Intelligent hospital division and guidance method and system based on knowledge graph and natural language processing

Abstract

The invention provides an intelligent hospital division and guidance method and system based on a knowledge graph and natural language processing, and the method comprises the steps: obtaining consultation information inputted by a to-be-guided user, the consultation information at least comprising symptom description and / or medical history information; identifying a target symptom entity from the consultation information; determining a target department entity matched with the target symptom entity from a pre-constructed medical knowledge graph, wherein the medical knowledge graph is constructed based on the medical data and a natural language processing technology; and recommending the doctor-seeing department name corresponding to the target department entity to the to-be-guided user. According to the scheme, the target symptom entity is extracted from the consultation information input by the to-be-guided user, then the target department entity matched with the target symptom entity is determined from the medical knowledge graph, the doctor-seeing department name corresponding to the target department entity is recommended to the to-be-guided user, and the medical guidance service is provided for the user through the medical knowledge graph. And the hospital guiding efficiency and the hospital guiding accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and in particular to an intelligent diagnosis and guidance method and system based on knowledge graph and natural language processing. Background Art

[0002] In the medical environment, patients often face problems such as difficulty in choosing a department and long waiting times when seeking medical treatment. Patients often find it difficult to accurately determine the department corresponding to their symptoms.

[0003] Currently, the triage and guidance method mainly relies on manual consultation, but the triage and guidance personnel need to face a large number of patient consultations, and the triage and guidance personnel rely on their own experience to provide consulting services to patients. The efficiency and accuracy of triage and guidance are poor. Summary of the Invention

[0004] In view of this, an embodiment of the present invention provides an intelligent triage method and system based on knowledge graph and natural language processing to solve the problems of poor triage efficiency and poor accuracy in the existing method of triage that mainly relies on manual consultation.

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0006] A first aspect of an embodiment of the present invention discloses an intelligent diagnosis and guidance method based on knowledge graph and natural language processing, the method comprising:

[0007] Obtaining consultation information to be input by the user, wherein the consultation information at least includes a symptom description and / or medical history information;

[0008] identifying a target symptom entity from the consultation information;

[0009] Determining a target department entity that matches the target symptom entity from a pre-constructed medical knowledge graph, wherein the medical knowledge graph is constructed based on medical data and natural language processing technology;

[0010] The name of the medical department corresponding to the target department entity is recommended to the user to be guided.

[0011] Preferably, identifying a target symptom entity from the consultation information includes:

[0012] If the consultation information is text information, using natural language processing technology to identify the target symptom entity from the consultation information;

[0013] If the consultation information is voice information, performing a first preprocessing on the consultation information, the first preprocessing at least including noise reduction and endpoint detection;

[0014] converting the consultation information after the first preprocessing into text information;

[0015] Utilizing natural language processing technology, target symptom entities are identified from the consultation information converted into text information.

[0016] Preferably, the process of constructing a medical knowledge graph based on medical data and natural language processing technology includes:

[0017] Collect medical data;

[0018] performing a second preprocessing on the medical data, the second preprocessing at least including data cleaning, labeling, and desensitization;

[0019] Extracting medical entities and relationships between entities from the medical data after the second preprocessing using natural language processing technology, wherein the medical entities include at least disease entities, symptom entities, examination item entities, and department entities;

[0020] A medical knowledge graph is constructed through the medical entities and the relationships between the entities.

[0021] Preferably, the medical knowledge graph includes at least disease entities, symptom entities, examination item entities, department entities, and relationships between entities;

[0022] Determine the target department entity that matches the target symptom entity from the pre-built medical knowledge graph, including:

[0023] Using the relationship between entities, determine the target disease entity corresponding to the target symptom entity from the pre-built medical knowledge graph;

[0024] Utilizing the relationship between the entities and the medical history information of the user to be guided, a target department entity corresponding to the target disease entity is determined from the medical knowledge graph.

[0025] Preferably, recommending the name of the medical department corresponding to the target department entity to the user to be guided includes:

[0026] Identify the target hospital where the user to be guided is located, and determine the target standardized department name corresponding to the target department entity;

[0027] Determine the name of the treatment department in the target hospital that corresponds to the target standardized department name using a preset department classification system, wherein the department classification system includes: a mapping relationship between each treatment department name in each hospital and each standardized department name;

[0028] The name of the medical department corresponding to the target standardized department name is recommended to the user to be guided.

