601,设置为从待检测的医嘱病历数据中获取第一医学实体和第二医学实体;其中,第一医学实体是指实体类型为症状的医学实体,第二医学实体是指实体类型为药品的医学实体;质控模块602,设置为采用实体关系图,确定第一医学实体和第二医学实体之间的质控信息;其中,实体关系图采用上述构建医学实体关系图的方法构建。FIG. 6 is a schematic structural diagram of a device for quality control of medical orders provided by an embodiment of the present application. As shown in FIG. 6 , the device includes: an entity acquisition module 601 configured to acquire a first medical entity from medical order medical record data to be detected and a second medical entity; wherein, the first medical entity refers to a medical entity whose entity type is a symptom, and the second medical entity refers to a medical entity whose entity type is a drug; the quality control module 602 is configured to use an entity relationship diagram to determine the first medical entity. Quality control information between a medical entity and a second medical entity; wherein, the entity relationship diagram is constructed by using the above method for constructing a medical entity relationship diagram.
在上述实施例的基础上,可选的,质控模块包括:On the basis of the above embodiment, optionally, the quality control module includes:
遍历单元,设置为采用预设遍历算法,从实体关系图中确定以第一医学实体为起始节点,以第二医学实体为终止节点的至少一条连通路径;筛选计算单元,设置为从至少一条连通路径中确定路径最短的目标连通路径,并计算目标连通路径的关联得分;质控单元,设置为根据关联得分,确定第一医学实体和第二医学实体之间的质控信息。The traversal unit is set to adopt a preset traversal algorithm to determine from the entity relationship graph at least one connected path with the first medical entity as the starting node and the second medical entity as the terminating node; the screening calculation unit is set to start from at least one path The shortest target connection path among the connection paths is determined, and the correlation score of the target connection path is calculated; the quality control unit is set to determine the quality control information between the first medical entity and the second medical entity according to the correlation score.
在上述实施例的基础上,可选的,若目标连通路径是由第一医学实体、第二医学实体和至少一个药物分类标签组成的,则筛选计算单元包括:On the basis of the above embodiment, optionally, if the target connection path is composed of a first medical entity, a second medical entity and at least one drug classification label, the screening calculation unit includes:
概率值确定子单元,设置为确定第一医学实体关联药物分类标签的第一概率值,以及第二医学实体关联药物分类标签的第二概率值;计算子单元,设置为对第一概率值和第二概率值进行加权处理,得到关联得分。The probability value determination subunit is set to determine the first probability value of the associated drug classification label of the first medical entity, and the second probability value of the associated drug classification label of the second medical entity; the calculation subunit is set to determine the first probability value and the second probability value of the associated drug classification label; The second probability value is weighted to obtain an association score.
在上述实施例的基础上,可选的,实体获取模块包括:On the basis of the foregoing embodiment, optionally, the entity acquisition module includes:
实体识别单元,设置为基于医学自然语言处理模型,对待检测的医嘱病历数据进行分词处理,以及对分词结果进行实体识别,得到至少两个医学实体;类型识别单元,设置为对医学实体的实体类型进行识别;其中,实体类型包括如下至少一项:症状、疾病、检查、检验、手术和药品;获取单元,设置为根据实体类型识别结果,获取第一医学实体和第二医学实体。The entity recognition unit is set to perform word segmentation processing on the medical order medical record data to be detected based on the medical natural language processing model, and performs entity recognition on the word segmentation results to obtain at least two medical entities; the type recognition unit is set to the entity type of the medical entity. performing identification; wherein, the entity type includes at least one of the following: symptoms, diseases, examinations, tests, operations and medicines; the acquiring unit is configured to acquire the first medical entity and the second medical entity according to the entity type identification result.
