Medical named entity interpretation method, medical named entity construction method, medical named entity interpretation system and related device

Abstract

The embodiment of the invention provides a medical named entity interpretation method, a medical named entity construction method, a medical named entity interpretation system and a medical named entity construction system and related devices. The interpretation method of the medical named entity comprises the steps of determining a selected target medical text according to a user selection operation in a display interface displaying a medical knowledge text; identifying a medical named entity in the target medical text; acquiring a context text of the medical named entity in the medical knowledge text; wherein the context text of the medical knowledge text is semantically associated with the medical knowledge text; and according to a prompt instruction constructed by the medical named entity and the context text, calling the medical interpretation large language model to indicate the medical interpretation large language model to output a target medical interpretation that the medical named entity accords with the context text. The interpretation method, the construction method, the system and the related device of the medical named entity can improve the semantic matching degree of the medical interpretation of the medical named entity and the medical knowledge text in which the medical named entity is located to a certain extent.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence technology, and in particular to a method for interpreting, constructing, system and related devices of medical named entities. Background Art

[0002] In the prior art, medical personnel read some medical knowledge texts during their work to learn medical knowledge from the medical knowledge texts.

[0003] In some cases, there may be some medical named entities in the medical knowledge text that medical personnel do not understand very well. For example, when a doctor sees a patient whose disease he is not very good at, the medical knowledge text of the relevant disease may involve medical named entities that the doctor does not understand very well.

[0004] In the prior art, medical personnel may consult some medical dictionaries to understand the meaning of medical named entities. However, the explanations in the medical dictionaries are fixed and independent of the medical knowledge texts read by medical personnel. As a result, in the prior art, the medical explanations consulted through medical dictionaries are inconsistent with the medical knowledge texts. Summary of the invention

[0005] In view of this, multiple embodiments of the present application are committed to providing a method for interpreting, constructing, system and related devices for medical named entities, which can improve the degree of semantic consistency between the medical interpretation of the medical named entities and the medical knowledge text in which the medical named entities are located to a certain extent.

[0006] One embodiment of the present application provides a method for interpreting a medical named entity, comprising: determining a selected target medical text according to a user selection operation in a display interface displaying a medical knowledge text; identifying a medical named entity in the target medical text; obtaining a context text of the medical named entity in the medical knowledge text; wherein the context text of the medical knowledge text is semantically associated with the medical knowledge text; and calling a medical interpretation large language model according to a prompt instruction constructed from the medical named entity and the context text to instruct the medical interpretation large language model to output a target medical interpretation of the medical named entity that matches the context text.

[0007] One embodiment of the present application provides a method for constructing a medical named entity interpretation system, comprising: obtaining a medical big language model base; using different prompt instruction sets to train the medical big language model base to obtain an entity extraction big model and a medical interpretation big language model; wherein the prompt instruction set includes an entity extraction instruction set for training the entity extraction big language model and a medical interpretation instruction set for training the medical interpretation big language model; the entity extraction instruction set includes a plurality of entity extraction prompt instructions; the entity extraction prompt instructions are used to train the medical big language model base to extract preliminary screening named entities from target medical texts, and provide location information of the preliminary screening named entities in the medical knowledge text to form the entity extraction big language model; the medical interpretation instruction set includes a plurality of medical interpretation prompt instructions, and the medical interpretation prompt instructions are used to train the medical big language model base to generate a target medical interpretation of the medical named entity for the input medical named entity and the context text obtained based on the location information.

[0008] One embodiment of the present application provides a medical named entity interpretation system, including: a determination module, used to determine a selected target medical text according to a user selection operation in a display interface displaying a medical knowledge text; an identification module, used to identify a medical named entity in the target medical text; a context acquisition module, used to acquire a context text of the medical named entity in the medical knowledge text; wherein the context text of the medical knowledge text is semantically associated with the medical knowledge text; and a calling module, used to call a medical interpretation large language model according to a prompt instruction constructed from the medical named entity and the context text, so as to instruct the medical interpretation large language model to output a target medical interpretation of the medical named entity that matches the context text.

[0009] One embodiment of the present application also provides a device for constructing a medical named entity interpretation system, comprising: a base acquisition module, used to acquire a medical large language model base; a training module, used to train the medical large language model base using different prompt instruction sets to obtain an entity extraction large model and a medical interpretation large language model; wherein the prompt instruction set includes an entity extraction instruction set for training the entity extraction large language model and a medical interpretation instruction set for training the medical interpretation large language model; the entity extraction instruction set includes a plurality of entity extraction prompt instructions; the entity extraction prompt instructions are used to train the medical large language model base to extract preliminary screening named entities from medical texts, and provide location information of the preliminary screening named entities in medical knowledge texts to form the entity extraction large language model; the medical interpretation instruction set includes a plurality of medical interpretation prompt instructions, and the medical interpretation prompt instructions are used to train the medical large language model base to generate a target medical interpretation of the medical named entity based on the input medical named entity and the context text obtained based on the location information.

[0010] One embodiment of the present application provides a computer device, which includes a memory and a processor, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the aforementioned method for interpreting medical named entities, or to implement the aforementioned method for constructing a system for interpreting medical named entities.

