CN118227776B - Disease science popularization method and system based on artificial intelligence - Google Patents

Abstract

The invention relates to the technical field of medical information, and provides a disease science popularization method and system based on artificial intelligence, comprising the following steps: inputting a diagnosis report of a user, and extracting a disease result keyword from the diagnosis report; collecting a synonym set of keywords of disease results based on an extreme learning machine classifier technology; if the disease result keyword is not in the synonym set, updating the synonym set; according to the synonym set of the disease result keywords, crawling medical materials to be subjected to science popularization; and after translating the medical materials to be subjected to science popularization into the language set by the user, recommending the language to the user for reading. The invention provides science popularization materials related to the diseases for patients and provides users with multiple aspects of understanding of the diseases.

Description

Disease popularization method and system based on artificial intelligence

Technical Field

The present invention relates to the field of medical information technology, and in particular to a disease popular science method and system based on artificial intelligence.

Background technique

Disease popularization is an important channel for the public to understand diseases. For common cancer patients, medical institutions have invested a lot of resources in producing popular science materials (articles, videos, audio, etc.). However, due to the complexity and diversity of disease types, there is a problem that the popular science content is not well matched with disease patients. Patients need to search for relevant information on the Internet through the disease keywords in the diagnosis report. For uncommon diseases and diseases with many synonymous terms, the patient's search process is difficult and has great limitations. Therefore, automatically providing patients with disease-related popular science is an urgent problem to be solved.

Summary of the invention

The purpose of the present invention is to provide patients with popular science materials related to their diseases, so that users can understand the diseases in many aspects, and to provide a disease popular science method and system based on artificial intelligence.

In order to achieve the above-mentioned object of the invention, the embodiment of the present invention provides the following technical solutions:

The disease popularization method based on artificial intelligence includes the following steps:

Step 1, input the user's diagnosis report and extract disease result keywords from the diagnosis report;

Step 2: Based on the extreme learning machine classifier technology, a synonym set of the disease result keyword is collected; if the disease result keyword is not in the synonym set, the synonym set is updated;

Step 3: crawl medical materials to be popularized based on the synonym set of the disease result keywords;

Step 4: After translating the medical materials to be popularized into the language set by the user, the materials are recommended to the user for reading.

Before step 1, the steps include:

Building a disease dictionary , which contains the results of named diseases worldwide, F represents the disease dictionary, is the vector form of F, _fj represents the jth disease result in the disease dictionary, is the vector form of _fj , m represents the number of disease results in the disease dictionary, j=1,2,...,m.

The step 1 specifically comprises the following steps:

Segment the user's diagnostic report to obtain the segmented diagnostic report , where X represents the diagnostic report after word segmentation, is the vector form of X, _xi represents the i-th word in the diagnostic report after word segmentation, is the vector form of _xi , n represents the number of segmented words in the diagnostic report after segmentation, i=1,2,...,n;

Each word vector exist Traverse and calculate and Ratings between:

in, express and Ratings between; represents the compensation coefficient corresponding to the mth disease result; represents the compensation coefficient corresponding to the m-1th disease result; Indicates the compensation coefficient corresponding to the first disease result; represents the weight corresponding to the mth disease outcome, and T represents the matrix transpose; Indicates the weight corresponding to the first disease result; Represents the bias corresponding to the first disease result; represents the bias corresponding to the mth disease result;

The rating Sort by large to small, and users select disease result keywords based on the ratings.

In step 2, the step of collecting synonym sets of disease result keywords based on extreme learning machine classifier technology includes:

Assuming that the disease result keyword is x _i , the context set A of the disease result keyword x _i is constructed based on the diagnosis report X = {x ₁ ,x ₂ ,...,x _n } after word segmentation:

Where k represents the context degree, k<i and k<n-i;

Pre-build the synonym database Base = {B ₁ ,B ₂ ,...,B _R } of various disease results, where R is the number of synonym sets in the synonym database, B _r is the synonym set corresponding to the rth disease result, r = 1, 2,..., R, and we have:

in, represents the lth synonym in the synonym set _Br , L represents the number of synonyms in the synonym set _Br , l=1,2,...,L;

The correlation coefficient between the disease result keyword _xi and the synonym set _Br is calculated based on the extreme learning machine classifier:

Where C _r represents the correlation coefficient between the disease result keyword _xi and the rth synonym set; sigmoid represents the activation function; ELM represents the extreme learning machine classifier;

Preset Threshold ,like , it means that the synonym set _Br is the synonym set of the disease result keyword _xi ; otherwise, it means that the synonym set _Br is not the synonym set of the disease result keyword _xi .