[0029] Preferably, after recommending the name of the medical department corresponding to the target department entity to the user to be guided, the method further includes:

[0030] Determining a target examination item entity corresponding to the target disease entity from the medical knowledge graph;

[0031] The inspection items corresponding to the target inspection item entity are recommended to the user to be guided.

[0032] A second aspect of an embodiment of the present invention discloses an intelligent diagnosis and guidance system based on knowledge graph and natural language processing, the system comprising:

[0033] an acquiring unit, configured to acquire consulting information to be input by the user to be guided, wherein the consulting information at least includes a symptom description and / or medical history information;

[0034] an identification unit, configured to identify a target symptom entity from the consultation information;

[0035] a determination unit, configured to determine a target department entity that matches the target symptom entity from a pre-constructed medical knowledge graph, wherein the medical knowledge graph is constructed based on medical data and natural language processing technology;

[0036] The recommendation unit is used to recommend the name of the medical department corresponding to the target department entity to the user to be guided.

[0037] Preferably, the identification unit includes:

[0038] A first recognition module is configured to, if the consultation information is text information, use natural language processing technology to identify a target symptom entity from the consultation information;

[0039] a preprocessing module, configured to perform a first preprocessing on the consultation information if the consultation information is voice information, the first preprocessing including at least noise reduction and endpoint detection;

[0040] a conversion module, configured to convert the consultation information after the first preprocessing into text information;

[0041] The second recognition module is used to use natural language processing technology to identify the target symptom entity from the consultation information converted into text information.

[0042] Preferably, the determining unit includes:

[0043] Collection module, used to collect medical data;

[0044] a preprocessing module, configured to perform a second preprocessing on the medical data, wherein the second preprocessing includes at least data cleaning, labeling, and desensitization;

[0045] an extraction module, configured to extract medical entities and inter-entity relationships from the medical data after the second preprocessing using natural language processing technology, wherein the medical entities include at least disease entities, symptom entities, examination item entities, and department entities;

[0046] A construction module is used to construct a medical knowledge graph through the medical entities and the relationships between the entities.

[0047] Preferably, the medical knowledge graph includes at least disease entities, symptom entities, examination item entities, department entities, and relationships between entities;

[0048] The determination unit is specifically used to: utilize the relationship between entities to determine the target disease entity corresponding to the target symptom entity from a pre-constructed medical knowledge graph; utilize the relationship between entities and the medical history information of the user to be guided to determine the target department entity corresponding to the target disease entity from the medical knowledge graph.

[0049] Based on the above-mentioned embodiment of the present invention, an intelligent diagnosis and guidance method and system based on knowledge graph and natural language processing is provided. The method is as follows: obtaining consulting information input by the user to be guided, the consulting information at least including symptom description and / or medical history information; identifying the target symptom entity from the consulting information; determining the target department entity that matches the target symptom entity from the pre-constructed medical knowledge graph, the medical knowledge graph is constructed based on medical data and natural language processing technology; and recommending the name of the medical department corresponding to the target department entity to the user to be guided. This solution extracts the target symptom entity from the consulting information input by the user to be guided, and then determines the target department entity that matches the target symptom entity from the medical knowledge graph, and recommends the name of the medical department corresponding to the target department entity to the user to be guided. The medical knowledge graph is used to provide diagnosis and guidance services to the user, thereby improving the efficiency and accuracy of diagnosis and guidance. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0051] Figure 1 A flowchart of an intelligent diagnosis and guidance method based on knowledge graph and natural language processing provided by an embodiment of the present invention;

[0052] Figure 2A flowchart for constructing a medical knowledge graph provided by an embodiment of the present invention;

[0053] Figure 3 An overall flow chart of an intelligent diagnosis and guidance method based on knowledge graph and natural language processing provided by an embodiment of the present invention;

[0054] Figure 4 This is a structural diagram of an intelligent diagnosis and guidance system based on knowledge graph and natural language processing provided by an embodiment of the present invention;

[0055] Figure 5 Another structural block diagram of an intelligent diagnosis and guidance system based on knowledge graph and natural language processing provided by an embodiment of the present invention;

[0056] Figure 6 Another structural block diagram of an intelligent diagnosis and guidance system based on knowledge graph and natural language processing provided by an embodiment of the present invention;

[0057] Figure 7 Another structural block diagram of an intelligent diagnosis and guidance system based on knowledge graph and natural language processing provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] In this application, the terms "comprises," "comprising," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not preclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0060] In the medical environment, patients often face difficulties in selecting a department and long wait times when seeking treatment. This is especially true in large hospitals, where numerous departments are specialized and specialized. Patients often struggle to accurately determine the department that best suits their symptoms. This leads to overcrowded triage staff, low efficiency, and high triage costs, impacting the patient experience and the optimal use of medical resources.

