Medical term normalization method, computer equipment and storage medium

Abstract

The invention provides a medical term normalization method, computer equipment and a storage medium, and relates to the technical field of computers.The method comprises the steps that input new medical terms are obtained, a pre-trained medical vectorization model is adopted, the new medical terms are vectorized, semantic vectors of the new medical terms are obtained, and the semantic vectors of the new medical terms are obtained; and according to the semantic vector of the new medical term and the semantic vector of each medical term in a preset medical term dictionary, calculating the semantic similarity between the new medical term and each medical term, and if the semantic similarity between the new medical term and a target medical term in each medical term is greater than or equal to a preset semantic similarity threshold, determining that the target medical term is the target medical term. And if yes, determining representative medical terms of the target term category corresponding to the target medical terms as target normalized medical terms corresponding to the new medical terms. According to the method, manual operation is not needed, the new medical terms are normalized automatically, and the normalization efficiency of the medical terms is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a medical term normalization method, a computer device, and a storage medium. Background Art

[0002] In the field of modern medicine, with the continuous advancement of science and technology, the number of medical literature and research reports has exploded. These documents and research reports use a large number of medical terms, which may be different in different documents, leading to obstacles in information exchange and knowledge sharing. In order to overcome this problem, medical terminology normalization technology came into being.

[0003] At present, all medical terms are mainly extracted manually from a large amount of medical literature and research, and are manually classified and sorted, the extracted medical terms are standardized, medical terms in different forms of expression are unified into standardized medical terms, and the standardized medical terms are stored in a terminology database.

[0004] However, medical terminology is rich and diverse and is constantly updated. Manual classification has limited coverage, and established rules are often targeted at terminology in specific fields and cannot cover all situations. In addition, the maintenance cost of the terminology database is high, and it is difficult to maintain the rules. New and undefined terms cannot be effectively handled, resulting in poor flexibility. Summary of the invention

[0005] The purpose of this application is to provide a medical terminology normalization method, computer equipment and storage medium to address the deficiencies in the above-mentioned prior art, which can automatically normalize new medical terms without manual operation and improve the efficiency of medical terminology normalization.

[0006] To achieve the above purpose, the technical solution adopted in the embodiment of the present application is as follows:

[0007] In a first aspect, an embodiment of the present application provides a medical term normalization method, the method comprising:

[0008] Get input of new medical terms;

[0009] Using a pre-trained medical vectorization model to vectorize the new medical term to obtain a semantic vector of the new medical term;

[0010] Calculating the semantic similarity between the new medical term and each medical term in a preset medical term dictionary based on the semantic vector of the new medical term;

[0011] If the semantic similarity between the new medical term and the target medical term in the medical terms is greater than or equal to a preset semantic similarity threshold, a representative medical term of the target term category corresponding to the target medical term is determined as the target normalized medical term corresponding to the new medical term.

[0012] Optionally, the adopting a pre-trained medical vectorization model to vectorize the new medical term to obtain a semantic vector of the new medical term includes:

[0013] The new medical term is vectorized using the medical vectorization model to map the new medical term into a vector space corresponding to the medical vectorization model, thereby obtaining a dense semantic vector of the new medical term. Multiple elements of the dense semantic vector of the new medical term respectively represent the semantic features of the new medical term in multiple feature dimensions.

[0014] Optionally, before the medical vectorization model is used to vectorize the new medical term to map the new medical term to a vector space corresponding to the medical vectorization model to obtain a dense semantic vector of the new medical term, the method further includes:

[0015] Sample medical terms are used to train an initial large language model with multiple feature dimensions to obtain the medical vectorization model.

[0016] Optionally, before calculating the semantic similarity between the new medical term and each medical term in a preset medical term dictionary based on the semantic vector of the new medical term, the method further includes:

[0017] Adjusting the length of the semantic vector of the new medical term to a preset vector length, wherein the length of the semantic vector of each medical term in the preset medical term dictionary is the preset vector length;

[0018] The step of calculating the semantic similarity between the new medical term and each medical term in a preset medical term dictionary according to the semantic vector of the new medical term includes:

[0019] The semantic similarity between the new medical term and the medical terms is calculated based on the length-adjusted semantic vector of the new medical term and the semantic vectors of the medical terms.

[0020] Optionally, before calculating the semantic similarity between the new medical term and each medical term in a preset medical term dictionary based on the semantic vector of the new medical term, the method further includes:

[0021] Get multiple medical terms from a pre-set medical corpus;

[0022] Using the medical vectorization model, vectorize each medical term to obtain a semantic vector of each medical term;

[0023] Determining the semantic similarity between every two medical terms in the medical terms according to the semantic vectors of the medical terms;

[0024] Performing cluster analysis on the medical terms according to the semantic similarity between each two medical terms to obtain term categories corresponding to the medical terms;

[0025] According to the semantic similarity of all medical terms under each term category, determine the medical term with the highest similarity to other medical terms as the representative medical term of each term category;

[0026] The preset medical term dictionary is constructed according to the semantic vectors of the medical terms, the term categories corresponding to the medical terms, and the representative medical terms of each term category.