本申请实施例提供的医嘱质控的装置可执行本申请任意实施例提供的医嘱质控的方法,具备执行方法相应的功能模块和效果。本实施例中未详尽描述的内容可以参考本申请任意方法实施例中的描述。The device for quality control of a doctor's order provided by the embodiment of the present application can execute the method for quality control of a doctor's order provided by any embodiment of the present application, and has functional modules and effects corresponding to the execution method. For the content not described in detail in this embodiment, reference may be made to the description in any method embodiment of this application.
本申请的技术方案中,所涉及的用户个人信息的获取,存储和应用等,均符合相关法律法规的规定,且不违背公序良俗。In the technical solution of this application, the acquisition, storage and application of the user's personal information involved are in compliance with the relevant laws and regulations, and do not violate public order and good customs.
根据本申请的实施例,本申请还提供了一种电子设备、一种可读存储介质和一种计算机程序产品。According to the embodiments of the present application, the present application further provides an electronic device, a readable storage medium, and a computer program product.
图7是本申请实施例提供的一种用来实现构建医学实体关系图的方法和医嘱质控的方法的电子设备700的框图。电子设备700旨在表示多种形式的数字计算机,诸如,膝上型计算机、台式计算机、工作台、个人数字助理、服务器、刀片式服务器、大型计算机、和其它适合的计算机。电子设备700还可以表示多种形式的移动装置,诸如,个人数字处理、蜂窝电话、智能电话、可穿戴设备和其它类似的计算装置。本文所示的部件、它们的连接和关系、以及它们的功能仅仅作为示例,并且不意在限制本文中描述的和/或者要求的本申请的实现。FIG. 7 is a block diagram of an electronic device 700 for implementing a method for constructing a medical entity relationship diagram and a method for quality control of a doctor's order provided by an embodiment of the present application. Electronic device 700 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 700 may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are by way of example only, and are not intended to limit implementations of the application described and/or claimed herein.
如图7所示,电子设备700包括计算单元701,其可以根据存储在只读存储器(Read-Only Memory,ROM)702中的计算机程序或者从存储单元708加载到随机访问存储器(Random Access Memory,RAM)703中的计算机程序,来执行多种适当的动作和处理。在RAM 703中,还可存储电子设备700操作所需的多种程序和数据。计算单元701、ROM 702以及RAM 703通过总线704彼此相连。输入/输出(Input/Output,I/O)接口705也连接至总线704。As shown in FIG. 7 , the electronic device 700 includes a computing unit 701, which can be loaded into a random access memory (Random Access Memory) according to a computer program stored in a read-only memory (Read-Only Memory, ROM) 702 or from a storage unit 708, A computer program in RAM) 703 to perform various appropriate actions and processes. In the RAM 703, various programs and data required for the operation of the electronic device 700 can also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704. An Input/Output (I/O) interface 705 is also connected to the bus 704 .
电子设备700中的多个部件连接至I/O接口705,包括:输入单元706,例如键盘、鼠标等;输出单元707,例如多种类型的显示器、扬声器等;存储单元708,例如磁盘、光盘等;以及通信单元709,例如网卡、调制解调器、无线通信收发机等。通信单元709允许电子设备700通过诸如因特网的计算机网络和/或多种电信网络与其他设备交换信息/数据。Various components in the electronic device 700 are connected to the I/O interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a magnetic disk, an optical disk etc.; and a communication unit 709, such as a network card, modem, wireless communication transceiver, and the like. The communication unit 709 allows the electronic device 700 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks.
计算单元701可以是多种具有处理和计算能力的通用和/或专用处理组件。计算单元701的一些示例包括但不限于中央处理单元(Central Processing Unit,CPU)、图形处理单元(Graphics Processing Unit,GPU)、多种专用的人工智能(Artificial Intelligence,AI)计算芯片、多种运行机器学习模型算法的计算单元、数字信号处理器(Digital Signal Processor,DSP)、以及任何适当的处理器、控制器、微控制器等。计算单元701执行上文所描述的方法和处理,例如医嘱质控的方法或构建医学实体关系图的方法。例如,在一些实施例中,医嘱质控的或构建医学实体关系图的方法可被实现为计算机软件程序,其被有形地包含于机器可读介质,例如存储单元708。在一些实施例中,计算机程序的部分或者全部可以经由ROM 702和/或通信单元709而被载入和/或安装到电子设备700上。当计算机程序加载到RAM 703并由计算单元701执行时,可以执行上文描述的医嘱质控的方法或构建医学实体关系图的方法的一个或多个步骤。备选地, 在其他实施例中,计算单元701可以通过其他任何适当的方式(例如,借助于固件)而被配置为执行医嘱质控的方法或构建医学实体关系图的方法。 