[0061] For example: A survey of a large tertiary hospital showed that patients spent an average of more than 30 minutes looking for a suitable department.

[0062] Existing triage methods mainly rely on manual consultation or simple symptom classification, which has problems such as low efficiency and poor accuracy. For example, the accuracy rate of manual triage is only about 60%, and it is easily affected by the experience of the triage personnel.

[0063] Research has found that with the development of artificial intelligence and big data technologies, natural language-based intelligent triage technology has become a research hotspot. However, existing technologies still face numerous challenges. Regarding data quality, medical data comes from a wide range of sources and formats, and data standards across different hospital information systems are inconsistent. This leads to errors and omissions in the data, which in turn affects the accuracy of intelligent triage. Furthermore, most existing intelligent triage systems lack the ability to deeply understand and reason about medical knowledge, making them unable to effectively handle the complex problem of matching symptoms with departments.

[0064] To this end, this solution proposes an intelligent triage method and system based on knowledge graph and natural language processing. The target symptom entity is extracted from the consultation information input by the user to be guided, and then the target department entity matching the target symptom entity is determined from the medical knowledge graph. The name of the medical department corresponding to the target department entity is recommended to the user to be guided. The medical knowledge graph constructed based on medical data and natural language processing technology is used to provide users with triage services, thereby improving the efficiency and accuracy of triage.

[0065] This solution effectively solves the data quality problem by introducing a combination of medical knowledge graphs and natural language processing technology, and achieves accurate triage and guidance through deep semantic understanding and reasoning capabilities, significantly improving the efficiency and accuracy of triage and guidance. The following explains this solution through various examples.

[0066] See also Figure 1 , shows a flow chart of an intelligent diagnosis and guidance method based on knowledge graph and natural language processing provided by an embodiment of the present invention, the intelligent diagnosis and guidance method comprising:

[0067] Step S101: Obtain consultation information to be input by the user.

[0068] In the specific implementation of step S101 , consulting information to be input by the user is obtained. The consulting information at least includes symptom description and / or medical history information. The medical history information also includes corresponding examination information.

[0069] In actual applications, the user to be guided can input the consultation information through "text input" or "voice input".

[0070] When consulting information is input through "text input", the type of consulting information is text information; when consulting information is input through "voice input", the type of consulting information is voice information (or voice signal).

[0071] Step S102: Identify the target symptom entity from the consultation information.

[0072] In the specific implementation of step S102 , after obtaining the consultation information to be input by the user, a target symptom entity is identified from the consultation information. The target symptom entity is: a symptom entity that matches the consultation information.

[0073] Specifically, if the consultation information is text information, natural language processing technology is used to identify the target symptom entity from the consultation information.

[0074] If the consultation information is voice information, the consultation information is subjected to a first preprocessing, and the first preprocessing includes at least noise reduction and endpoint detection; the consultation information after the first preprocessing is converted into text information; and natural language processing technology is used to identify the target symptom entity from the consultation information converted into text information.

[0075] Among them, the specific implementation method of identifying the target symptom entity from the consulting information converted into text information is: using natural language processing technology to perform word segmentation, part-of-speech tagging, named entity recognition and other operations on the consulting information converted into text information, so as to extract the target symptom entity.

[0076] It can be understood that after extracting the target symptom entity, the target symptom entity is semantically understood in combination with context information to ensure the accuracy of the recognition result.

[0077] Step S103: Determine the target department entity that matches the target symptom entity from the pre-built medical knowledge graph.

[0078] It should be noted that the medical knowledge graph is constructed based on medical data and natural language processing technology. The medical knowledge graph includes at least disease entities, symptom entities, examination item entities, department entities, and relationships between entities.

[0079] In the specific implementation of step S103, the relationship between entities is used to determine the target disease entity corresponding to the target symptom entity from the pre-constructed medical knowledge graph.

[0080] That is, through the relationship between the symptom entity and the department entity, the target disease entity corresponding to the target symptom entity is determined from the medical knowledge graph. The target disease entity is: the disease entity corresponding to the target symptom entity.

[0081] By utilizing the relationship between entities and the medical history information of the user to be guided, the target department entity corresponding to the target disease entity is determined from the medical knowledge graph.