[0027] Optionally, constructing the preset medical term dictionary according to the semantic vectors of the medical terms, the term categories corresponding to the medical terms, and the representative medical terms of each term category includes:

[0028] According to the similarity between other medical terms under each term category and the representative medical term, determining other medical terms whose semantic similarity with the representative medical term is greater than or equal to the preset semantic similarity threshold as synonymous medical terms of the representative medical term of each term category;

[0029] According to the similarity between other medical terms under each term category and the representative medical term, determining other medical terms whose semantic similarity with the representative medical term is less than the preset semantic similarity threshold as synonymous medical terms of the representative medical term of each term category;

[0030] The preset medical term dictionary is constructed according to the semantic vectors of the medical terms, the term categories corresponding to the medical terms, the representative medical terms, synonymous medical terms and near-synonymous medical terms of each term category.

[0031] Optionally, the method further comprises:

[0032] If the semantic similarity between the new medical term and the target medical term is greater than or equal to the preset semantic similarity threshold, and the semantic similarity between the new medical term and other medical terms in the target term category is also greater than or equal to the preset semantic similarity threshold, then the new medical term is determined to be a synonymous medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary;

[0033] If the semantic similarity between the new medical term and the target medical term is greater than or equal to the preset semantic similarity threshold, and the semantic similarity between the new medical term and other medical terms in the target term category is less than the preset semantic similarity threshold, the new medical term is determined to be an approximate medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary.

[0034] Optionally, the method further comprises:

[0035] If the semantic similarity between the new medical term and each of the medical terms is less than the preset semantic similarity threshold, the new medical term is determined to be a newly added medical term, and the term category corresponding to the newly added medical term is a newly added term category, and the semantic vector of the newly added medical term and the newly added term category are added to the preset medical term dictionary.

[0036] In a second aspect, another embodiment of the present application provides a medical term normalization device, the device comprising:

[0037] An acquisition module, used to acquire new medical terms as input;

[0038] A vectorization module, used to use a pre-trained medical vectorization model to vectorize the new medical term to obtain a semantic vector of the new medical term;

[0039] A calculation module, used for calculating the semantic similarity between the new medical term and each medical term in a preset medical term dictionary according to the semantic vector of the new medical term;

[0040] A determination module is used to determine, if the semantic similarity between the new medical term and the target medical term among the medical terms is greater than or equal to a preset semantic similarity threshold, a representative medical term of the target term category corresponding to the target medical term as the target normalized medical term corresponding to the new medical term.

[0041] In the third aspect, another embodiment of the present application provides a computer device, comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the computer device is running, the processor communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to perform the steps of the medical terminology normalization method as described in any one of the first aspects above.

[0042] In a fourth aspect, another embodiment of the present application provides a storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the medical term normalization method as described in any one of the above-mentioned first aspects are executed.

[0043] The beneficial effects of this application are:

[0044] The present application provides a medical terminology normalization method, a computer device, and a storage medium. By obtaining an input new medical term, a pre-trained medical vectorization model is used to vectorize the new medical term to obtain a semantic vector of the new medical term. According to the semantic vector of the new medical term and the semantic vector of each medical term in the preset medical term dictionary, the semantic similarity between the new medical term and each medical term is calculated. If the semantic similarity between the new medical term and the target medical term in each medical term is greater than or equal to the preset semantic similarity threshold, the representative medical term of the target term category corresponding to the target medical term is determined as the target normalized medical term corresponding to the new medical term. The medical terminology normalization method of the present application does not require manual operation, and automatically normalizes the new medical term. It can be applied to multiple scenarios and reduce dependence on manual annotation and rule-making. It can also improve the accuracy of medical mathematical normalization, help doctors better understand patients' symptoms, and improve the accuracy of diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A schematic diagram of a process for normalizing medical terms provided in an embodiment of the present application;

[0047] Figure 2 A schematic diagram of a process for determining semantic similarity in a medical term normalization method provided in an embodiment of the present application;

[0048] Figure 3A schematic diagram of a process for constructing a medical term dictionary in a medical term normalization method provided in an embodiment of the present application;

[0049] Figure 4 A schematic diagram of a process for constructing a medical term dictionary in another medical term normalization method provided in an embodiment of the present application;

[0050] Figure 5 A schematic diagram of a process of adding a new medical term in a medical term normalization method provided in an embodiment of the present application;

[0051] Figure 6 An architectural diagram of a medical terminology normalization method provided in an embodiment of the present application;

[0052] Figure 7 A schematic diagram of the structure of a medical term normalization device provided in an embodiment of the present application;

[0053] Figure 8 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme 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. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0055] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0056] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0057] To clearly describe the medical terminology normalization method provided in the embodiment of the present application, the medical terminology normalization method provided in the embodiment of the present application is described below with reference to a plurality of drawings. Figure 1 A schematic diagram of a medical terminology normalization method provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method includes:

[0058] Step 101: Obtain input new medical terms.