In step 2, if the disease result keyword is not in the synonym set, the step of updating the synonym set includes:

The correlation coefficient C _r is limited to 0 ≤ _{C r} ≤ 1 by the extreme learning machine classifier, and the threshold The value range is , if C _r = 1, it means that the disease result keyword x _i is exactly in the synonym set B _r ; if C _r = 0 and/or , it means that the disease result keyword _xi has no synonym set in the synonym library Base, so a new synonym set _Br+1 is created, and _xi is added to the synonym set Br ₊₁ , and _Br+1 ={ _xi }; if , then query whether the synonym _xi is in the synonym set B _r . If not, add _xi to the synonym set B _r .

The step 3 also includes the following steps:

Recommended values for calculating medical materials:

in, Indicates the recommendation value of the medical material; G indicates the number of times the medical material has been read; H indicates the expert score; I indicates reading behavior, if the medical material has been shared, then I=1, otherwise I=0; J indicates collection behavior, if the medical material has been collected, then J=1, otherwise J=0;

Calculate the interest value of the medical material:

in, Indicates the interest value of medical materials; O indicates reading frequency;

Calculate the applicable value of the medical material:

in, represents the applicable value of medical material; e represents the Euler number; Q represents the error coefficient, and its value range is -1≤Q≤1;

Setting Thresholds ,like , then the medical material will be used as medical material suitable for popular science for patients; or the applicable value of all medical materials Sort them from large to small, and select the first S as medical materials suitable for popularizing knowledge to patients.

The step 3 specifically comprises the following steps:

Medical materials suitable for popularizing knowledge to patients are , where D represents the medical material collection, is the vector form of D, _ds represents the sth medical material in the medical material set, is the vector form of _ds , S is the number of medical materials in the medical material set, s=1,2,...,S;

The medical material collection is constructed as graph structure data. is a node, and the edge between nodes indicates that there is a reference relationship between the two nodes;

Extracting medical materials based on graph attention mechanism Global features:

in, For medical materials Softplus is the activation function; U is the number of attention heads, u represents the u-th attention head, u=1,2,...,U; W _u represents the weight matrix of the u-th attention head; Z _s is the set of neighbor nodes with edges to node s, z is the neighbor node of node s, z∈Z _s ; Indicates the medical material corresponding to the neighbor node z; Represents the association parameters of neighbor node z to node s, which are:

Among them, exp represents the exponential function with e as the base, e represents the Euler number; SoftShrink is the activation function; is the attention coefficient matrix, T represents the matrix transpose; p is the neighbor node of node s, p∈Z _s and p≠z; represents the medical material corresponding to the neighbor node p; the symbol || represents the gluing of tensors;

Compute synonym set B _r relative to medical material Related metrics:

in, is the vector form of _Br ; Represents the synonym set B _r relative to the medical material ; Represents the sigmoid activation function; represents the measurement coefficient;

Setting Thresholds ,like , then select the medical material d _s as the medical material to be popularized.

The disease popular science system based on artificial intelligence is applied to the disease popular science method based on artificial intelligence described in any of the above embodiments, including:

A keyword extraction module is used to extract disease result keywords from the diagnosis report input by the user;

The synonym collection module is used to collect synonym sets of disease result keywords based on extreme learning machine classifier technology;

A synonym updating module, used for judging whether the disease result keyword extracted by the keyword extracting module is in the synonym set collected by the synonym collecting module, and if not, updating the synonym set;

The material popular science module is used to crawl medical materials to be popularized based on the synonym set of disease result keywords;

The material translation module is used to translate the medical materials to be popularized into the language set by the user and recommend them to the users for reading.

A dictionary building module is also included to build disease dictionaries.

Compared with the prior art, the present invention has the following beneficial effects: the present invention extracts disease result keywords in user diagnosis reports, collects synonym sets of the disease result keywords, crawls relevant materials on the Internet and recommends them to users, thereby enabling users to understand disease-related information in many aspects. Compared with the method in which users independently search on the Internet for disease results represented by one word in the diagnosis report, the present invention can provide users with more comprehensive popular science information.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention 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 creative work.

FIG1 is a flow chart of the method of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

The present invention is implemented by the following technical solution. As shown in FIG1 , the disease popularization method based on artificial intelligence comprises the following steps:

Step 1: Input the user's diagnosis report and extract disease result keywords from the diagnosis report.

Pre-built disease dictionaries , which contains the results of named diseases worldwide, F represents the disease dictionary, is the vector form of F, _fj represents the jth disease result in the disease dictionary, is the vector form of _fj , m represents the number of disease results in the disease dictionary, j=1,2,...,m.