[0011] One embodiment of the present application provides a computer-readable storage medium, in which at least one computer program is stored. When the at least one computer program is executed by a processor, it can implement the aforementioned method for interpreting medical named entities, or implement the aforementioned method for constructing a system for interpreting medical named entities.

[0012] One embodiment of the present application provides a computer program product, which is used to implement the aforementioned method for interpreting medical named entities, or to implement the aforementioned method for constructing a system for interpreting medical named entities.

[0013] In the multiple implementations provided in the present application, the target medical text is determined according to the user's selection operation, and the medical named entity is identified in the text, thereby obtaining the context text where the medical named entity is located, and constructing a prompt instruction to call the medical interpretation large language model, so that the medical interpretation large language model can output a target medical interpretation that fits the context according to the context text. In this way, it is achieved that the professionalism of interpreting the medical named entity is taken into account, and the target medical interpretation is closely related to the context. In this way, the degree of semantic fit between the medical interpretation of the medical named entity and the medical knowledge text where the medical named entity is located is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A schematic diagram of a medical named entity interpretation system provided for one embodiment of the present application.

[0015] Figure 2 A schematic diagram of a display interface of a medical named entity interpretation system provided for one embodiment of the present application.

[0016] Figure 3 A schematic diagram of a display interface of a medical named entity interpretation system provided for one embodiment of the present application.

[0017] Figure 4 A flowchart of a method for interpreting medical named entities provided for one embodiment of the present application.

[0018] Figure 5 A flowchart of a method for constructing a medical named entity interpretation system provided for one embodiment of the present application.

[0019] Figure 6 A schematic diagram of the modules of a medical named entity interpretation system provided for one embodiment of the present application.

[0020] Figure 7 A schematic diagram of modules of a construction device of a medical named entity interpretation system provided for one embodiment of the present application.

[0021] Figure 8 A schematic diagram of a computer device provided for one embodiment of the present application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments.

[0023] In the description of the embodiments of the present application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0024] In the related art, in the field of interpretation of medical named entities, traditional interpretation methods usually rely on static medical dictionaries to provide term interpretations.

[0025] Specifically, when medical personnel encounter unfamiliar medical named entities in medical knowledge texts, they often need to use medical dictionaries to query the fixed definition of the named entity. However, this method has certain limitations: the explanations in medical dictionaries are predefined and lack adaptability to specific contexts. Especially when the same medical term has different meanings in different contexts, the explanations in medical dictionaries are often difficult to fit the semantics well. In addition, knowledge in the medical field is updated relatively quickly, and medical dictionaries cannot reflect the latest medical developments in a timely manner. This makes it difficult for doctors to obtain high-quality and context-related explanations when faced with new terms or diseases they have never encountered, which is not conducive to doctors' efficient understanding of text content.

[0026] With the development of natural language processing technology, researchers have applied large language models (LLMs) to the interpretation of medical named entities. Large language models have absorbed a lot of knowledge during the training process and have good natural language understanding and content generation capabilities. However, when performing entity interpretation tasks, existing large language models may still have the phenomenon of "AI hallucination", that is, generating interpretations that are inconsistent with the actual semantics or deviate from the context. This situation will affect the accuracy of medical interpretations and cause trouble for doctors. For example, after receiving an entity interpretation task, the large language model may directly use the provided entity as a medical named entity and generate an incorrect medical interpretation that looks correct. Alternatively, the large language model may also understand the medical named entity as a non-medical field and generate an incorrect interpretation that looks correct. As a result, there is still a problem that the interpretation of medical named entities is not consistent with the medical knowledge text. Even the interpretation obtained is wrong content.

[0027] In summary, it is necessary to provide a technical solution that can improve the semantic fit between the medical interpretation of medical named entities and the medical knowledge text in which the medical named entities are located.

[0028] See also Figure 1. In multiple embodiments provided in the present application, the interpretation method of medical named entities can be applied to the interpretation system of medical named entities. The interpretation system may include a client and a server. The client may be an electronic device with network access capability. Specifically, for example, the client may be a desktop computer, a laptop computer, a tablet computer, a smart phone, a digital assistant, a smart wearable device, a shopping guide terminal, a smart speaker, a microphone and other devices. Among them, smart wearable devices include but are not limited to smart bracelets, smart watches, smart glasses, smart helmets, smart necklaces, etc. Alternatively, the client may also be a software application that can run in the above-mentioned electronic devices. The server may be an electronic device with certain computing and processing capabilities, and has hardware resources such as a network communication module, a processor and a memory. The server may also be a distributed server, including multiple processors, memories, network communication modules, etc., which work together to realize various functions. Alternatively, the server may also be a server cluster formed by several servers, which has higher computing and data processing capabilities. With the development of science and technology, the server can also be implemented by new forms of technical means, such as a new type of "server" based on quantum computing.

[0029] One embodiment of the present application provides an application scenario example of a method for interpreting medical named entities. The interpretation system can interpret medical named entities in medical knowledge texts and ultimately present the target medical interpretation to the user. In this scenario example, taking a doctor reading medical literature as an example, the interpretation system helps the doctor understand unfamiliar medical terms in the text and provides medical interpretations that fit the context.