[0082] That is, through the relationship between the disease entity and the department entity, combined with the medical history information of the user to be guided, the target department entity corresponding to the target disease entity is determined from the medical knowledge graph. The target department entity is: the department entity corresponding to the target disease entity.

[0083] Step S104: recommending the name of the medical department corresponding to the target department entity to the user to be guided.

[0084] In the specific implementation of step S104, the target hospital where the user to be guided is located is identified, and the target standardized department name corresponding to the target department entity is determined. The target hospital is the hospital where the user to be guided goes for treatment, and the target standardized department name is: the standardized department name corresponding to the target department entity.

[0085] The preset department classification system is used to determine the treatment department name corresponding to the "target standardized department name" in the target hospital. The department classification system includes: the mapping relationship between each treatment department name in each hospital and each standardized department name.

[0086] The name of the medical department corresponding to the target standardized department name is recommended to the user to be guided.

[0087] It should be noted that different hospitals have different divisions of departments. For example, the granularity of department division in large hospitals is finer than that in small hospitals.

[0088] For example, small hospitals only have the "Internal Medicine" department, while large hospitals have multiple departments under the "Internal Medicine" department.

[0089] In order to ensure that the accurate name of the medical department can be recommended to the user to be guided after the target department entity is identified, this solution pre-establishes a standardized department classification system, which includes: the mapping relationship between the names of each medical department in each hospital and the standardized department names.

[0090] That is, the content in the department classification system represents: which medical department name in a hospital corresponds to the standardized department name.

[0091] This department classification system, combined with dynamic department matching, addresses differences in department settings between hospitals. After identifying the target department entity, the target hospital where the user to be guided is located is identified and the target standardized department name corresponding to the target department entity is determined. Then, the department classification system is used to determine the name of the treatment department in the target hospital that corresponds to the target standardized department name. At this point, the determined treatment department name is the name of a department in the target hospital (that is, the actual department visited). The treatment department name corresponding to the target standardized department name is recommended to the user to be guided, thus achieving accurate recommendations.

[0092] In some specific embodiments, the department classification system is updated regularly to ensure that the department recommendation function is always kept up to date.

[0093] In some specific embodiments, after the name of the medical department corresponding to the target department entity is recommended to the user to be guided, the target examination item entity corresponding to the target disease entity is determined from the medical knowledge graph, and the examination items corresponding to the target examination item entity are recommended to the user to be guided.

[0094] Specifically, by utilizing the relationship between the disease entity and the examination item entity, the target examination item entity corresponding to the target disease entity is determined from the medical knowledge graph, and the examination items corresponding to the target examination item entity are recommended to the user to be guided. The target examination item entity is: the examination item entity corresponding to the target disease entity.

[0095] In an embodiment of the present invention, a target symptom entity is extracted from the consultation information input by the user to be guided, and then a target department entity matching the target symptom entity is determined from the medical knowledge graph, and the name of the medical department corresponding to the target department entity is recommended to the user to be guided. The medical knowledge graph is used to provide users with diagnosis and guidance services, thereby improving the efficiency and accuracy of diagnosis and guidance.

[0096] For the above embodiments of the present invention Figure 1 The medical knowledge graph involved in step S103 is shown in Figure 2 , which shows a flowchart of constructing a medical knowledge graph provided by an embodiment of the present invention, Figure 2 The steps include:

[0097] Step S201: Collect medical data.

[0098] In the specific implementation of step S201 , a massive amount of medical data is collected, and the medical data includes but is not limited to multi-source data such as medical literature, clinical guidelines, disease databases, and electronic medical records.

[0099] It should be noted that when collecting medical data, authorization has been obtained from the relevant data parties, that is, this plan collects medical data on the premise of legality and compliance.

[0100] Step S202: performing a second preprocessing on the medical data.

[0101] In the specific implementation of step S202, the collected medical data is subjected to a second preprocessing to ensure the accuracy and security of the medical data. The second preprocessing includes at least data cleaning, labeling, and desensitization.

[0102] That is to say, medical data is cleaned, labeled, and desensitized.

[0103] Data cleaning is specifically implemented by using data cleaning algorithms to remove duplicate, erroneous, and incomplete data records. For example, duplicate data can be identified and deleted by comparing key information such as patient ID and visit time. Incorrect data can be corrected using data validation rules and logical checks. Missing data can be supplemented or marked based on data characteristics and statistical patterns.

[0104] The specific implementation method of desensitization is: follow strict data desensitization specifications, use encryption, hash functions and other technologies to process patient sensitive information, such as name, ID number, contact information, etc., to protect patient privacy and make the data meet safety standards.