[0059] Among them, medical terms refer to professional terms in the medical field, which are used to describe the structure, function, disease, medical procedure, medicine and various concepts in medical research of the human body. For example, medical terms can be hypertension, diabetes, asthma, appendicitis, etc., which are not limited in the embodiments of the present application.

[0060] Optionally, obtain a new medical term input by any medical staff, or obtain a data table of new medical terms input by any medical system. In other words, you can input a new medical term or multiple medical terms at the same time, and process multiple medical terms at the same time. In this application, it can be a new medical term based on manual input or a new medical term input after the system is automatically triggered, and the embodiments of this application are not limited to this.

[0061] Step 102: Use a pre-trained medical vectorization model to vectorize the new medical term to obtain a semantic vector of the new medical term.

[0062] Among them, the pre-trained medical vectorization model can be a generative pre-trained model based on the unidirectional Transformer architecture. When the dimension of the generative pre-trained model based on the unidirectional Transformer architecture is higher, the expressive power is stronger, and the high-dimensional vector space can capture more fine-grained semantic information, but it also means that the model has more parameters and the amount of calculation is greater; the lower the dimension, the insufficient expressive power, and the low-dimensional vector space may not be able to fully represent the rich semantics of the vocabulary. When the amount of data is too large, a high-dimensional model is required. For this reason, the generative pre-trained model based on the unidirectional Transformer architecture can be a 128- or 256-dimensional GEMMA2 model. Semantic vector is a representation method in natural language processing.

[0063] Optionally, the new medical term is input into a pre-trained medical vectorization model, and the new medical term is vectorized by the pre-trained medical vectorization model to obtain a semantic vector corresponding to the new medical term.

[0064] Step 103: Calculate the semantic similarity between the new medical term and each medical term in the preset medical term dictionary based on the semantic vector of the new medical term.

[0065] Among them, the preset medical term dictionary is pre-built to ensure that the use of medical terms follows a unified standard and reduces ambiguity and misunderstanding. The preset medical term dictionary includes multiple representative medical terms and multiple synonyms and near synonyms of the representative medical terms.

[0066] The similarity algorithm can be a similarity algorithm such as cosine similarity, Euclidean distance, Manhattan distance, Jaccard similarity, etc. Cosine similarity only focuses on the directional relationship between vectors, that is, the angle between them. If the directions of two semantic vectors are very close, then their cosine similarity will be close to 1, indicating that their semantic similarity is very high. Even if the lengths of two medical terms are different, as long as their semantic vectors are similar, cosine similarity still calculates a higher value. The similarity algorithm in this application can be cosine similarity.

[0067] Optionally, the semantic vectors of the new medical terms and the semantic vectors of each medical term in the preset medical term dictionary are calculated respectively by a cosine similarity algorithm to obtain calculation results of the semantic similarities between the multiple new medical terms and each medical term.

[0068] If the semantic similarity between the new medical term and the target medical term in each medical term is greater than or equal to the preset semantic similarity threshold, step 104 is executed.

[0069] Step 104: Determine a representative medical term of the target term category corresponding to the target medical term as the target normalized medical term corresponding to the new medical term.

[0070] The representative medical term is the term with the highest similarity to other medical terms in the target term category. The preset medical term dictionary includes multiple term categories, and each term category has a representative medical term.

[0071] Optionally, if the semantic similarity between the new medical term and the target medical term in each medical term is greater than or equal to a preset semantic similarity threshold, it means that the new medical term is similar to the target medical term in each medical term, indicating that the new medical term can be standardized. At this time, the representative medical term of the target term category corresponding to the target medical term is determined as the target normalized medical term corresponding to the new medical term.

[0072] The present application provides a medical terminology normalization method, which obtains input new medical terms, uses a pre-trained medical vectorization model to vectorize the new medical terms, obtains the semantic vector of the new medical terms, and calculates the semantic similarity between the new medical terms and each medical term in the preset medical term dictionary based on the semantic vector of the new medical term. If the semantic similarity between the new medical term and the target medical term in each medical term is greater than or equal to the preset semantic similarity threshold, the representative medical term of the target term category corresponding to the target medical term is determined as the target normalized medical term corresponding to the new medical term. The medical terminology normalization method of the present application does not require manual operation, and automatically normalizes new medical terms. It can be applied to multiple scenarios, reducing dependence on manual annotation and rule-making. It can also improve the accuracy of medical mathematical normalization, help doctors better understand patients' symptoms, and improve the accuracy of diagnosis.

[0073] Based on the above embodiment, the present application further provides a process for determining a semantic vector in a medical term normalization method. In the above step 102, a pre-trained medical vectorization model is used to vectorize a new medical term to obtain a semantic vector of the new medical term, including:

[0074] The new medical terms are vectorized using a medical vectorization model to map the new medical terms into the vector space corresponding to the medical vectorization model, thereby obtaining dense semantic vectors of the new medical terms.