Segment the user's diagnostic report to obtain the segmented diagnostic report , where X represents the diagnostic report after word segmentation, is the vector form of X, _xi represents the i-th word in the diagnostic report after word segmentation, is the vector form of _xi , n represents the number of segmented words in the diagnostic report after segmentation, i=1,2,...,n.

Each word vector exist Traverse and calculate and Ratings between:

in, express and Ratings between; represents the compensation coefficient corresponding to the mth disease result; represents the compensation coefficient corresponding to the m-1th disease result; Indicates the compensation coefficient corresponding to the first disease result; represents the weight corresponding to the mth disease outcome, and T represents the matrix transpose; Indicates the weight corresponding to the first disease result; Represents the bias corresponding to the first disease result; Represents the bias corresponding to the mth disease result.

Since there may be more than one disease result in a diagnostic report, users can select the disease result that is the same as, most similar to, or of interest to their own disease result as the disease result keyword based on the ranking from top to bottom.

Step 2: Based on the extreme learning machine classifier technology, the synonym set of the disease result keyword is collected. If the disease result keyword is not in the synonym set, the synonym set is updated.

Assuming that the disease result keyword selected by the user is x _i , the context set A of the disease result keyword x _i is constructed based on the diagnosis report X = {x ₁ ,x ₂ ,...,x _n } after word segmentation:

Where k represents the context degree, k<i and k<n-i.

Pre-build the synonym database Base = {B ₁ ,B ₂ ,...,B _R } of various disease results, where R is the number of synonym sets in the synonym database, B _r is the synonym set corresponding to the rth disease result, r = 1, 2,..., R, and we have:

in, represents the lth synonym in the synonym set B _r , L represents the number of synonyms in the synonym set B _r , l=1,2,...,L. For example, if the rth disease result is coronary heart disease, then the synonym set B _r may include synonyms such as "coronary heart disease", "coronary atherosclerotic heart disease", "ischemic heart disease", "myocardial infarction", "heart failure", "primary cardiac arrest", etc. It should be noted that the synonyms referred to in this scheme include full names, abbreviations, common names, hyponyms, hypernyms, etc., and may also include symptoms.

The correlation coefficient between the disease result keyword _xi and the synonym set _Br is calculated based on the extreme learning machine classifier (ELM):

Among them, C _r represents the correlation coefficient between the disease result keyword _xi and the rth synonym set; sigmoid represents the activation function; ELM represents the extreme learning machine classifier. Since there are R synonym sets, R correlation coefficients can be calculated, and the preset threshold ,like , it means that the synonym set _Br is the synonym set of the disease result keyword _xi , and one disease result keyword is allowed to have multiple synonym sets; otherwise, it means that the synonym set _Br is not the synonym set of the disease result keyword _xi .

In addition, the correlation coefficient C _r is limited to 0 ≤ C _r ≤ 1 by the extreme learning machine classifier, and the threshold The value range is , if C _r = 1, it means that the disease result keyword x _i is exactly in the synonym set B _r ; if C _r = 0 and/or , it means that the disease result keyword _xi has no synonym set in the synonym library Base, then we can create a new synonym set Br ₊₁ , and add _xi to the synonym set Br ₊₁ , so _Br+1 ={ _xi }; if , then query whether the synonym _xi is in the synonym set _Br . If not, add _xi to the synonym set _Br to realize the dynamic update of the synonym base Base.

It is easy to understand that this step assumes that only one disease result keyword x _i is selected. When the user selects multiple disease result keywords, the corresponding synonym sets are obtained in the same way.

Step 3: crawl medical materials to be popularized based on the synonym set of disease result keywords.

Since there are a lot of medical materials on the Internet, but not all medical materials are suitable for patients to read, this step crawls medical materials suitable for popularizing science for patients based on factors such as the number of readings of a medical material and expert ratings. First, calculate the recommendation value of a medical material:

in, It represents the recommendation value of the medical material; G represents the number of times the medical material has been read; H represents the expert score (unit: percentage); I represents the reading behavior, if the medical material has been shared, then I=1, otherwise I=0; J represents the collection behavior, if the medical material has been collected, then J=1, otherwise J=0.

Then calculate the interest value of the medical material:

in, Indicates the interest value of medical materials; O indicates the reading frequency (times/hour).

Finally, calculate the applicable value of the medical material:

Where, e represents the Euler number; It represents the applicable value of medical material; Q represents the error coefficient, and its value range is -1≤Q≤1.

Setting Thresholds ,like , then the medical material will be used as medical material suitable for popular science for patients; or the applicable value of all medical materials Sort them from large to small, and select the first S as medical materials suitable for popularizing knowledge to patients.