[0105] The specific implementation method of labeling is: using professional labeling tools, based on unified medical labeling standards, to label medical entities such as diseases, symptoms, examination items, etc. in medical data.

[0106] Step S203: using natural language processing technology, extracting medical entities and relationships between entities from the medical data after the second preprocessing.

[0107] In the specific implementation of step S203, named entity recognition, relationship extraction and other methods in natural language processing technology are used to extract medical entities and inter-entity relationships from the medical data after the second preprocessing. Medical entities include at least disease entities, symptom entities, examination item entities, and department entities. The inter-entity relationships represent the mutual relationships between the medical entities.

[0108] Among them, the disease entity includes information such as the name, definition, symptoms, cause, and treatment of the disease.

[0109] Symptom entity: describes the various symptoms that a patient may experience and their characteristics.

[0110] Examination item entity: covers various medical examination items and their applicable scope, examination methods, etc.

[0111] Department entity: represents the different departments of the hospital and their professional areas.

[0112] Relationships between entities: including the association between diseases and symptoms, the correspondence between diseases and examination items, the affiliation between diseases and departments, etc.

[0113] Step S204: Construct a medical knowledge graph through medical entities and relationships between entities.

[0114] In the specific implementation of step S204, the extracted medical entities and the relationships between entities are used to construct a medical knowledge graph.

[0115] The above embodiments of the present invention Figure 2 , which is an explanation about building a medical knowledge graph.

[0116] The implementation of this solution involves the following parts: "Construction of medical knowledge graph", "Development of symptom department matching algorithm", "Development of voice symptom entity recognition algorithm", "Development of department difference matching algorithm", and "Intelligent diagnosis and guidance". Figure 3 The overall flow chart shown explains the above parts. Figure 3 The steps include:

[0117] Step S301: Constructing a medical knowledge graph.

[0118] In the specific implementation of step S301, machine learning, big models and natural language processing technologies are used to extract medical entities and relationships between entities in medical data (raw data), thereby constructing a medical knowledge graph and dynamically updating the medical knowledge graph.

[0119] Specifically, the construction of medical knowledge graph is mainly divided into two steps: "data preprocessing" and "constructing knowledge graph".

[0120] Data preprocessing: Collect massive amounts of medical data and perform secondary preprocessing (data cleaning, labeling, desensitization, etc.) on the collected medical data.

[0121] Constructing a knowledge graph: Using methods such as named entity recognition and relationship extraction in natural language processing technology, medical entities and relationships between entities are extracted from the medical data after the second preprocessing, and a medical knowledge graph is constructed based on this.

[0122] Step S302: Development of symptom-department matching algorithm.

[0123] In the specific implementation of step S302, a matching algorithm between symptoms and departments is developed based on the constructed medical knowledge graph to achieve accurate department recommendations.

[0124] Based on the constructed medical knowledge graph, accurate diagnosis and guidance are achieved through symptom matching and department recommendation. At the same time, personalized diagnosis and guidance suggestions are provided based on the patient's medical history information and examination item requirements.

[0125] Specifically, the development of the symptom-department matching algorithm is mainly divided into three steps: "symptom matching", "department recommendation" and "examination item recommendation".

[0126] Symptom matching: Match the symptom description, examination information, and other consulting information entered by the user to be guided (i.e., the patient) with the relevant medical entities in the medical knowledge graph to find the disease entity most relevant to the symptoms of the user to be guided (i.e., find the target disease entity).

[0127] Department recommendation: Based on the relationship between the disease entity and the department entity (i.e., the relationship between entities), combined with the medical history information of the user to be guided, the most suitable department for the user to be guided is recommended (i.e., the target department entity is determined).

[0128] Examination item recommendation: Based on the diagnosis needs of the disease and combined with the examination item entities in the medical knowledge graph, necessary examination items are recommended to the user to be guided (i.e., the target examination item entities are determined), which can also assist doctors in making a quick diagnosis.

[0129] Step S303: Development of speech symptom entity recognition algorithm.

[0130] In the specific implementation of step S303, based on massive data of triage and guidance dialogues and doctor-patient dialogues, the voice symptom entity recognition algorithm is trained to improve the accuracy and efficiency of symptom recognition, that is, to achieve rapid processing of voice input and accurate recognition of symptom entities.

[0131] Specifically, the development of speech symptom entity recognition algorithm is mainly divided into four steps: "speech collection and preprocessing", "speech recognition", "symptom entity recognition" and "semantic understanding".