Computing unit 701 may be a variety of general-purpose and/or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU), various dedicated artificial intelligence (Artificial Intelligence, AI) computing chips, various operating A computational unit, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc., for the algorithm of the machine learning model. The computing unit 701 executes the methods and processes described above, such as a method for quality control of a doctor's order or a method for constructing a medical entity relationship diagram. For example, in some embodiments, a method of quality control or constructing a medical entity relationship diagram may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 708 . In some embodiments, part or all of the computer program may be loaded and/or installed on electronic device 700 via ROM 702 and/or communication unit 709 . When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the above-described method for quality control of a doctor's order or a method for constructing a medical entity relationship diagram can be performed. Alternatively, in other embodiments, the computing unit 701 may be configured in any other suitable manner (eg, by means of firmware) to perform a method of quality control of a doctor's order or a method of constructing a medical entity relationship diagram.
本文中以上描述的系统和技术的多种实施方式可以在数字电子电路系统、集成电路系统、场可编程门阵列(Field Programmable Gate Array,FPGA)、专用集成电路(Application Specific Integrated Circuit,ASIC)、专用标准产品(Application Specific Standard Parts,ASSP)、芯片上系统的系统(System on Chip,SoC)、复杂可编程逻辑设备(Complex Programming Logic Device,CPLD)、计算机硬件、固件、软件、和/或它们的组合中实现。这些多种实施方式可以包括:实施在一个或者多个计算机程序中,该一个或者多个计算机程序可在包括至少一个可编程处理器的可编程系统上执行和/或解释,该可编程处理器可以是专用或者通用可编程处理器,可以从存储系统、至少一个输入装置、和至少一个输出装置接收数据和指令,并且将数据和指令传输至该存储系统、该至少一个输入装置、和该至少一个输出装置。Various implementations of the systems and techniques described herein above may be implemented in digital electronic circuitry, integrated circuit systems, Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Parts (ASSP), System on Chip (SoC), Complex Programming Logic Device (CPLD), computer hardware, firmware, software, and/or them implemented in a combination. These various embodiments may include implementation in one or more computer programs executable and/or interpretable on a programmable system including at least one programmable processor that The processor, which may be a special purpose or general-purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device an output device.
用于实施本申请的方法的程序代码可以采用一个或多个编程语言的任何组合来编写。这些程序代码可以提供给通用计算机、专用计算机或其他可编程数据处理装置的处理器或控制器,使得程序代码当由处理器或控制器执行时使流程图和/或框图中所规定的功能/操作被实施。程序代码可以完全在机器上执行、部分地在机器上执行,作为独立软件包部分地在机器上执行且部分地在远程机器上执行或完全在远程机器或服务器上执行。Program code for implementing the methods of the present application may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer or other programmable data processing apparatus, such that the program code, when executed by the processor or controller, performs the functions/functions specified in the flowcharts and/or block diagrams. Action is implemented. The program code may execute entirely on the machine, partly on the machine, partly on the machine and partly on a remote machine as a stand-alone software package or entirely on the remote machine or server.
在本申请的上下文中,机器可读介质可以是有形的介质,其可以包含或存储以供指令执行系统、装置或设备使用或与指令执行系统、装置或设备结合地使用的程序。机器可读介质可以是机器可读信号介质或机器可读储存介质。机器可读介质可以包括但不限于电子的、磁性的、光学的、电磁的、红外的、或半导体系统、装置或设备,或者上述内容的任何合适组合。机器可读存储介质的示例会包括基于一个或多个线的电气连接、便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦除可编程只读存储器(Erasable Programmable Read-Only Memory,EPROM或快闪存储器)、光纤、便捷式紧凑盘只读存储器(Compact Disc Read-Only Memory,CD-ROM)、光学储存设备、磁储存设备、或上述内容的任何合适组合。In the context of this application, a machine-readable medium may be a tangible medium that may contain or store the program for use by or in connection with the instruction execution system, apparatus or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or devices, or any suitable combination of the foregoing. Examples of machine-readable storage media would include one or more wire-based electrical connections, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (Erasable Programmable Read-Only Memory, EPROM or flash memory), fiber optics, Compact Disc Read-Only Memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing .
