Method for converting SNOMED-CT medical concept into vector

Abstract

The invention relates to the technical field of bioinformatics, in particular to a method for converting an SNOMED-CT medical concept into a vector. The problem that the quality of the medical concept vector is not high is solved. According to the technical scheme, the method comprises the steps of S1, data collection, training and preprocessing; s2, carrying out medical concept embedding development based on a Node2Vec method; s3, semantic information of the medical concept is calculated through the development text description vector; and S4, performing vector representation on the medical concept by using text description. The method has the advantages that the vector representation of the SNOMED-CT medical concept is improved, and the accuracy of medical care and biomedical application is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of bioinformatics, and in particular to a method for converting SNOMED-CT medical concepts into vectors. Background Art

[0002] SNOMED-CT medical concept vector representation plays an important role for many AI-based healthcare and biomedical applications, such as clinical decision support systems, disease diagnosis, text classification, and information retrieval. By improving the quality of SNOMED-CT medical concept vector representation, the performance of these methods can be improved.

[0003] Reference URL: https: / / snap.stanford.edu / node2vec / The existing Node2Vec method is a graph embedding technology that is used to learn vector representations of graph nodes. The ordinary Node2Vec method only relies on the graph structure of SNOMED-CT, which limits its efficiency. SNOMED-CT medical concepts provide text descriptions and explanations for them.

[0004] The present invention solves the following problems:

[0005] (1) How to use the medical concept text descriptions provided by SNOMED-CT to calculate the semantic information of medical concepts;

[0006] (2) How to integrate the semantic information in the text description of medical concepts provided by SNOMED-CT and its graph embedding based on the Node2Vec method. Summary of the invention

[0007] The purpose of the present invention is to provide a method for converting SNOMED-CT medical concepts into vectors. The present invention improves the vector representation of medical concepts by combining graph embedding based on the Node2Vec method and semantic information obtained from the medical concept text description vector provided by SNOMED-CT, thereby improving the quality of the medical concept vector.

[0008] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is specifically as follows:

[0009] A method for converting SNOMED-CT medical concepts into vectors, comprising the following steps:

[0010] S1: data collection, training, and preprocessing;

[0011] S2: Development of medical concept embedding based on Node2Vec approach;

[0012] S3: Computing semantic information of medical concepts by developing text description vectors;

[0013] S4: Vector representation of medical concepts using text descriptions.

[0014] Step S1 specifically uses the graph structure of SNOMED-CT to learn the graph embedding of medical concepts based on the Node2Vec method. The text description of medical concepts provided by SNOMED-CT is used to derive the semantic information of medical concepts. Python programming language version 3.10 and Python's Natural Language Toolkit (NLTK) library are used for data preprocessing.

[0015] Step S2 is specifically as follows:

[0016] The Node2Vec method is applied to the SNOMED-CT graph to generate graph embeddings of medical concepts, as shown in formula (1):

[0017]

[0018] Among them: MedCon represents the medical concept in SNOMED-CT, function Node2Vec SNOMED-CTgraph The graph structure of SNOMED-CT is used to generate the graph embedding of the medical concept MedCon. The graph embedding of the medical concept is represented as

[0019] Step S3 is specifically as follows:

[0020] The text description vector of the medical concept is used to calculate its semantic information, as shown in formula (2):

[0021]

[0022] in: represents the text description vector of the medical concept MedCon based on the semantic information provided by SNOMED-CT. The function (f|MedCon,d|) represents the frequency of the medical concept MedCon in a specific text description d. The symbol γ is a linear contribution parameter. The function ∑ W∈d T|W,d| calculates the total number of text words in the text description d, where the symbol e represents a natural constant. The function |D| represents the total number of text descriptions in SNOMED-CT. The function |d∈D:MedCon∈d| calculates the total number of text descriptions D in which the medical concept MedCon appears.

[0023] Step S4 is specifically as follows:

[0024] The medical concepts of SNOMED-CT are represented as vectors by combining its graph embedding based on the Node2Vec method and text description, as shown in formula (3):

[0025]

[0026] in: By combining its graph embedding based on Node2Vec method and text description vector, it represents the vector representation of SNOMED medical concept development; the symbol α is a linear contribution parameter, which determines the function in formula (3) and the relative importance of

[0027] Relative importance means which function, that is, and Which is more important should be given more weight. The binary operator (+) represents the concatenation of two vectors; the binary operator (*) represents the product of a scalar value and a vector.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] (1) This paper uses the graphical structure and textual description of the SNOMED-CT medical ontology to develop a new method to improve the vector representation of SNOMED-CT medical concepts. The proposed method can improve the accuracy of healthcare and biomedical applications, including clinical decision support systems, disease diagnosis, and information retrieval;

[0030] (2) Using formula (2) to calculate the semantic information of medical concepts using the medical concept text description provided by SNOMED-CT;

[0031] (2) Formula (3) is used to integrate the semantic information in the text description of medical concepts provided by SNOMED-CT and its graph embedding based on the Node2Vec method. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0033] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Of course, the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.