[0132] Voice collection and preprocessing: Voice information such as the symptom description of the user to be guided is collected through the voice input device, and the voice information is preprocessed (noise reduction and endpoint detection).

[0133] Speech recognition: converting the speech information after the first preprocessing into text information.

[0134] Symptom entity recognition: Use natural language processing technology to perform operations such as word segmentation, part-of-speech tagging, and named entity recognition on text information to extract symptom entities.

[0135] Semantic understanding: Combined with the context, semantic understanding of symptom entities is performed to ensure the accuracy of recognition results.

[0136] Step S304: Development of department difference matching algorithm.

[0137] In the specific implementation of step S304, differences in department settings between different hospitals are resolved based on the mapping relationship between standardized departments and department settings of specific hospitals.

[0138] That is, establish a standardized department classification system, combined with dynamic department matching, to resolve the differences in department settings between hospitals.

[0139] Specifically, the development of the department difference matching algorithm is mainly divided into three steps: "standardized department mapping", "dynamic department adaptation" and "department information update".

[0140] Standardized department mapping: Establish a standardized department classification system and map the department settings of each hospital to this department classification system.

[0141] Dynamic department adaptation: Dynamically adjust the department recommendation results based on the specific department settings of the hospital where the user to be guided is located (target hospital) to ensure that the recommended department is consistent with the actual department settings of the hospital.

[0142] Department information update: Regularly update the department setting information of each hospital in the department classification system to ensure that the department recommendation function is always kept up to date.

[0143] Step S305: Intelligent diagnosis and guidance.

[0144] In the specific implementation of step S305, the above algorithms are used to achieve rapid intelligent triage and guidance of patients, and provide personalized triage and guidance suggestions. That is, the above algorithms are used to achieve rapid intelligent triage and guidance, and provide information such as recommended departments and examination items.

[0145] Specifically, intelligent triage is mainly divided into four steps: "patient information input", "information processing and analysis", "triage result generation", and "result display and feedback".

[0146] Patient information input: The user is guided to input consultation information through voice or text.

[0147] Information processing and analysis: Pre-process the consultation information, use the voice symptom entity recognition algorithm to extract the target disease entity, and combine it with the medical knowledge graph for semantic understanding.

[0148] Generation of triage results: Based on the symptom-department matching algorithm and the department difference matching algorithm, triage results are generated for the user to be guided. The triage results include the recommended department name and examination items.

[0149] Result display and feedback: The triage results will be displayed to the user to be guided, and a feedback mechanism will be provided to allow the user to be guided to evaluate and provide feedback on the triage results in order to optimize and improve the department recommendation function.

[0150] It should be noted that Figure 3 For the execution principle of each step, please refer to the above embodiment of the present invention. Figure 1 and Figure 2 The content will not be repeated here.

[0151] In general, this solution achieves rapid and intelligent patient triage and guidance by building a medical knowledge graph and developing related algorithms, combined with voice symptom entity recognition technology. This has the following beneficial effects:

[0152] Improve the efficiency of diagnosis and treatment: Through intelligent analysis and recommendation, it reduces the time patients spend searching for departments in the hospital and improves the efficiency of medical treatment.

[0153] Improve the accuracy of diagnosis and guidance: Based on the deep semantic understanding and reasoning capabilities of the medical knowledge graph, combined with natural language processing technology, it can more accurately judge the patient's condition and the corresponding department, reducing misdiagnosis and missed diagnosis caused by human error.

[0154] Optimize the allocation of medical resources: Rationally guide patients to go to appropriate departments for treatment, avoid patients blindly concentrating in certain popular departments, and improve the utilization rate of medical resources.

[0155] Enhance patient medical experience: Provide patients with personalized diagnosis and guidance suggestions, which enhances their satisfaction and trust in medical services.

[0156] Solve the problem of differences in department settings: Through standardized department mapping and dynamic department adaptation, the problem of differences in department settings between different hospitals is solved.

[0157] Corresponding to the intelligent diagnosis and guidance method based on knowledge graph and natural language processing provided by the above embodiment of the present invention, see Figure 4 An embodiment of the present invention also provides a structural block diagram of an intelligent diagnosis and guidance system based on knowledge graph and natural language processing. The intelligent diagnosis and guidance system includes: an acquisition unit 100, an identification unit 200, a determination unit 300 and a recommendation unit 400.