为了提供与用户的交互,可以在计算机上实施此处描述的系统和技术,该计算机具有:设置为向用户显示信息的显示装置(例如,阴极射线管CRT(Cathode Ray Tube,CRT)或者液晶显示器(Liquid Crystal Display,LCD)监视器);以及键盘和指向装置(例如,鼠标或者轨迹球),用户可以通过该键盘和该指向装置来将输入提供给计算机。其它种类的装置还可以设置为提供与用户的交 互;例如,提供给用户的反馈可以是任何形式的传感反馈(例如,视觉反馈、听觉反馈、或者触觉反馈);并且可以用任何形式(包括声输入、语音输入或者、触觉输入)来接收来自用户的输入。To provide interaction with a user, the systems and techniques described herein may be implemented on a computer having a display device (eg, a Cathode Ray Tube (CRT) or a liquid crystal display) configured to display information to the user (Liquid Crystal Display, LCD monitor); and a keyboard and pointing device (eg, mouse or trackball) through which a user can provide input to the computer. Other kinds of devices may also be configured to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (eg, visual feedback, auditory feedback, or tactile feedback); and may be in any form (including acoustic input, voice input, or tactile input) to receive input from the user.
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如,作为数据服务器)、或者包括中间件部件的计算系统(例如,应用服务器)、或者包括前端部件的计算系统(例如,具有图形用户界面或者网络浏览器的用户计算机,用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如,通信网络)来将系统的部件相互连接。通信网络的示例包括:局域网(Local Area Network,LAN)、广域网(Wide Area Network,WAN)、区块链网络和互联网。The systems and techniques described herein may be implemented on a computing system that includes back-end components (eg, as a data server), or a computing system that includes middleware components (eg, an application server), or a computing system that includes front-end components (eg, a user's computer having a graphical user interface or web browser through which a user may interact with implementations of the systems and techniques described herein), or including such backend components, middleware components, Or any combination of front-end components in a computing system. The components of the system may be interconnected by any form or medium of digital data communication (eg, a communication network). Examples of communication networks include: Local Area Network (LAN), Wide Area Network (WAN), blockchain network, and the Internet.
计算机系统可以包括客户端和服务器。客户端和服务器一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并且彼此具有客户端-服务器关系的计算机程序来产生客户端和服务器的关系。服务器可以是云服务器,又称为云计算服务器或云主机,是云计算服务体系中的一项主机产品,以解决了传统物理主机与虚拟专用服务器(Virtual Private Server,VPS)服务中,存在的管理难度大,业务扩展性弱的缺陷。服务器也可以为分布式系统的服务器,或者是结合了区块链的服务器。A computer system can include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server arises by computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or a cloud host. It is a host product in the cloud computing service system to solve the problems existing in traditional physical host and virtual private server (Virtual Private Server, VPS) services. The management is difficult and the business expansion is weak. The server can also be a server of a distributed system, or a server combined with a blockchain.
可以使用上面所示的多种形式的流程,重新排序、增加或删除步骤。例如,本申请中记载的多个步骤可以并行地执行也可以顺序地执行也可以不同的次序执行,只要能够实现本申请公开的技术方案所期望的结果,本文在此不进行限制。Steps can be reordered, added, or removed using the various forms of flow shown above. For example, multiple steps described in the present application may be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in the present application can be achieved, no limitation is imposed herein.