[0030] like Figure 2 As shown. For example, a doctor is reading a medical paper on liver disease treatment on a client application and encounters the following paragraph: "In patients with non-alcoholic steatohepatitis (NASH), treatment with obeticholic acid has shown good results in reducing liver fibrosis. However, some patients have experienced adverse reactions such as itching." The doctor is not familiar with the terms "non-alcoholic steatohepatitis", "obeticholic acid", and "itching". At this point, the doctor selects the text by marking the words in the display interface. Based on this, the interpretation system determines the text selected by the user as the target medical text.

[0031] The interpretation system can identify medical named entities in the target medical text. Specifically, the interpretation system can construct prompt instructions based on the target medical text to instruct the entity extraction model to extract the initial screening named entities from the target medical text. The prompt instructions may include role settings, task definitions, task requirements, and target medical texts. For example, the prompt instructions may be: "You are an excellent expert in medical-related term recognition (role setting). Your job is to extract the corresponding medical-related entities from the target medical text (task definition). Extract entities from the text without making any changes. The entities may be discontinuous. When extracting entities, give the initial location of the entity (task requirements): '[target medical text]'." After receiving the prompt instructions, the entity extraction model analyzes the target medical text, extracts the initial screening named entities, such as "non-alcoholic steatohepatitis", "obeticholic acid", "itch", etc., and provides the location information of these entities in the medical knowledge text.

[0032] The interpretation system matches the initial screening named entities with the specified medical named entity library to obtain the medical named entities. The medical named entity library is a pre-built database that contains named entities in the medical field and their corresponding medical definition texts. Through matching, the interpretation system confirms that these initial screening named entities are all medical named entities.

[0033] Furthermore, the interpretation system can use the location information provided by the entity extraction model to obtain the context text of the medical named entities from the medical knowledge text. In this scenario example, the context text is the entire text containing these medical named entities, that is, the text segment initially selected by the doctor.

[0034] In this scenario example, the interpretation system can construct prompt instructions based on medical named entities, medical definition texts and context texts, and call the medical interpretation large language model. The prompt instructions may include role settings, task definitions, task requirements, RAG requirements, corresponding medical named entities and medical definition texts. For example, the prompt instruction may be: "You are an excellent expert in the interpretation of medical-related terms (role setting). Your job is to give the meaning of the medical entity words in this sentence based on the medical entity interpretation and related context (task definition). Ensure the completeness of the reply and do not omit any medical entity words. Please strictly follow the paradigm and explain the meaning of the entity words in this sentence in combination with the context when replying (task requirements), [context text]' (RAG requirements), medical named entity: 'obeticholic acid', medical definition text: 'obeticholic acid is a synthetic bile acid analog used to treat certain liver diseases.'".

[0035] After receiving the prompt instruction, the medical explanation big language model generated the following target medical explanation: "Obetocholic acid is a drug used to treat liver diseases, especially in patients with non-alcoholic steatohepatitis (NASH), which can effectively reduce liver fibrosis. In the above case, obeticholic acid showed good therapeutic effects, but attention should be paid to possible adverse reactions such as itching."

[0036] The interpretation system receives the target medical interpretation generated by the medical interpretation big language model. Since the display interface includes a medical knowledge text display area and a medical interpretation display area side by side, the interpretation system displays the target medical interpretation in the area near "obeticholic acid" in the medical interpretation display area. In this way, doctors can easily view the corresponding medical interpretation while reading the medical knowledge text, which improves the reading experience and comprehension efficiency.

[0037] When the doctor selects a different medical named entity, the medical explanation display area will be dynamically updated to the corresponding target medical explanation. Figure 3 As shown in the figure, for "pruritus", the interpretation system may generate: "Pruritus is a skin sensory abnormality that is common in many diseases and may be a manifestation of adverse drug reactions. In the above case, the patient developed pruritus after using obeticholic acid and needs to be closely observed and treated accordingly."

[0038] In this scenario example, the interpretation system can achieve accurate recognition of medical named entities and professional interpretation that fits the context by calling the entity extraction big model and medical interpretation big language model trained based on the same medical big language model base. The medical big language model base is trained using different prompt instruction sets to form the entity extraction big model and the medical interpretation big language model, respectively, ensuring the consistency of the two models in medical knowledge and language understanding, and improving the overall performance and reliability of the system.

[0039] In addition, the designated medical named entity library records the medical definition text of the medical named entity, which improves the professionalism of the target medical explanation. By combining the medical definition text and the context text, the target medical explanation generated by the medical explanation large language model is both accurate and targeted, helping doctors to understand the medical knowledge text more deeply and improve work efficiency.

[0040] In an application scenario example, the medical named entity library can be a pre-built database that contains named entities in the medical field and their corresponding medical definition texts. The construction process of the medical named entity library is as follows.

[0041] The medical big language model is used to extract medical named entities and medical definition texts from the designated medical knowledge base to form a first medical entity definition set. For example, the designated medical knowledge base may include medical papers, clinical guidelines, and medical textbooks. The medical big language model can extract medical definition texts of medical named entities such as "non-alcoholic steatohepatitis" and "obeticholic acid" from the designated medical knowledge base.