[0158] The acquisition unit 100 is used to acquire consultation information to be input by the user, where the consultation information at least includes symptom description and / or medical history information.

[0159] The identification unit 200 is used to identify the target symptom entity from the consultation information.

[0160] The determination unit 300 is used to determine the target department entity that matches the target symptom entity from a pre-constructed medical knowledge graph, where the medical knowledge graph is constructed based on medical data and natural language processing technology.

[0161] In some embodiments, the medical knowledge graph includes at least disease entities, symptom entities, examination item entities, department entities, and relationships between entities; the determination unit 300 is specifically used to: utilize the relationships between entities to determine the target disease entity corresponding to the target symptom entity from the pre-constructed medical knowledge graph; utilize the relationships between entities and the medical history information of the user to be guided to determine the target department entity corresponding to the target disease entity from the medical knowledge graph.

[0162] The recommendation unit 400 is used to recommend the name of the medical department corresponding to the target department entity to the user to be guided.

[0163] Preferably, the determination unit 300 is further used to determine the target examination item entity corresponding to the target disease entity from the medical knowledge graph.

[0164] The recommendation unit 400 is further configured to recommend the examination items corresponding to the target examination item entity to the user to be guided.

[0165] In an embodiment of the present invention, a target symptom entity is extracted from the consultation information input by the user to be guided, and then a target department entity matching the target symptom entity is determined from the medical knowledge graph, and the name of the medical department corresponding to the target department entity is recommended to the user to be guided. The medical knowledge graph is used to provide users with diagnosis and guidance services, thereby improving the efficiency and accuracy of diagnosis and guidance.

[0166] Preferably, see Figure 5 , shows another structural block diagram of an intelligent diagnosis and guidance system based on knowledge graph and natural language processing provided by an embodiment of the present invention, wherein the recognition unit 200 includes:

[0167] The first recognition module 201 is used to identify the target symptom entity from the consultation information by using natural language processing technology if the consultation information is text information.

[0168] The pre-processing module 202 is configured to perform a first pre-processing on the consultation information if the consultation information is voice information. The first pre-processing at least includes noise reduction and endpoint detection.

[0169] The conversion module 203 is configured to convert the consultation information after the first pre-processing into text information.

[0170] The second recognition module 204 is used to use natural language processing technology to identify target symptom entities from the consultation information converted into text information.

[0171] Preferably, see Figure 6 , shows another structural block diagram of an intelligent diagnosis and guidance system based on knowledge graph and natural language processing provided by an embodiment of the present invention, wherein the determination unit 300 includes:

[0172] The collection module 301 is used to collect medical data.

[0173] The preprocessing module 302 is used to perform a second preprocessing on the medical data, and the second preprocessing at least includes data cleaning, labeling and desensitization.

[0174] The extraction module 303 is used to extract medical entities and inter-entity relationships from the medical data after the second preprocessing using natural language processing technology. The medical entities include at least disease entities, symptom entities, examination item entities, and department entities.

[0175] The construction module 304 is used to construct a medical knowledge graph through medical entities and relationships between entities.

[0176] Preferably, see Figure 7 , shows another structural block diagram of an intelligent diagnosis and guidance system based on knowledge graph and natural language processing provided by an embodiment of the present invention, wherein the recommendation unit 400 includes:

[0177] The processing module 401 is used to identify the target hospital where the user to be guided is located, and determine the target standardized department name corresponding to the target department entity.

[0178] The determination module 402 is used to determine the treatment department name corresponding to the target standardized department name in the target hospital using a preset department classification system. The department classification system includes: a mapping relationship between each treatment department name in each hospital and each standardized department name.

[0179] The recommendation module 403 is used to recommend the name of the medical department corresponding to the target standardized department name to the user to be guided.

[0180] In summary, the embodiments of the present invention provide an intelligent triage method and system based on knowledge graph and natural language processing, which extracts the target symptom entity from the consultation information input by the user to be guided, and then determines the target department entity matching the target symptom entity from the medical knowledge graph, and recommends the name of the medical department corresponding to the target department entity to the user to be guided. The medical knowledge graph is used to provide users with triage services, thereby improving the efficiency and accuracy of triage.

[0181] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple. For relevant parts, refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0182] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0183] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent diagnosis and guidance method based on knowledge graph and natural language processing, characterized in that: The method comprises: Obtaining consultation information to be input by the user, wherein the consultation information at least includes a symptom description and / or medical history information; identifying a target symptom entity from the consultation information; Determining a target department entity that matches the target symptom entity from a pre-constructed medical knowledge graph, wherein the medical knowledge graph is constructed based on medical data and natural language processing technology; The name of the medical department corresponding to the target department entity is recommended to the user to be guided.