Claims (21)

  1. 一种构建医学实体关系图的方法,包括:A method for building a medical entity relationship diagram, comprising:
    对医学相关数据进行实体识别,得到至少两个医学实体;Perform entity recognition on medical-related data to obtain at least two medical entities;
    确定所述至少两个医学实体所关联的药物分类标签;determining a drug classification label associated with the at least two medical entities;
    根据所述至少两个医学实体和所述至少两个医学实体所关联的药物分类标签,构建医学实体关系图。A medical entity relationship graph is constructed according to the at least two medical entities and the drug classification labels associated with the at least two medical entities.
  2. 根据权利要求1所述的方法,其中,所述根据所述至少两个医学实体和所述药物分类标签,构建医学实体关系图,包括:The method according to claim 1, wherein the constructing a medical entity relationship diagram according to the at least two medical entities and the drug classification label comprises:
    将每个医学实体和所述每个医学实体所关联的药物分类标签作为节点,将所述至少两个医学实体之间的关系、每个医学实体与所述每个医学实体所关联的药物分类标签之间的关系和所述至少两个医学实体所关联的药物分类标签之间的关系作为边,构建所述医学实体的关系图。Use each medical entity and the drug classification label associated with each medical entity as a node, and classify the relationship between the at least two medical entities and the drug classification associated with each medical entity and each medical entity The relationship between the tags and the relationship between the drug classification tags associated with the at least two medical entities are used as edges to construct a relationship graph of the medical entities.
  3. 根据权利要求1所述的方法,其中,所述对医学相关数据进行实体识别,得到至少两个医学实体,包括:The method according to claim 1, wherein, performing entity identification on medical-related data to obtain at least two medical entities, comprising:
    基于医学自然语言处理模型,对所述医学相关数据进行分词处理,以及对分词结果进行实体识别,得到所述至少两个医学实体;Based on a medical natural language processing model, word segmentation is performed on the medical-related data, and entity recognition is performed on the word segmentation result to obtain the at least two medical entities;
    其中,所述医学相关数据包括如下至少一项:药物说明书、医学书籍、医学指南和用户的医嘱病历数据;所述医学自然语言处理模型是基于双向长短期记忆网络、注意力机制和条件随机场构建的深层网络模型。Wherein, the medical-related data includes at least one of the following: drug instructions, medical books, medical guides, and user's medical order data; the medical natural language processing model is based on a bidirectional long short-term memory network, an attention mechanism, and a conditional random field The built deep network model.
  4. 根据权利要求1所述的方法,其中,所述确定所述至少两个医学实体所关联的药物分类标签,包括:The method of claim 1, wherein the determining of the drug classification labels associated with the at least two medical entities comprises:
    利用预先训练的标签预测模型对所述至少两个医学实体进行标签预测,根据预测结果确定所述至少两个医学实体所关联的药物分类标签。Use a pre-trained label prediction model to perform label prediction on the at least two medical entities, and determine the drug classification labels associated with the at least two medical entities according to the prediction results.
  5. 根据权利要求4所述的方法,其中,所述标签预测模型为知识增强的语义表示模型;训练所述标签预测模型的样本数据为从药物说明书中确定的实体描述数据和所述实体描述数据对应的药物分类标签。The method according to claim 4, wherein the label prediction model is a knowledge-enhanced semantic representation model; the sample data for training the label prediction model is the entity description data determined from the drug instructions corresponding to the entity description data drug classification labels.
  6. 一种医嘱质控的方法,包括:A method for quality control of doctor's orders, including:
    从待检测的医嘱病历数据中获取第一医学实体和第二医学实体;其中,第一医学实体是指实体类型为症状的医学实体,所述第二医学实体是指实体类型为药品的医学实体;Obtain a first medical entity and a second medical entity from the medical order medical record data to be detected; wherein, the first medical entity refers to a medical entity whose entity type is symptom, and the second medical entity refers to a medical entity whose entity type is medicine ;
    采用实体关系图,确定所述第一医学实体和所述第二医学实体之间的质控信息;其中,所述实体关系图采用如权利要求1-5中任一项所述的构建医学实体 关系图的方法构建。Using an entity relationship diagram, the quality control information between the first medical entity and the second medical entity is determined; wherein, the entity relationship diagram adopts the construction medical entity according to any one of claims 1-5. The method of building a relationship diagram.
  7. 根据权利要求6所述的方法,其中,所述采用实体关系图,确定所述第一医学实体和所述第二医学实体之间的质控信息,包括:The method according to claim 6, wherein the using an entity relationship diagram to determine the quality control information between the first medical entity and the second medical entity comprises:
    采用预设遍历算法,从所述实体关系图中确定以所述第一医学实体为起始节点,以所述第二医学实体为终止节点的至少一条连通路径;Using a preset traversal algorithm, determine from the entity relationship graph at least one connected path with the first medical entity as a starting node and the second medical entity as a termination node;
    从所述至少一条连通路径中确定路径最短的目标连通路径,并计算所述目标连通路径的关联得分;Determine a target connected path with the shortest path from the at least one connected path, and calculate an association score of the target connected path;
    根据所述关联得分,确定所述第一医学实体和所述第二医学实体之间的质控信息。According to the correlation score, quality control information between the first medical entity and the second medical entity is determined.