[0042] Furthermore, medical named entities can be obtained from a designated medical knowledge base using entity extraction technology, and supplemented by a noun term dictionary. The medical named entities in the noun term dictionary and the medical named entities extracted from the designated medical knowledge base are deduplicated to obtain a list of medical named entities. For example, terms such as "itching" and "liver fibrosis" are collected. Then, the medical definition text corresponding to the medical named entity can be obtained through an open source term interpretation library (such as a medical encyclopedia, professional dictionary, etc.) to form a second medical definition set. The first medical definition set and the second medical definition set can be integrated into one medical definition set.

[0043] Provide the medical definition set for manual review to improve the accuracy of the data in the medical definition set. The reviewed medical named entities and medical definition texts can be stored in a designated medical knowledge base. Furthermore, synonym expansion can be performed for the medical named entities in the designated medical knowledge base to enrich the specific expressions of the medical named entities. Furthermore, the medical named entities in the designated medical knowledge base can be translated from the original language into another language. For example, medical named entities originally in Chinese are translated into English. In this way, the content of the designated medical knowledge base can be enriched to a greater extent.

[0044] See also Figure 4 The embodiment of the present application provides a method for interpreting a medical named entity. The method for interpreting a medical named entity can be applied to an interpretation system for a medical named entity. The method for interpreting a medical named entity can include the following steps.

[0045] Step S110: Determine a selected target medical text according to a user selection operation in a display interface displaying medical knowledge texts.

[0046] Step S120: Identify medical named entities in the target medical text.

[0047] Step S130: Acquire the context text of the medical named entity in the medical knowledge text; wherein the context text of the medical knowledge text is semantically associated with the medical knowledge text.

[0048] Step S140: According to the prompt instruction constructed by the medical named entity and the context text, the medical interpretation large language model is called to instruct the medical interpretation large language model to output the target medical interpretation of the medical named entity that matches the context text.

[0049] In this embodiment, the medical knowledge text may be text information used to introduce knowledge in the medical field. Specifically, the medical knowledge text may include, but is not limited to, various types of medical information such as medical papers, clinical guidelines, case reports, and medical textbooks. The display interface may be an application program interface on the client, which is used to present the medical knowledge text for users to read. For example, when medical personnel (such as doctors, nurses, or medical students) read medical knowledge texts on the client, they may encounter unfamiliar medical named entities. At this point, the medical personnel can select operations by clicking, marking words, highlighting, etc., and select the text content of interest. The interpretation system can use the text selected by the medical personnel as the target medical text.

[0050] In this embodiment, the interpretation system can identify medical named entities in the target medical text. Medical named entities can refer to entity names with specific meanings in the medical field. For example, medical named entities can include but are not limited to disease names, drug names, anatomical structures, physiological processes, medical test items, etc. The interpretation system can use Named Entity Recognition (NER) technology to analyze the target medical text and automatically extract the medical named entities therein. For example, the target medical text is "The patient was admitted to the hospital due to acute myocardial infarction and was treated with aspirin and clopidogrel." Through NER technology, the interpretation system can identify medical named entities such as "acute myocardial infarction", "aspirin", and "clopidogrel".

[0051] In this embodiment, the context text may refer to the contextual information associated with the medical named entity. Specifically, the context text may include sentences, paragraphs or chapters in the medical knowledge text that are adjacent to the medical named entity. In this way, the context text is semantically associated with the medical knowledge text, which helps to understand the meaning of the medical named entity in a specific context. For example, the context text may be an entire sentence or related paragraph containing the medical named entity, such as "The patient was admitted to the hospital due to acute myocardial infarction and was treated with aspirin and clopidogrel." The context text provides the usage context of the medical named entity and helps to eliminate ambiguity.

[0052] In this embodiment, the interpretation system can construct prompt instructions based on medical named entities and context texts, and call the medical interpretation big language model. The prompt instruction can have a preset format, which is used to combine medical named entities and context texts to guide the medical interpretation big language model to generate medical interpretations that fit the context. The medical interpretation big language model can be a large language model that has been trained with a large amount of medical data and has the ability to understand medical semantics and generate professional medical content. For example, the prompt instruction may be: "Please explain the meaning of the medical term 'acute myocardial infarction' based on the following context: 'The patient was admitted to the hospital due to acute myocardial infarction and was treated with aspirin and clopidogrel.'" In this way, the medical interpretation big language model can combine the context to generate medical interpretations relevant to the current context.

[0053] In this embodiment, after receiving the prompt instruction, the medical explanation big language model can output the target medical explanation that fits the context. For example, the medical explanation big language model may generate the following explanation: "Acute myocardial infarction refers to a serious cardiovascular disease in which the coronary arteries of the heart are suddenly blocked, leading to myocardial ischemia and necrosis. In the above case, the patient was admitted to the hospital due to acute myocardial infarction, and the treatment plan included the use of aspirin and clopidogrel to inhibit platelet aggregation and prevent thrombosis."

[0054] In this embodiment, the interpretation system realizes the interpretation of medical named entities combined with contextual text, which enhances the accuracy and practicality of medical interpretation. Compared with the static interpretation provided by traditional medical dictionaries, this method can provide dynamic interpretation that is more in line with the actual context, helping medical personnel to understand medical knowledge text more deeply.