[0035] Example 1

[0036] like Figure 1 shown.

[0037] A method for converting SNOMED-CT medical concepts into vectors, comprising the following steps:

[0038] S1: data collection, training, and preprocessing;

[0039] S2: Development of medical concept embedding based on Node2Vec approach;

[0040] S3: Computing semantic information of medical concepts by developing text description vectors;

[0041] S4: Vector representation of medical concepts using text descriptions.

[0042] Step S1 specifically uses the graph structure of SNOMED-CT to learn the graph embedding of medical concepts based on the Node2Vec method. The text description of medical concepts provided by SNOMED-CT is used to derive the semantic information of medical concepts. Python programming language version 3.10 and Python's Natural Language Toolkit (NLTK) library are used for data preprocessing.

[0043] Step S2 is specifically as follows:

[0044] The Node2Vec method is applied to the SNOMED-CT graph to generate graph embeddings of medical concepts, as shown in formula (1):

[0045]

[0046] Among them: MedCon represents the medical concept in SNOMED-CT, function Node2Vec SNOMED-CTgraph The graph structure of SNOMED-CT is used to generate the graph embedding of the medical concept MedCon. The graph embedding of the medical concept is represented as

[0047] Step S3 is specifically as follows:

[0048] The text description vector of the medical concept is used to calculate its semantic information, as shown in formula (2):

[0049]

[0050] in: represents the text description vector of the medical concept MedCon based on the semantic information provided by SNOMED-CT. The function (f|MedCon, d|) represents the frequency of the medical concept MedCon in a specific text description d. The symbol Υ is a linear contribution parameter. In this invention, Υ=0.30. The function ∑ W∈dT|W,d| calculates the total number of text words in the text description d, where the symbol e represents a natural constant. The function |D| represents the total number of text descriptions in SNOMED-CT. The function |d∈D:MedCon∈d| calculates the total number of text descriptions D in which the medical concept MedCon appears.

[0051] Step S4 is specifically as follows:

[0052] The medical concepts of SNOMED-CT are represented as vectors by combining its graph embedding based on the Node2Vec method and text description, as shown in formula (3):

[0053]

[0054] in: By combining graph embeddings based on its Node2Vec approach and text description vectors, it represents the vector representation of SNOMED medical concept development; the symbol α is a linear contribution parameter that determines the function of the company (3) and the relative importance of

[0055] Relative importance means which function, that is, and Which one is more important should be given a greater weight. In the present invention, the value of α is set to 0.50 and the two functions have equal importance. The binary operator (+) represents the concatenation of two vectors; the binary operator (*) represents the product of a scalar value and a vector.

[0056] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for converting SNOMED-CT medical concepts into vectors, characterized in that: The following steps are involved: S1: data collection, training, and preprocessing; S2: Development of medical concept embedding based on Node2Vec approach; S3: Computing semantic information of medical concepts by developing text description vectors; S4: Vector representation of medical concepts using text descriptions.

2. A method for representing gene ontology medical term vectors according to claim 1, characterized in that: Step S2 is specifically as follows: The Node2Vec method is applied to the SNOMED-CT graph to generate graph embeddings of medical concepts, as shown in formula (1): Among them: MedCon represents the medical concept in SNOMED-CT, function Node2Vec SNOMED-CTgraph The graph structure of SNOMED-CT is used to generate the graph embedding of the medical concept MedCon; the graph embedding of the medical concept is represented as 3. A method for converting SNOMED-CT medical concepts into vectors according to claim 1, characterized in that: Step S3 is specifically as follows: The text description vector of the medical concept is used to calculate its semantic information, as shown in formula (2): in: represents the text description vector of the medical concept MedCon based on the semantic information provided by SNOMED-CT. The function (f|MedCon,d|) represents the frequency of the medical concept MedCon in a specific text description d. The symbol γ is a linear contribution parameter. The function ∑ W∈d T|W,d| calculates the total number of text words in the text description d, where the symbol e represents a natural constant. The function |D| represents the total number of text descriptions in SNOMED-CT. The function |d∈D:MedCon∈d| calculates the total number of text descriptions D in which the medical concept MedCon appears.

4. A method for converting SNOMED-CT medical concepts into vectors according to claim 1, characterized in that: Step S4 is specifically as follows: The medical concepts of SNOMED-CT are represented as vectors by combining its graph embedding based on the Node2Vec method and text description, as shown in formula (3): in: By combining its graph embedding based on Node2Vec method and text description vector, it represents the vector representation of SNOMED medical concept development; the symbol α is a linear contribution parameter, which determines the function in formula (3) and The binary operator (+) represents the concatenation of two vectors; the binary operator (*) represents the multiplication of a scalar value and a vector.

Images