  8. 根据权利要求7所述的方法,其中,在所述目标连通路径是由所述第一医学实体、所述第二医学实体和至少一个药物分类标签组成的情况下,所述计算所述目标连通路径的关联得分,包括:7. The method of claim 7, wherein the calculating the target connectivity is performed where the target connectivity path is composed of the first medical entity, the second medical entity, and at least one drug classification label The associated score for the path, including:
    确定所述第一医学实体关联所述至少一个药物分类标签的第一概率值,以及所述第二医学实体关联所述至少一个药物分类标签的第二概率值;determining a first probability value associated with the at least one drug classification label by the first medical entity, and a second probability value associated with the at least one drug classification label by the second medical entity;
    对所述第一概率值和所述第二概率值进行加权处理,得到所述关联得分。The first probability value and the second probability value are weighted to obtain the association score.
  9. 根据权利要求6所述的方法,其中,所述从待检测的医嘱病历数据中获取第一医学实体和第二医学实体,包括:The method according to claim 6, wherein the obtaining the first medical entity and the second medical entity from the medical order medical record data to be detected comprises:
    基于医学自然语言处理模型,对所述待检测的医嘱病历数据进行分词处理,以及对分词结果进行实体识别,得到至少两个医学实体;Based on a medical natural language processing model, word segmentation is performed on the medical order medical record data to be detected, and entity recognition is performed on the word segmentation result to obtain at least two medical entities;
    对所述至少两个医学实体的实体类型进行识别;其中,所述实体类型包括如下至少一项:症状、疾病、检查、检验、手术和药品;Identifying entity types of the at least two medical entities; wherein the entity types include at least one of the following: symptoms, diseases, examinations, tests, surgery, and medicines;
    根据实体类型识别结果,获取所述第一医学实体和所述第二医学实体。Obtain the first medical entity and the second medical entity according to the entity type identification result.
  10. 一种构建医学实体关系图的装置,包括:A device for constructing a medical entity relationship diagram, comprising:
    实体识别模块,设置为对医学相关数据进行实体识别,得到至少两个医学实体;The entity recognition module is set to perform entity recognition on medical-related data to obtain at least two medical entities;
    标签预测模块,设置为确定所述至少两个医学实体所关联的药物分类标签;The label prediction module is configured to determine the drug classification labels associated with the at least two medical entities;
    关系图构建模块,设置为根据所述至少两个医学实体和所述至少两个医学实体所关联的药物分类标签,构建医学实体关系图。The relationship graph building module is configured to build a medical entity relationship graph according to the at least two medical entities and the drug classification labels associated with the at least two medical entities.
  11. 根据权利要求10所述的装置,其中,所述关系图构建模块,设置为:The apparatus of claim 10, wherein the relationship graph building module is configured to:
    将每个医学实体和所述每个医学实体所关联的药物分类标签作为节点,将所述至少两个医学实体之间的关系、每个医学实体与所述每个医学实体所关联 的药物分类标签之间的关系和所述至少两个医学实体所关联的药物分类标签之间的关系作为边,构建所述医学实体的关系图。Use each medical entity and the drug classification label associated with each medical entity as a node, and classify the relationship between the at least two medical entities and the drug classification associated with each medical entity and each medical entity The relationship between the tags and the relationship between the drug classification tags associated with the at least two medical entities are used as edges to construct a relationship graph of the medical entities.
  12. 根据权利要求10所述的装置,其中,所述实体识别模块,设置为:The device according to claim 10, wherein the entity identification module is configured to:
    基于医学自然语言处理模型,对所述医学相关数据进行分词处理,以及对分词结果进行实体识别,得到所述至少两个医学实体;Based on a medical natural language processing model, word segmentation is performed on the medical-related data, and entity recognition is performed on the word segmentation result to obtain the at least two medical entities;
    其中,所述医学相关数据包括如下至少一项:药物说明书、医学书籍、医学指南和用户的医嘱病历数据;所述医学自然语言处理模型是基于双向长短期记忆网络、注意力机制和条件随机场构建的深层网络模型。Wherein, the medical-related data includes at least one of the following: drug instructions, medical books, medical guides, and user's medical order data; the medical natural language processing model is based on a bidirectional long short-term memory network, an attention mechanism, and a conditional random field The built deep network model.
  13. 根据权利要求10所述的装置,其中,所述标签预测模块,设置为:The device according to claim 10, wherein the label prediction module is configured to:
    利用预先训练的标签预测模型对所述至少两个医学实体进行标签预测,根据预测结果确定所述至少两个医学实体所关联的药物分类标签。Use a pre-trained label prediction model to perform label prediction on the at least two medical entities, and determine drug classification labels associated with the at least two medical entities according to the prediction results.
  14. 根据权利要求13所述的装置,其中,所述标签预测模型为知识增强的语义表示模型;训练所述标签预测模型的样本数据为从药物说明书中确定的实体描述数据和所述实体描述数据对应的药物分类标签。The device according to claim 13, wherein the label prediction model is a knowledge-enhanced semantic representation model; the sample data for training the label prediction model is the entity description data determined from the drug instructions corresponding to the entity description data drug classification labels.