[0075] Among them, multiple elements of the dense semantic vector of the new medical term represent the semantic features of the new medical term in multiple feature dimensions. The dense semantic vector has values ​​in each dimension and has a low dimension, which can better capture the semantic relationship between medical terms.

[0076] Optionally, by setting corresponding prompt words and using a medical vectorization model, the new medical term is vectorized, so that the medical vectorization model maps the new medical term to a vector space corresponding to the medical vectorization model, thereby obtaining a dense semantic vector of the new medical term.

[0077] In the embodiment of the present application, a medical vectorization model is used to vectorize new medical terms, so as to map the new medical terms to the vector space corresponding to the medical vectorization model, and obtain a dense semantic vector of the new medical terms. More features and semantic differences of the new medical terms can be captured, so as to more comprehensively express the meaning of the new medical terms.

[0078] Based on the above embodiment, the present application further provides a process for determining a medical vectorization model in a medical terminology normalization method, wherein before the above step of using the medical vectorization model to vectorize the new medical term to map the new medical term to the vector space corresponding to the medical vectorization model and obtaining the dense semantic vector of the new medical term, the process further includes:

[0079] Sample medical terms are used to train an initial large language model with multiple feature dimensions to obtain a medical vectorization model.

[0080] The initial large language model with multiple characteristic dimensions may be an initial large language model with 124 dimensions or a large language model with 256 dimensions, which is determined according to actual usage.

[0081] Optionally, multiple sample medical terms are obtained, the sample medical terms are divided into training sets and validation sets, an initial large language model with multiple feature dimensions is trained, the initial large language model with multiple feature dimensions is optimized according to the training results, and the optimized initial large language model with multiple feature dimensions is used as a medical vectorization model for vectorizing medical terms.

[0082] In the embodiment of the present application, sample medical terms are used to train an initial large language model with multiple characteristic dimensions to obtain the medical vectorization model, so that the obtained medical vectorization model is more consistent with the current scenario, thereby improving the accuracy of medical term vectorization.

[0083] Based on the above embodiments, the present application also provides a process for determining semantic similarity in a medical term normalization method. Figure 2 A schematic diagram of a process for determining semantic similarity in a medical term normalization method provided in an embodiment of the present application, such as Figure 2 As shown, before calculating the semantic similarity between the new medical term and each medical term in the preset medical term dictionary according to the semantic vector of the new medical term in step 103, the method further includes:

[0084] Step 201: Adjust the length of the semantic vector of the new medical term to a preset vector length. The length of the semantic vector of each medical term in the preset medical term dictionary is the preset vector length.

[0085] The preset vector length may be 1, which is determined according to actual conditions.

[0086] For example, when the preset vector length is 1, the semantic vector length of the new medical term is normalized so that the semantic vector of the new medical term is the preset semantic length. Accordingly, when constructing the preset medical term dictionary, the length of the semantic vector of each medical term in the preset medical term dictionary is normalized so that the length of the semantic vector of each medical term in the preset medical term dictionary is the preset vector length.

[0087] In the above step 103, the semantic similarity between the new medical term and each medical term in the preset medical term dictionary is calculated according to the semantic vector of the new medical term, including:

[0088] Step 202: Calculate the semantic similarity between the new medical term and each medical term based on the length-adjusted semantic vector of the new medical term and the semantic vector of each medical term.

[0089] Optionally, the semantic vector of the new medical term after length adjustment and the semantic vector of each medical term are used to calculate the semantic similarity between the new medical term and each medical term by cosine similarity, so as to obtain semantic similarity results of multiple new medical terms and each medical term.

[0090] In the embodiment of the present application, the length of the semantic vector of the new medical term is adjusted to a preset vector length, and the semantic similarity between the new medical term and each medical term is calculated based on the semantic vector of the new medical term after the length adjustment and the semantic vector of each medical term. The influence of the length of the semantic vector on the semantic similarity can be eliminated, thereby improving the calculation efficiency.

[0091] Based on the above embodiments, the present application also provides a process for constructing a medical term dictionary in a medical term normalization method. Figure 3 A schematic diagram of a process for constructing a medical term dictionary in a medical term normalization method provided in an embodiment of the present application, such as Figure 3 As shown, before calculating the semantic similarity between the new medical term and each medical term in the preset medical term dictionary according to the semantic vector of the new medical term in step 103, the method further includes:

[0092] Step 301: Acquire multiple medical terms from a preset medical corpus.

[0093] Among them, the preset medical corpus stores a plurality of standardized medical documents and medical terms.

[0094] Optionally, when a preset medical corpus stores multiple medical terms, the multiple medical terms are directly obtained from the preset medical corpus; when the preset medical corpus stores multiple standardized medical documents, the multiple standardized medical documents are obtained, and multiple medical terms are extracted from the multiple standardized medical documents.

[0095] Step 302: Use a medical vectorization model to vectorize each medical term to obtain a semantic vector of each medical term.

[0096] Optionally, multiple medical terms are input into a medical vector model, and each medical term is vectorized by the medical vector model to obtain a semantic vector of each medical term.