[0055] In some embodiments, the interpretation method can be applied to a variety of medical scenarios, such as electronic medical record systems, medical literature retrieval tools, clinical decision support systems, etc. By improving the semantic fit between the interpretation of medical named entities and medical knowledge texts, it can help medical personnel improve work efficiency and reduce misunderstandings and errors.

[0056] In some embodiments, the display interface includes a medical knowledge text display area and a medical explanation display area side by side. The explanation system can receive a target medical explanation of the medical explanation large language model corresponding to the medical named entity; and display the target medical explanation in an area of ​​the medical explanation display area adjacent to the medical named entity.

[0057] In this embodiment, the display interface may include a medical knowledge text display area and a medical interpretation display area side by side. In this way, the user can view the medical knowledge text and the corresponding medical named entity interpretation on the same interface at the same time, improving the reading experience and understanding efficiency. Specifically, the interpretation system can display the target medical interpretation generated by the medical interpretation large language model in the medical interpretation display area after receiving it. The position of the medical interpretation display area is adjacent to the medical knowledge text display area, which is convenient for users to quickly obtain relevant interpretations of medical named entities while reading the medical knowledge text.

[0058] For example, when medical staff read "The patient was admitted to the hospital due to acute myocardial infarction and was given aspirin and clopidogrel" in the medical knowledge text display area, they select "Unfamiliar medical named entity" - "clopidogrel". The explanation system can call the medical explanation big language model to generate a target medical explanation about "clopidogrel", for example: "Clopidogrel is an antiplatelet aggregation drug used to prevent thrombosis and is commonly used in the treatment of patients with coronary heart disease, myocardial infarction, etc." Then, the explanation system displays the target medical explanation in the area near "clopidogrel" in the medical explanation display area.

[0059] In some embodiments, when the user selects a different medical named entity, the medical explanation display area will be dynamically updated to the corresponding target medical explanation, ensuring that the user always obtains information related to the current reading content, bringing convenience to the user.

[0060] In some embodiments, the interpretation system can construct prompt instructions based on the target medical text to instruct the entity extraction model to extract preliminary screening named entities from the target medical text; match the preliminary screening named entities in a specified medical named entity library to obtain the medical named entities in the preliminary screening named entities.

[0061] In this embodiment, the entity extraction large model can be a large language model trained with a large amount of medical data, which has the ability to identify possible named entities from the text. The prompt instruction can adopt a preset format, take the target medical text as input, and guide the entity extraction large model to perform the entity recognition task. For example, the prompt instruction can be: "Please extract all named entities from the following text: '[target medical text]'." After receiving the prompt instruction, the entity extraction large model will analyze the target medical text and extract the initial screening named entities. These initial screening named entities may include terms in the medical field, drug names, disease names, etc., as well as other possible named entities.

[0062] In this embodiment, the medical named entity library can be a pre-constructed database containing a set of named entities in the medical field. For example, such as disease names, drug names, anatomical structures, physiological processes, etc. Through matching, named entities in non-medical fields can be filtered out to ensure that the objects of subsequent processing are medical named entities. For example, the initial screening named entities may include "acute myocardial infarction", "aspirin", "blood pressure", "patient", etc. By matching with the medical named entity library, it can be confirmed that "acute myocardial infarction" and "aspirin" are medical named entities, while "blood pressure" may be regarded as a general term, and "patient" may be excluded. The interpretation system can accurately identify medical named entities from the target medical text, providing an accurate data basis for calling the medical interpretation large language model to generate the target medical interpretation, thereby improving the accuracy of the target medical interpretation to a certain extent.

[0063] In this embodiment, the interpretation system utilizes the recognition capability of the entity extraction large model and the professionalism of the medical named entity library to improve the accuracy and reliability of medical named entity recognition.

[0064] In some embodiments, the entity extraction model also provides the location information of the initial screening named entity in the medical knowledge text corresponding to the initial screening named entity. The interpretation system can obtain the context text of the initial screening named entity in the medical knowledge text according to the location information.

[0065] In this embodiment, the entity extraction model outputs the position information of these pre-screening named entities in the medical knowledge text while extracting the pre-screening named entities from the target medical text. The position information may include the starting position and the ending position of each pre-screening named entity in the text, or other forms of position information identification. The interpretation system can obtain the context text of the pre-screening named entity in the medical knowledge text according to the position information. It can be understood that the interpretation system can accurately locate the position of the pre-screening named entity in the medical knowledge text according to the position information of the pre-screening named entity, and then extract the context text related to the pre-screening named entity. The context text can be a whole sentence containing the medical named entity, the paragraph in which it is located, or several words before and after it. For example, the pre-screening named entity can be "clopidogrel", and the position information indicates that its character position in the medical knowledge text is the 25th to 29th character. The interpretation system can extract sentences containing "clopidogrel" from the medical knowledge text according to the position information, such as "The patient was admitted to the hospital for acute myocardial infarction and given aspirin and clopidogrel treatment." as context text. In this way, the relevance of the acquired context text to the medical named entities can be improved, providing accurate context information for the subsequent construction of prompt instructions to call the medical interpretation large language model.