  15. 一种医嘱质控的装置,包括:A device for quality control of doctor's orders, comprising:
    实体获取模块,设置为从待检测的医嘱病历数据中获取第一医学实体和第二医学实体;其中,第一医学实体是指实体类型为症状的医学实体,所述第二医学实体是指实体类型为药品的医学实体;The entity acquisition module is configured to acquire a first medical entity and a second medical entity from the medical order medical record data to be detected; wherein, the first medical entity refers to a medical entity whose entity type is symptom, and the second medical entity refers to an entity a medical entity of type drug;
    质控模块,设置为采用实体关系图,确定所述第一医学实体和所述第二医学实体之间的质控信息;其中,所述实体关系图采用如权利要求1-5中任一项所述的构建医学实体关系图的方法构建。A quality control module, configured to use an entity relationship diagram to determine the quality control information between the first medical entity and the second medical entity; wherein, the entity relationship diagram adopts any one of claims 1-5 The method for constructing a medical entity relationship diagram is constructed.
  16. 根据权利要求15所述的装置,其中,所述质控模块,包括:The apparatus of claim 15, wherein the quality control module comprises:
    遍历单元,设置为采用预设遍历算法,从所述实体关系图中确定以所述第一医学实体为起始节点,以所述第二医学实体为终止节点的至少一条连通路径;A traversal unit, configured to adopt a preset traversal algorithm, and determine, from the entity relationship graph, at least one connected path with the first medical entity as a starting node and the second medical entity as a termination node;
    筛选计算单元,设置为从所述至少一条连通路径中确定路径最短的目标连通路径,并计算所述目标连通路径的关联得分;A screening calculation unit, configured to determine a target connected path with the shortest path from the at least one connected path, and calculate an association score of the target connected path;
    质控单元,设置为根据所述关联得分,确定所述第一医学实体和所述第二医学实体之间的质控信息。A quality control unit, configured to determine quality control information between the first medical entity and the second medical entity according to the correlation score.
  17. 根据权利要求16所述的装置,其中,在所述目标连通路径是由所述第一医学实体、所述第二医学实体和至少一个药物分类标签组成的情况下,所述筛选计算单元,包括:The apparatus according to claim 16, wherein, in the case where the target communication path is composed of the first medical entity, the second medical entity and at least one drug classification label, the screening calculation unit comprises: :
    概率值确定子单元,设置为确定所述第一医学实体关联所述至少一个药物分类标签的第一概率值,以及所述第二医学实体关联所述至少一个药物分类标签的第二概率值;a probability value determination subunit, configured to determine a first probability value associated with the at least one drug classification label by the first medical entity, and a second probability value associated with the at least one drug classification label by the second medical entity;
    计算子单元,设置为对所述第一概率值和所述第二概率值进行加权处理,得到所述关联得分。The calculation subunit is configured to perform weighting processing on the first probability value and the second probability value to obtain the association score.
  18. 根据权利要求15所述的装置,其中,所述实体获取模块,包括:The apparatus of claim 15, wherein the entity acquisition module comprises:
    实体识别单元,设置为基于医学自然语言处理模型,对所述待检测的医嘱病历数据进行分词处理,以及对分词结果进行实体识别,得到至少两个医学实体;an entity recognition unit, configured to perform word segmentation processing on the medical order medical record data to be detected based on a medical natural language processing model, and perform entity recognition on the word segmentation result to obtain at least two medical entities;
    类型识别单元,设置为对所述至少两个医学实体的实体类型进行识别;其中,所述实体类型包括如下至少一项:症状、疾病、检查、检验、手术和药品;a type identification unit, configured to identify the entity types of the at least two medical entities; wherein, the entity types include at least one of the following: symptoms, diseases, examinations, tests, surgery, and medicines;
    获取单元,设置为根据实体类型识别结果,获取所述第一医学实体和所述第二医学实体。The obtaining unit is configured to obtain the first medical entity and the second medical entity according to the entity type identification result.
  19. 一种电子设备,包括:An electronic device comprising:
    至少一个处理器;以及at least one processor; and
    与所述至少一个处理器通信连接的存储器;其中,a memory communicatively coupled to the at least one processor; wherein,
    所述存储器存储有可被所述至少一个处理器执行的指令,所述指令被所述至少一个处理器执行,以使所述至少一个处理器能够执行权利要求1-5或者权利要求6-9中任一项所述的方法。the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute claims 1-5 or claims 6-9 The method of any of the above.
  20. 一种存储有计算机指令的非瞬时计算机可读存储介质,其中,所述计算机指令用于使所述计算机执行根据权利要求1-5或者权利要求6-9中任一项所述的方法。A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the method of any one of claims 1-5 or 6-9.
  21. 一种计算机程序产品,包括计算机程序,所述计算机程序在被处理器执行时实现根据权利要求1-5或者权利要求6-9中任一项所述的方法。A computer program product comprising a computer program which, when executed by a processor, implements the method of any of claims 1-5 or 6-9.
PCT/CN2021/106769 2021-01-29 2021-07-16 Method and apparatus for constructing medical entity relationship diagram, method and apparatus for medical order quality control, device, and medium WO2022160614A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110129851.6A CN112786144B (en) 2021-01-29 2021-01-29 Knowledge graph method, doctor's advice quality control method, device, equipment and medium
CN202110129851.6 2021-01-29