[0097] Step 303: Determine the semantic similarity between every two medical terms in each medical term according to the semantic vector of each medical term.

[0098] Optionally, according to the semantic vectors of the medical terms, the semantic similarity between every two medical terms in the medical terms is determined by a similarity algorithm.

[0099] Step 304: Perform cluster analysis on each medical term according to the semantic similarity between every two medical terms to obtain the term category corresponding to each medical term.

[0100] Among them, cluster analysis is used to classify each medical term into multiple categories, so that the similarity of each medical term in each category is high, and the similarity of each medical term in different categories is low.

[0101] Optionally, cluster analysis is performed on each medical term based on the semantic similarity between every two medical terms to obtain the term category corresponding to each medical term, so that the similarity of each medical term in the same term category is higher, and the similarity of each medical term in different categories is lower.

[0102] Step 305: According to the semantic similarity of all medical terms in each term category, determine the medical term with the highest similarity to other medical terms as the representative medical term of each term category.

[0103] Optionally, based on the semantic similarities between all medical terms in each term category, the medical term with the highest similarity to other medical terms is determined, and the medical term with the highest similarity to other medical terms is used as the representative medical term of each term category.

[0104] Step 306: construct a preset medical term dictionary based on the semantic vector of each medical term, the term category corresponding to each medical term, and the representative medical term of each term category.

[0105] Optionally, a preset medical terminology dictionary is constructed based on the semantic vectors of each medical term, the term categories corresponding to each medical term, and the representative medical terms of each term category, so that the medical terminology dictionary includes multiple categories of medical terms, and each category of medical terms includes representative medical terms, semantic vectors of representative medical terms, synonyms and near-synonyms of representative medical terms, and semantic vectors of synonyms and near-synonyms of representative medical terms.

[0106] In the embodiment of the present application, multiple medical terms are obtained from a preset medical corpus, and a preset medical term dictionary is constructed based on the multiple medical terms. This can ensure that the use of medical terms follows the same standards, reduce ambiguity and misunderstanding, and ensure the accuracy and authority of the terms.

[0107] Based on the above embodiment, the present application also provides another process of constructing a medical term dictionary in a medical term normalization method. Figure 4 A schematic diagram of a process for constructing a medical term dictionary in another medical term normalization method provided in an embodiment of the present application, such as Figure 4 As shown, in the above step 306, a preset medical term dictionary is constructed according to the semantic vector of each medical term, the term category corresponding to each medical term, and the representative medical term of each term category, including:

[0108] Step 401: Based on the similarity between other medical terms and the representative medical term in each term category, other medical terms whose semantic similarity with the representative medical term is greater than or equal to a preset semantic similarity threshold are determined as synonymous medical terms of the representative medical term in each term category.

[0109] The synonymous medical term is a medical term with the same meaning as the representative medical term. The preset semantic similarity threshold may be 95% or other higher values.

[0110] Optionally, based on the similarity between other medical terms and the representative medical terms under each term category, when the similarity between other medical term categories and the representative medical terms is greater than or equal to a preset semantic similarity threshold, the other medical terms are used as synonymous medical terms of the representative medical terms of each term category.

[0111] Step 402: Based on the similarity between other medical terms and the representative medical term in each term category, other medical terms whose semantic similarity to the representative medical term is less than a preset semantic similarity threshold are determined as synonymous medical terms of the representative medical term in each term category.

[0112] Optionally, when the similarity between other medical term categories and the representative medical terms is greater than or equal to a preset semantic similarity threshold, the other medical terms are not synonymous medical terms of the representative medical terms of each term category, and other medical terms whose semantic similarity with the representative medical terms is less than the preset semantic similarity threshold are determined as synonymous medical terms of the representative medical terms of each term category.

[0113] Step 403: construct a preset medical term dictionary based on the semantic vector of each medical term, the term category corresponding to each medical term, the representative medical term of each term category, the synonymous medical term and the near-synonymous medical term.

[0114] Optionally, a preset medical term dictionary is constructed based on the semantic vectors of each medical term, the term categories corresponding to each medical term, the representative medical terms, synonymous medical terms and near-synonymous medical terms of each term category. The preset medical term dictionary includes the term categories corresponding to each medical term, the representative medical terms, synonymous medical terms and near-synonymous medical terms under the term categories corresponding to each medical term, and the semantic vectors of the representative medical terms, synonymous medical terms and near-synonymous medical terms.

[0115] In an embodiment of the present application, by presetting a similarity threshold, synonymous medical terms and near-synonymous medical terms representing medical terms are determined, thereby improving the accuracy and consistency of synonyms and near-synonyms of medical terms. The use of synonyms and near-synonyms can make the expression of medical terms richer and more diverse, help avoid repetition and improve the vividness of the language.