[0066] In some embodiments, the medical named entity in the designated medical named entity library has a medical definition text recorded therein. The interpretation system can construct a prompt instruction based on the medical named entity, the medical definition text of the medical named entity, and the context text to instruct the medical interpretation large language model to output a target medical interpretation of the medical named entity that matches the context text based on the medical definition text.

[0067] In this embodiment, the medical named entity library can be a pre-constructed database, which contains medical named entities and medical definition texts corresponding to the medical named entities. The medical definition text can be a standard interpretation or definition of the medical named entity. The interpretation system can add the medical definition text to the prompt instruction so that the medical interpretation large language model can output an interpretation that fits the specific context based on the medical definition text and combined with the context text when generating the target medical interpretation. For example, after obtaining the medical named entity "clopidogrel", its corresponding medical definition text is extracted from the medical named entity library, such as "clopidogrel is an antiplatelet aggregation drug used to prevent thrombosis". Subsequently, the interpretation system constructs a prompt instruction containing the medical named entity, its medical definition text and context text. The prompt instruction can be constructed as: "Please explain the meaning of the medical term 'clopidogrel' based on the following medical definition and context. Medical definition: 'Clopidogrel is an antiplatelet aggregation drug used to prevent thrombosis'. Context: 'The patient was admitted to the hospital for acute myocardial infarction and was treated with aspirin and clopidogrel'." By providing medical definition text, the medical explanation big language model can use the medical definition text as a basis when generating target medical explanations, avoiding possible understanding bias or errors. At the same time, combined with the context text, the model can output an explanation that is more suitable for the current context. For example, the medical explanation big language model may generate the following target medical explanation: "Clopidogrel is an antiplatelet aggregation drug used to prevent thrombosis. In the above case, the patient was admitted to the hospital due to acute myocardial infarction. The use of clopidogrel helps prevent the aggravation of thrombosis and reduce the risk of recurring cardiovascular events."

[0068] In this embodiment, the professionalism of the target medical interpretation is improved by using the medical definition text. Furthermore, the target medical interpretation is made more relevant and targeted by combining the context text, helping medical personnel to understand the medical knowledge text more deeply.

[0069] See also Figure 5 The present application also provides a method for constructing a medical named entity interpretation system. The construction method may include the following steps.

[0070] Step S210: Acquire the medical large language model base.

[0071] Step S220: The medical big language model base is trained with different prompt instruction sets to obtain an entity extraction big model and a medical interpretation big language model; wherein the prompt instruction set includes an entity extraction instruction set for training the entity extraction big language model and a medical interpretation instruction set for training the medical interpretation big language model; the entity extraction instruction set includes a plurality of entity extraction prompt instructions; the entity extraction prompt instructions are used to train the medical big language model base to extract preliminary screening named entities from the target medical text, and provide the position information of the preliminary screening named entities in the medical knowledge text to form the entity extraction big language model; the medical interpretation instruction set includes a plurality of medical interpretation prompt instructions, and the medical interpretation prompt instructions are used to train the medical big language model base to generate a target medical interpretation of the medical named entity based on the input medical named entity and the context text obtained based on the position information.

[0072] In this embodiment, the construction device may first obtain a medical big language model base. The medical big language model base is a pre-trained model that has been trained with a large amount of medical data and has the ability to understand medical language and generate professional medical content. Furthermore, the construction device may use different prompt instruction sets to train the medical big language model base to obtain an entity extraction big model and a medical interpretation big language model. Specifically, the prompt instruction set may include an entity extraction instruction set for training an entity extraction big model and a medical interpretation instruction set for training a medical interpretation big language model.

[0073] In this embodiment, the entity extraction instruction set includes multiple entity extraction prompt instructions. The entity extraction prompt instruction can be used to train the medical large language model base, instructing the medical large language model base to extract preliminary screening named entities from the target medical text, and provide the location information of these preliminary screening named entities in the medical knowledge text, thereby forming an entity extraction large model. The entity extraction prompt instruction can adopt a preset format to guide the model to perform entity recognition tasks. For example, the entity extraction prompt instruction can be: "Please extract all named entities from the following text and mark their location information in the text: '[medical text]'.".

[0074] In this embodiment, the medical interpretation instruction set includes multiple medical interpretation prompt instructions. The medical interpretation prompt instruction can be used to train the medical big language model base to generate the target medical interpretation of the medical named entity based on the input medical named entity and the context text obtained according to the location information, thereby forming a medical interpretation big language model. For example, the medical interpretation prompt instruction can be constructed as: "Please generate the medical interpretation of the entity based on the following medical named entity and context. Medical named entity: '[medical named entity]'; context: '[context text]'.". Through the above training process, the construction of an entity extraction big model specifically for entity extraction and a medical interpretation big language model for medical interpretation is realized. The entity extraction big model can efficiently identify medical named entities from medical texts and provide corresponding location information, and the medical interpretation big language model can combine the medical definition text and context text of the medical named entity to generate an accurate and contextual target medical interpretation.