Publications (1)

Publication Number Publication Date
WO2022160614A1 true WO2022160614A1 (en) 2022-08-04

Family

ID=75760006

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/106769 WO2022160614A1 (en) 2021-01-29 2021-07-16 Method and apparatus for constructing medical entity relationship diagram, method and apparatus for medical order quality control, device, and medium

Country Status (2)

Country Link
CN (1) CN112786144B (en)
WO (1) WO2022160614A1 (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112786144B (en) * 2021-01-29 2024-04-02 北京百度网讯科技有限公司 Knowledge graph method, doctor's advice quality control method, device, equipment and medium
CN114743693A (en) * 2022-03-21 2022-07-12 北京左医科技有限公司 Doctor-patient dialogue based center quality control method and center quality control device
CN115050441B (en) * 2022-08-16 2022-11-01 北京嘉和美康信息技术有限公司 Treatment scheme display method and device, electronic equipment and medium

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103955531A (en) * 2014-05-12 2014-07-30 南京提坦信息科技有限公司 Online knowledge map based on named entity library
US20160259899A1 (en) * 2015-03-04 2016-09-08 Expeda ehf Clinical decision support system for diagnosing and monitoring of a disease of a patient
CN109102855A (en) * 2018-07-03 2018-12-28 北京康夫子科技有限公司 Drug recommended method
CN109920508A (en) * 2018-12-28 2019-06-21 安徽省立医院 prescription auditing method and system
CN110534168A (en) * 2019-08-30 2019-12-03 北京百度网讯科技有限公司 Medicine advises indicating risk method, apparatus, electronic equipment and storage medium
CN111738014A (en) * 2020-06-16 2020-10-02 北京百度网讯科技有限公司 Drug classification method, device, equipment and storage medium
CN112786144A (en) * 2021-01-29 2021-05-11 北京百度网讯科技有限公司 Knowledge map method, medical advice quality control method, device, equipment and medium

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110532360A (en) * 2019-07-19 2019-12-03 平安科技(深圳)有限公司 Medical field knowledge mapping question and answer processing method, device, equipment and storage medium

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103955531A (en) * 2014-05-12 2014-07-30 南京提坦信息科技有限公司 Online knowledge map based on named entity library
US20160259899A1 (en) * 2015-03-04 2016-09-08 Expeda ehf Clinical decision support system for diagnosing and monitoring of a disease of a patient
CN109102855A (en) * 2018-07-03 2018-12-28 北京康夫子科技有限公司 Drug recommended method
CN109920508A (en) * 2018-12-28 2019-06-21 安徽省立医院 prescription auditing method and system
CN110534168A (en) * 2019-08-30 2019-12-03 北京百度网讯科技有限公司 Medicine advises indicating risk method, apparatus, electronic equipment and storage medium
CN111738014A (en) * 2020-06-16 2020-10-02 北京百度网讯科技有限公司 Drug classification method, device, equipment and storage medium
CN112786144A (en) * 2021-01-29 2021-05-11 北京百度网讯科技有限公司 Knowledge map method, medical advice quality control method, device, equipment and medium

Also Published As

Publication number Publication date
CN112786144B (en) 2024-04-02
CN112786144A (en) 2021-05-11

Similar Documents

Publication Publication Date Title
US11942221B2 (en) Disambiguation of ambiguous portions of content for processing by automated systems
Pezoulas et al. Medical data quality assessment: On the development of an automated framework for medical data curation
WO2022160614A1 (en) Method and apparatus for constructing medical entity relationship diagram, method and apparatus for medical order quality control, device, and medium
US11200968B2 (en) Verifying medical conditions of patients in electronic medical records
US10614196B2 (en) System for automated analysis of clinical text for pharmacovigilance
US11823798B2 (en) Container-based knowledge graphs for determining entity relations in non-narrative text
CN111801741B (en) Adverse drug reaction analysis
US20180089383A1 (en) Container-Based Knowledge Graphs for Determining Entity Relations in Medical Text
US10431338B2 (en) System and method for weighting manageable patient attributes during criteria evaluations for treatment
WO2018201772A1 (en) Method and system for inferring potential disease from medical text, and readable storage medium
US11495332B2 (en) Automated prediction and answering of medical professional questions directed to patient based on EMR
CN108920453A (en) Data processing method, device, electronic equipment and computer-readable medium
JP7285893B2 (en) MEDICAL DATA VERIFICATION METHOD, DEVICE AND ELECTRONIC DEVICE
CN112528660A (en) Method, apparatus, device, storage medium and program product for processing text
US20220068482A1 (en) Interactive treatment pathway interface for guiding diagnosis or treatment of a medical condition
US20190340487A1 (en) Neural Network Architecture for Performing Medical Coding
US11532387B2 (en) Identifying information in plain text narratives EMRs
Wang et al. Artificial intelligence-empowered chatbot for effective COVID-19 information delivery to older adults
US11334720B2 (en) Machine learned sentence span inclusion judgments
Nakayama et al. Making sense of abbreviations in nursing notes: A case study on mortality prediction
CN112613313A (en) Method, device, equipment, storage medium and program product for quality control of medical orders
US10886030B2 (en) Presenting contextually relevant patient data in relation to other patients to a medical professional
GB2616369A (en) Sentiment detection using medical clues
Zhang et al. Medical Q&A statement NER based on ECA attention mechanism and lexical enhancement
US20240111999A1 (en) Segmenting and classifying unstructured text using multi-task neural networks

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21922211

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21922211

Country of ref document: EP

Kind code of ref document: A1