[0116] Based on the above embodiments, the present application also provides a process of adding new medical terms in a medical term normalization method. Figure 5 A schematic diagram of a process of adding a new medical term in a medical term normalization method provided in an embodiment of the present application, such as Figure 5 As shown, based on the above steps 101 to 104, the method further includes:

[0117] Step 501: If the semantic similarity between the new medical term and the target medical term is greater than or equal to a preset semantic similarity threshold, and the semantic similarity between the new medical term and other medical terms in the target term category is also greater than or equal to the preset semantic similarity threshold, the new medical term is determined to be a synonymous medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary.

[0118] Optionally, if the semantic similarity between the new medical term and the target medical term is greater than or equal to a preset semantic similarity threshold, the new medical term can be normalized, and if the semantic similarity of other medical terms in the target term category is also greater than or equal to the preset semantic similarity threshold, it means that the new medical term is a synonymous medical term of the representative medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary.

[0119] Step 502: If the semantic similarity between the new medical term and the target medical term is greater than or equal to a preset semantic similarity threshold, and the semantic similarity between the new medical term and other medical terms in the target term category is less than the preset semantic similarity threshold, the new medical term is determined to be an approximate medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary.

[0120] Optionally, if the semantic similarity between the new medical term and the target medical term is greater than or equal to a preset semantic similarity threshold, the new medical term can be normalized, and if the semantic similarity of other medical terms in the target term category is also less than the preset semantic similarity threshold, it means that the new medical term is a synonymous medical term of the representative medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary.

[0121] In the application embodiment, the target term category of the new medical term is determined by a preset semantic similarity threshold, and the new medical term is determined to be a synonym or near-synonym based on the similarity between the new medical term and other semantic terms in the target term category, thereby saving time and effort for manual identification, improving the efficiency of term management, and being able to more accurately identify the synonyms or near-synonyms of the new medical term, reducing the possibility of misjudgment.

[0122] Based on the above embodiment, the present application also provides another process of adding a new medical term in a medical term normalization method. Based on the above steps 101 to 104, the method further includes:

[0123] If the semantic similarity between the new medical term and each medical term is less than the preset semantic similarity threshold, the new medical term is determined to be a newly added medical term, and the term category corresponding to the newly added medical term is the newly added term category, and the semantic vector of the newly added medical term and the newly added term category are added to the preset medical term dictionary.

[0124] Optionally, if the semantic similarity between the new medical term and each medical term is less than a preset semantic similarity threshold, that is, the new medical term cannot be normalized through the preset medical semantic dictionary, then the new medical term is determined to be a newly added medical term, and the semantic category corresponding to the newly added medical term is used as the newly added semantic category, and the semantic vector of the newly added medical term and the newly added term category are added to the preset medical term dictionary, thereby optimizing the preset medical term dictionary.

[0125] In the embodiment of the present application, if the semantic similarity between the new medical term and each medical term is less than the preset semantic similarity threshold, the semantic vector of the newly added medical term and the newly added term category are added to the preset medical term dictionary. The preset medical term dictionary can be optimized without manual intervention, and the preset medical term dictionary can be updated, thereby improving the practicality of the preset medical term dictionary.

[0126] The following is an overall description of the medical terminology normalization method provided by the present application in conjunction with the accompanying drawings. Figure 6 The following is an architecture diagram of a medical terminology normalization method provided in an embodiment of the present application, such as Figure 6 As shown, multiple medical terms are obtained from a medical corpus, and each medical term is vectorized using a medical vectorization model. According to the semantic vector of each medical term, the semantic similarity between every two medical terms in each medical term is determined, so as to perform cluster analysis on each medical term to obtain the term category corresponding to each medical term, and then a preset medical term dictionary is constructed according to the semantic vector of each medical term, the term category corresponding to each medical term, and the representative medical term of each term category. When a new medical term is input, the pre-trained medical vectorization model is used to vectorize the new medical term to obtain the semantic vector of the new medical term, and the semantic similarity between the new medical term and each medical term in the preset medical term dictionary is calculated according to the semantic vector of the new medical term and the semantic vector of each medical term in the preset medical term dictionary. If the semantic similarity between the new medical term and the target medical term in each medical term is greater than or equal to the preset semantic similarity threshold, the representative medical term of the target term category corresponding to the target medical term is determined as the target normalized medical term corresponding to the new medical term; if the semantic similarity between the new medical term and each medical term is less than the preset semantic similarity threshold, the new medical term is determined to be a newly added medical term, and the term category corresponding to the newly added medical term is the newly added term category, and the semantic vector of the newly added medical term and the newly added term category are added to the preset medical term dictionary.

[0127] Based on the same inventive concept, a medical term normalization device corresponding to the medical term normalization method is also provided in the embodiment of the present application. Since the principle of solving the problem by the device in the embodiment of the present application is similar to the above-mentioned medical term normalization method in the embodiment of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0128] Figure 7 A schematic diagram of a medical term normalization device provided in an embodiment of the present application, the device comprising:

[0129] An acquisition module 701 is used to acquire input new medical terms;

[0130] A vectorization module 702, used to use a pre-trained medical vectorization model to vectorize the new medical term to obtain a semantic vector of the new medical term;

[0131] A calculation module 703, for calculating the semantic similarity between the new medical term and each medical term according to the semantic vector of the new medical term and the semantic vector of each medical term in the preset medical term dictionary;

[0132] The determination module 704 is used to determine the representative medical term of the target term category corresponding to the target medical term as the target normalized medical term corresponding to the new medical term if the semantic similarity between the new medical term and the target medical term in each medical term is greater than or equal to a preset semantic similarity threshold.