[0075] In this implementation, different prompt instruction sets are used to conduct specialized training on the same medical large language model base, which improves the accuracy and reliability of medical named entity recognition and interpretation. By building a dedicated entity extraction large model and a medical interpretation large language model, the interpretation system can better meet the needs of medical personnel for medical named entity interpretation when reading medical knowledge texts, thereby improving work efficiency.

[0076] Furthermore, the use of the same model base improves the consistency of the entity extraction big language model and the medical explanation big language model in medical knowledge and language understanding, making the entity extraction and medical interpretation processes highly coordinated. In this way, the medical named entities identified by the entity extraction big model are consistent with the input used by the medical interpretation big language model to generate explanations in terms of semantics and context, reducing information bias and error accumulation. Furthermore, training the entity extraction big language model and the medical interpretation big language model based on the same model base can make full use of the existing medical field knowledge in the pre-trained model and reduce the workload of repeated training. In addition, the trained entity extraction big language model and the medical interpretation big language model can better understand complex medical terms, improving the accuracy of medical named entity recognition and interpretation. This consistency helps to reduce the occurrence of the "AI hallucination" phenomenon and ensure that the generated target medical explanation is both professional and fits the context.

[0077] See also Figure 6 One embodiment of the present application also provides a medical named entity interpretation system. The interpretation system includes: a determination module, a recognition module, a context acquisition module and a call module.

[0078] The determination module is used to determine the selected target medical text according to the user selection operation in the display interface displaying the medical knowledge text.

[0079] A recognition module, used for identifying medical named entities in the target medical text;

[0080] A context acquisition module, used to acquire the context text of the medical named entity in the medical knowledge text; wherein the context text of the medical knowledge text is semantically associated with the medical knowledge text;

[0081] A calling module is used to call a medical interpretation large language model according to a prompt instruction constructed based on the medical named entity and the context text, so as to instruct the medical interpretation large language model to output a target medical interpretation of the medical named entity that matches the context text.

[0082] In this embodiment, the specific functions and effects achieved by the medical named entity interpretation system can be explained with reference to other embodiments of the present application and will not be repeated here.

[0083] See also Figure 7 One embodiment of the present application also provides a device for constructing a medical named entity interpretation system. The device comprises: a base acquisition module and a training module.

[0084] The base acquisition module is used to obtain the medical large language model base.

[0085] A training module is used to train the medical large language model base using different prompt instruction sets to obtain an entity extraction large model and a medical interpretation large language model; wherein the prompt instruction set includes an entity extraction instruction set for training the entity extraction large language model and a medical interpretation instruction set for training the medical interpretation large language model; the entity extraction instruction set includes multiple entity extraction prompt instructions; the entity extraction prompt instructions are used to train the medical large language model base to extract preliminary screening named entities from medical texts, and provide location information of the preliminary screening named entities in the medical knowledge text to form the entity extraction large language model; the medical interpretation instruction set includes multiple medical interpretation prompt instructions, and the medical interpretation prompt instructions are used to train the medical large language model base to generate a target medical interpretation of the medical named entity based on the input medical named entity and the context text obtained based on the location information.

[0086] In this embodiment, the specific functions and effects achieved by the construction device of the medical named entity interpretation system can be explained with reference to other embodiments of the present application and will not be repeated here.

[0087] The embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor implements the method for interpreting medical named entities as described above, or implements the method for constructing a system for interpreting medical named entities as described above.

[0088] The embodiment of the present application also provides a computer program product comprising instructions, which, when executed by a processor, implements the aforementioned method for interpreting medical named entities, or implements the aforementioned method for constructing a system for interpreting medical named entities.

[0089] See also Figure 8 . The present description embodiment may provide a computer device, the computer device comprising: a memory, and one or more processors communicatively connected to the memory; the memory stores instructions executable by the one or more processors, the instructions are executed by the one or more processors to enable the one or more processors to implement the aforementioned method for interpreting medical named entities, or to implement the aforementioned method for constructing a system for interpreting medical named entities.

[0090] In some embodiments, the computer device may include a processor, a non-volatile storage medium, an internal memory, a communication interface, a display device, and an input device connected by a system bus. The non-volatile storage medium may store an operating system and related computer programs.

[0091] The user information or user account information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, etc.) involved in multiple implementation methods of the present application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0092] It should be understood that the specific examples in this article are only intended to help those skilled in the art to better understand the embodiments of the present application, rather than to limit the scope of the present invention.

[0093] It can be understood that in the various implementations of the present application, the size of the serial number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the implementation methods of the present application.

[0094] It can be understood that the various embodiments described in this application can be implemented individually or in combination, and the embodiments of this application are not limited to this.

[0095] Unless otherwise stated, all technical and scientific terms used in the embodiments of the present application have the same meaning as those generally understood by those skilled in the art of the technical field of the present application. The terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the scope of the present application. The term "and / or" used in the present application includes any and all combinations of one or more related listed items. The singular forms of "a kind of", "above" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0096] It can be understood that the processor of the embodiment of the present application can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method implementation can be completed by the hardware integrated logic circuit or software instructions in the processor. The above processor can be a general processor, a digital signal processor (DigitalSignal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiment of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to perform, or the hardware and software modules in the decoding processor are combined and executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0097] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (programmable ROM, PROM), an erasable programmable read-only memory (erasable PROM, EPROM), an electrically erasable programmable read-only memory (EEPROM) or a flash memory. The volatile memory may be a random access memory (RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0098] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0099] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method implementation methods and will not be repeated here.