2. The method according to claim 1, characterized in that Identifying a target symptom entity from the consultation information includes: If the consultation information is text information, using natural language processing technology to identify the target symptom entity from the consultation information; If the consultation information is voice information, performing a first preprocessing on the consultation information, the first preprocessing at least including noise reduction and endpoint detection; converting the consultation information after the first preprocessing into text information; Utilizing natural language processing technology, target symptom entities are identified from the consultation information converted into text information.

3. The method according to claim 1, characterized in that The process of building a medical knowledge graph based on medical data and natural language processing technology includes: Collect medical data; performing a second preprocessing on the medical data, the second preprocessing at least including data cleaning, labeling, and desensitization; Extracting medical entities and relationships between entities from the medical data after the second preprocessing using natural language processing technology, wherein the medical entities include at least disease entities, symptom entities, examination item entities, and department entities; A medical knowledge graph is constructed through the medical entities and the relationships between the entities.

4. The method according to any one of claims 1 to 3, characterized in that The medical knowledge graph includes at least disease entities, symptom entities, examination item entities, department entities, and relationships between entities; Determine the target department entity that matches the target symptom entity from the pre-built medical knowledge graph, including: Using the relationship between entities, determine the target disease entity corresponding to the target symptom entity from the pre-built medical knowledge graph; Utilizing the relationship between the entities and the medical history information of the user to be guided, a target department entity corresponding to the target disease entity is determined from the medical knowledge graph.

5. The method according to any one of claims 1 to 3, characterized in that: Recommending the name of the medical department corresponding to the target department entity to the user to be guided includes: Identify the target hospital where the user to be guided is located, and determine the target standardized department name corresponding to the target department entity; Determine the name of the treatment department in the target hospital that corresponds to the target standardized department name using a preset department classification system, wherein the department classification system includes: a mapping relationship between each treatment department name in each hospital and each standardized department name; The name of the medical department corresponding to the target standardized department name is recommended to the user to be guided.

6. The method according to claim 4, characterized in that After recommending the name of the medical department corresponding to the target department entity to the user to be guided, the method further includes: Determining a target examination item entity corresponding to the target disease entity from the medical knowledge graph; The inspection items corresponding to the target inspection item entity are recommended to the user to be guided.

7. An intelligent diagnosis and guidance system based on knowledge graph and natural language processing, characterized by: The system comprises: an acquiring unit, configured to acquire consulting information to be input by the user to be guided, wherein the consulting information at least includes a symptom description and / or medical history information; an identification unit, configured to identify a target symptom entity from the consultation information; a determination unit, configured to determine a target department entity that matches the target symptom entity from a pre-constructed medical knowledge graph, wherein the medical knowledge graph is constructed based on medical data and natural language processing technology; The recommendation unit is used to recommend the name of the medical department corresponding to the target department entity to the user to be guided.

8. The system according to claim 7, characterized in that The identification unit includes: A first recognition module is configured to, if the consultation information is text information, use natural language processing technology to identify a target symptom entity from the consultation information; a preprocessing module, configured to perform a first preprocessing on the consultation information if the consultation information is voice information, the first preprocessing including at least noise reduction and endpoint detection; a conversion module, configured to convert the consultation information after the first preprocessing into text information; The second recognition module is used to use natural language processing technology to identify the target symptom entity from the consultation information converted into text information.

9. The system according to claim 7, wherein: The determining unit includes: Collection module, used to collect medical data; a preprocessing module, configured to perform a second preprocessing on the medical data, wherein the second preprocessing includes at least data cleaning, labeling, and desensitization; an extraction module, configured to extract medical entities and inter-entity relationships from the medical data after the second preprocessing using natural language processing technology, wherein the medical entities include at least disease entities, symptom entities, examination item entities, and department entities; A construction module is used to construct a medical knowledge graph through the medical entities and the relationships between the entities.

10. The system according to any one of claims 7 to 9, characterized in that: The medical knowledge graph includes at least disease entities, symptom entities, examination item entities, department entities, and relationships between entities; The determination unit is specifically used to: utilize the relationship between entities to determine the target disease entity corresponding to the target symptom entity from a pre-constructed medical knowledge graph; utilize the relationship between entities and the medical history information of the user to be guided to determine the target department entity corresponding to the target disease entity from the medical knowledge graph.

Images