[0133] Optionally, the vectorization module 702 is also used to: use a medical vectorization model to vectorize the new medical terminology to map the new medical terminology to a vector space corresponding to the medical vectorization model to obtain a dense semantic vector of the new medical terminology, wherein multiple elements of the dense semantic vector of the new medical terminology respectively represent the semantic features of the new medical terminology in multiple feature dimensions.

[0134] Optionally, the vectorization module 702 is further used to: use sample medical terms to train an initial large language model with multiple characteristic dimensions to obtain a medical vectorization model.

[0135] Optionally, the calculation module 703 is further used to: adjust the length of the semantic vector of the new medical term to a preset vector length, and the length of the semantic vector of each medical term in the preset medical term dictionary is the preset vector length;

[0136] According to the semantic vector of the new medical term and the semantic vector of each medical term in the preset medical term dictionary, the semantic similarity between the new medical term and each medical term is calculated, including:

[0137] The semantic similarity between the new medical term and each medical term is calculated based on the length-adjusted semantic vector of the new medical term and the semantic vector of each medical term.

[0138] Optionally, the calculation module 703 further includes: acquiring a plurality of medical terms from a preset medical corpus;

[0139] Using the medical vectorization model, each medical term is vectorized to obtain the semantic vector of each medical term;

[0140] Determine the semantic similarity between every two medical terms in each medical term according to the semantic vector of each medical term;

[0141] According to the semantic similarity between every two medical terms, cluster analysis is performed on each medical term to obtain the term category corresponding to each medical term;

[0142] According to the semantic similarity of all medical terms under each term category, the medical term with the highest similarity to other medical terms is determined as the representative medical term of each term category;

[0143] A preset medical term dictionary is constructed according to the semantic vector of each medical term, the term category corresponding to each medical term, and the representative medical term of each term category.

[0144] Optionally, the calculation module 703 is specifically used to: determine other medical terms whose semantic similarity with the representative medical term is greater than or equal to a preset semantic similarity threshold as synonymous medical terms of the representative medical term of each term category according to the similarity between other medical terms under each term category and the representative medical term;

[0145] According to the similarity between other medical terms under each term category and the representative medical term, other medical terms whose semantic similarity with the representative medical term is less than a preset semantic similarity threshold are determined as synonymous medical terms of the representative medical term of each term category;

[0146] A preset medical term dictionary is constructed based on the semantic vector of each medical term, the term category corresponding to each medical term, the representative medical term of each term category, the synonymous medical term and the near-synonymous medical term.

[0147] Optionally, the medical term normalization device further includes an adding module; the adding module is specifically used to: if the semantic similarity between the new medical term and the target medical term is greater than or equal to a preset semantic similarity threshold, and the semantic similarity between the new medical term and other medical terms in the target term category is also greater than or equal to the preset semantic similarity threshold, then determine that the new medical term is a synonymous medical term in the target term category, and add the semantic vector of the new medical term to the preset medical term dictionary;

[0148] If the semantic similarity between the new medical term and the target medical term is greater than or equal to a preset semantic similarity threshold, and the semantic similarity between the new medical term and other medical terms in the target term category is less than the preset semantic similarity threshold, the new medical term is determined to be an approximate medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary.

[0149] Optionally, an adding module is specifically used to: if the semantic similarity between the new medical term and each medical term is less than a preset semantic similarity threshold, determine that the new medical term is a newly added medical term, and the term category corresponding to the newly added medical term is a newly added term category, and add the semantic vector of the newly added medical term and the newly added term category to the preset medical term dictionary.

[0150] For descriptions of the processing flow of each module in the device and the interaction flow between each module, reference may be made to the relevant descriptions in the above method embodiment, which will not be described in detail here.

[0151] The present application also provides a computer device, Figure 8 A schematic diagram of the structure of a computer device provided in an embodiment of the present application is shown in FIG. Figure 8 As shown, the computer device includes: a processor 801, a memory 802, and optionally, a bus 803. The memory 802 stores machine-readable instructions executable by the processor 801. When the computer device is running, the processor 801 communicates with the memory 802 via the bus 803. When the machine-readable instructions are executed by the processor 801, the steps of the above-mentioned medical terminology normalization method are performed.

[0152] The embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned medical terminology normalization method are executed.

[0153] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the method embodiment, and will not be repeated in this application. In the several embodiments provided in this application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules 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 communication interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0154] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. 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 invention is essentially or part of the technical solution that contributes to the prior art or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several 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 invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), disk or optical disk and other media that can store program code.