[0100] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device implementation described above is only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0101] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0102] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0103] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0104] The above is only a specific implementation of the present application, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for interpreting medical named entities, characterized in that: include: Determining a selected target medical text according to a user selection operation in a display interface displaying medical knowledge texts; identifying medical named entities in the target medical text; Acquire the context text of the medical named entity in the medical knowledge text; wherein the context text of the medical knowledge text is semantically associated with the medical knowledge text; According to the prompt instruction constructed by the medical named entity and the context text, the medical interpretation large language model is called to instruct the medical interpretation large language model to output the target medical interpretation of the medical named entity that matches the context text.

2. The method according to claim 1, characterized in that The display interface includes a medical knowledge text display area and a medical explanation display area arranged side by side; The method further comprises: Receiving a target medical explanation of the medical explanation big language model corresponding to the medical named entity; The target medical explanation is displayed in an area adjacent to the medical named entity in the medical explanation display area.

3. The method according to claim 1, characterized in that The step of identifying medical named entities in the target medical text comprises: Constructing prompt instructions according to the target medical text to instruct the entity extraction large model to extract preliminary screening named entities from the target medical text; The initially screened named entities are matched in a designated medical named entity library to obtain the medical named entities in the initially screened named entities.

4. The method according to claim 3, characterized in that The entity extraction model also provides location information of the initial screening named entity in the medical knowledge text corresponding to the initial screening named entity; The step of obtaining the context text of the medical named entity in the medical knowledge text comprises: According to the location information, the context text of the initially screened named entity is obtained in the medical knowledge text.

5. The method according to claim 3, characterized in that: The designated medical named entity library contains a medical definition text corresponding to the medical named entity record; The step of constructing a prompt instruction according to the medical named entity and the context text to instruct the medical explanation large language model to output a target medical explanation of the medical named entity includes: A prompt instruction is constructed according to the medical named entity, the medical definition text of the medical named entity and the context text to instruct the medical interpretation large language model to output a target medical interpretation of the medical named entity that matches the context text based on the medical definition text.

6. A method for constructing a medical named entity interpretation system, characterized in that: include: Get the medical big language model base; The medical big language model base is trained with different prompt instruction sets to obtain an entity extraction big model and a medical interpretation big language model; wherein the prompt instruction set includes an entity extraction instruction set for training the entity extraction big language model and a medical interpretation instruction set for training the medical interpretation big language model; the entity extraction instruction set includes a plurality of entity extraction prompt instructions; the entity extraction prompt instructions are used to train the medical big language model base to extract preliminary screening named entities from the target medical text, and provide the position information of the preliminary screening named entities in the medical knowledge text to form the entity extraction big language model; the medical interpretation instruction set includes a plurality of medical interpretation prompt instructions, and the medical interpretation prompt instructions are used to train the medical big language model base to generate a target medical interpretation of the medical named entity based on the input medical named entity and the context text obtained based on the position information.

7. A medical named entity interpretation system, characterized in that: include: A determination module, used to determine a selected target medical text according to a user selection operation in a display interface displaying medical knowledge texts; A recognition module, used for identifying medical named entities in the target medical text; A context acquisition module, used to acquire the context text of the medical named entity in the medical knowledge text; wherein the context text of the medical knowledge text is semantically associated with the medical knowledge text; A calling module is used to call a medical interpretation large language model according to a prompt instruction constructed based on the medical named entity and the context text, so as to instruct the medical interpretation large language model to output a target medical interpretation of the medical named entity that matches the context text.

8. A device for constructing a medical named entity interpretation system, characterized in that: include: A base acquisition module is used to acquire the medical large language model base; A training module is used to train the medical large language model base using different prompt instruction sets to obtain an entity extraction large model and a medical interpretation large language model; wherein the prompt instruction set includes an entity extraction instruction set for training the entity extraction large language model and a medical interpretation instruction set for training the medical interpretation large language model; the entity extraction instruction set includes multiple entity extraction prompt instructions; the entity extraction prompt instructions are used to train the medical large language model base to extract preliminary screening named entities from medical texts, and provide location information of the preliminary screening named entities in the medical knowledge text to form the entity extraction large language model; the medical interpretation instruction set includes multiple medical interpretation prompt instructions, and the medical interpretation prompt instructions are used to train the medical large language model base to generate a target medical interpretation of the medical named entity based on the input medical named entity and the context text obtained based on the location information.

9. A computer device, characterized in that: The computer device includes a memory and a processor, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the method for interpreting medical named entities as described in any one of claims 1 to 5, or to implement the method for constructing a system for interpreting medical named entities as described in claim 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one computer program, which, when executed by a processor, can implement the method for interpreting a medical named entity as described in any one of claims 1 to 5, or implement the method for constructing a medical named entity interpretation system as described in claim 6.

11. A computer program product, characterized in that The computer program product is used to implement the method for interpreting a medical named entity as described in any one of claims 1 to 5, or to implement the method for constructing a system for interpreting a medical named entity as described in claim 6.

Images