[0155] The above are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be covered by the protection scope of the present application.

Claims

1. A medical term normalization method, characterized in that: The method comprises: Get input of new medical terms; Using a pre-trained medical vectorization model to vectorize the new medical term to obtain a semantic vector of the new medical term; Calculating the semantic similarity between the new medical term and each medical term in a preset medical term dictionary based on the semantic vector of the new medical term; If the semantic similarity between the new medical term and the target medical term in the medical terms is greater than or equal to a preset semantic similarity threshold, a representative medical term of the target term category corresponding to the target medical term is determined as the target normalized medical term corresponding to the new medical term.

2. The medical term normalization method according to claim 1, characterized in that: The method of using a pre-trained medical vectorization model to vectorize the new medical term to obtain a semantic vector of the new medical term includes: The new medical term is vectorized using the medical vectorization model to map the new medical term into a vector space corresponding to the medical vectorization model, thereby obtaining a dense semantic vector of the new medical term. Multiple elements of the dense semantic vector of the new medical term respectively represent the semantic features of the new medical term in multiple feature dimensions.

3. The medical term normalization method according to claim 2, characterized in that: Before the method adopts the medical vectorization model to vectorize the new medical term to map the new medical term to the vector space corresponding to the medical vectorization model to obtain the dense semantic vector of the new medical term, the method further includes: Sample medical terms are used to train an initial large language model with multiple feature dimensions to obtain the medical vectorization model.

4. The medical term normalization method according to claim 1, characterized in that: Before calculating the semantic similarity between the new medical term and each medical term in the preset medical term dictionary based on the semantic vector of the new medical term, the method further includes: Adjusting the length of the semantic vector of the new medical term to a preset vector length, wherein the length of the semantic vector of each medical term in the preset medical term dictionary is the preset vector length; The step of calculating the semantic similarity between the new medical term and each medical term in a preset medical term dictionary according to the semantic vector of the new medical term includes: The semantic similarity between the new medical term and the medical terms is calculated based on the length-adjusted semantic vector of the new medical term and the semantic vectors of the medical terms.

5. The medical term normalization method according to claim 1, characterized in that: Before calculating the semantic similarity between the new medical term and each medical term in the preset medical term dictionary based on the semantic vector of the new medical term, the method further includes: Get multiple medical terms from a pre-set medical corpus; Using the medical vectorization model, vectorize each medical term to obtain a semantic vector of each medical term; Determining the semantic similarity between every two medical terms in the medical terms according to the semantic vectors of the medical terms; Performing cluster analysis on the medical terms according to the semantic similarity between each two medical terms to obtain term categories corresponding to the medical terms; According to the semantic similarity of all medical terms under each term category, determine the medical term with the highest similarity to other medical terms as the representative medical term of each term category; The preset medical term dictionary is constructed according to the semantic vectors of the medical terms, the term categories corresponding to the medical terms, and the representative medical terms of each term category.

6. The medical term normalization method according to claim 5, characterized in that: The step of constructing the preset medical term dictionary according to the semantic vectors of the medical terms, the term categories corresponding to the medical terms, and the representative medical terms of each term category includes: According to the similarity between other medical terms under each term category and the representative medical term, determining other medical terms whose semantic similarity with the representative medical term is greater than or equal to the preset semantic similarity threshold as synonymous medical terms of the representative medical term of each term category; According to the similarity between other medical terms under each term category and the representative medical term, determining other medical terms whose semantic similarity with the representative medical term is less than the preset semantic similarity threshold as synonymous medical terms of the representative medical term of each term category; The preset medical term dictionary is constructed according to the semantic vectors of the medical terms, the term categories corresponding to the medical terms, the representative medical terms, synonymous medical terms and near-synonymous medical terms of each term category.

7. The medical term normalization method according to claim 1, characterized in that: The method further comprises: If the semantic similarity between the new medical term and the target medical term is greater than or equal to the preset semantic similarity threshold, and the semantic similarity between the new medical term and other medical terms in the target term category is also greater than or equal to the preset semantic similarity threshold, then the new medical term is determined to be a synonymous medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary; If the semantic similarity between the new medical term and the target medical term is greater than or equal to the preset semantic similarity threshold, and the semantic similarity between the new medical term and other medical terms in the target term category is less than the preset semantic similarity threshold, the new medical term is determined to be an approximate medical term in the target term category, and the semantic vector of the new medical term is added to the preset medical term dictionary.

8. The medical term normalization method according to claim 1, characterized in that: The method further comprises: If the semantic similarity between the new medical term and each of the medical terms is less than the preset semantic similarity threshold, the new medical term is determined to be a newly added medical term, and the term category corresponding to the newly added medical term is a newly added term category, and the semantic vector of the newly added medical term and the newly added term category are added to the preset medical term dictionary.

9. A computer device, characterized in that: include: A processor and a memory, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor executes the machine-readable instructions to perform the steps of the method for normalizing medical terms as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for normalizing medical terms as claimed in any one of claims 1 to 8 are executed.

Images