Thus, the medical materials suitable for popularizing knowledge to patients are , where D represents the medical material collection, is the vector form of D, _ds represents the sth medical material in the medical material set, is the vector form of _ds , S is the number of medical materials in the medical material set, s=1,2,...,S. The medical material set is constructed as graph structure data, each medical material is a node, and the edge between nodes indicates that there is a reference relationship between the two nodes.

Extracting medical materials based on graph attention mechanism Global features:

in, For medical materials Softplus is the activation function; U is the number of attention heads, u represents the u-th attention head, u=1,2,...,U; W _u represents the weight matrix of the u-th attention head; Z _s is the set of neighbor nodes with edges to node s, z is the neighbor node of node s, z∈Z _s ; Indicates the medical material corresponding to the neighbor node z; Represents the association parameters of neighbor node z to node s, which are:

Among them, exp represents the exponential function with e as the base, e represents the Euler number; SoftShrink is the activation function; is the attention coefficient matrix, T represents the matrix transpose; p is the neighbor node of node s, p∈Z _s and p≠z; Represents the medical material corresponding to the neighbor node p; the symbol || represents the concatenation of tensors.

Compute synonym set B _r relative to medical material Related metrics:

in, is the vector form of _Br ; Represents the synonym set B _r relative to the medical material ; Represents the sigmoid activation function; Indicates the measurement coefficient. Set the threshold ,like , then select the medical material d _s as the medical material to be popularized.

Step 4: After translating the medical materials to be popularized into the language set by the user, the materials are recommended to the user for reading.

Usually, there are multiple medical materials to be popularized, and users are accustomed to different languages, such as Chinese, English, and minority languages. Users can set their accustomed language and translate the medical materials to be popularized and recommend them to users for reading. The translation behavior can use existing technology and will not be elaborated here.

The present invention also proposes an artificial intelligence-based disease science popularization system, including: a dictionary construction module, a keyword extraction module, a synonym collection module, a synonym update module, a material crawling module, a material science popularization module, and a material translation module.

The dictionary building module is used to build a disease dictionary. In detail, the dictionary building module pre-builds a disease dictionary , which contains the results of named diseases worldwide, F represents the disease dictionary, is the vector form of F, _fj represents the jth disease result in the disease dictionary, is the vector form of _fj , m represents the number of disease results in the disease dictionary, j=1,2,...,m.

The keyword extraction module is used to extract disease result keywords from the diagnosis report input by the user. In detail, the keyword extraction module performs word segmentation on the user's diagnosis report to obtain a segmented diagnosis report. , where X represents the diagnostic report after word segmentation, is the vector form of X, _xi represents the i-th word in the diagnostic report after word segmentation, is the vector form of _xi , n represents the number of segmented words in the diagnostic report after segmentation, i=1,2,...,n.

Each word vector exist Traverse and calculate and Ratings between:

in, express and Ratings between; represents the compensation coefficient corresponding to the mth disease result; represents the compensation coefficient corresponding to the m-1th disease result; Indicates the compensation coefficient corresponding to the first disease result; represents the weight corresponding to the mth disease outcome, and T represents the matrix transpose; Indicates the weight corresponding to the first disease result; Represents the bias corresponding to the first disease result; Indicates the bias corresponding to the mth disease result. A total of n×m scores will be obtained , and sort them from large to small. The ones at the front are closer to the disease results in the diagnosis report.

The synonym collection module is used to collect synonym sets of disease result keywords based on extreme learning machine classifier technology. In detail, assuming that the disease result keyword selected by the user is x _i , the synonym collection module constructs a context set A of the disease result keyword x _i based on the diagnostic report X={x ₁ ,x ₂ ,...,x _n } after word segmentation:

Where k represents the context degree, k<i and k<n-i.

Pre-build the synonym database Base = {B ₁ ,B ₂ ,...,B _R } of various disease results, where R is the number of synonym sets in the synonym database, B _r is the synonym set corresponding to the rth disease result, r = 1, 2,..., R, and we have:

in, represents the lth synonym in the synonym set _Br , L represents the number of synonyms in the synonym set _Br , l=1,2,...,L.

The correlation coefficient between the disease result keyword _xi and the synonym set _Br is calculated based on the extreme learning machine classifier (ELM):

Among them, C _r represents the correlation coefficient between keyword _xi and the rth synonym set; sigmoid represents the activation function; ELM represents the extreme learning machine classifier. Since there are R synonym sets, R correlation coefficients can be calculated, and the preset threshold ,like , it means that the synonym set _Br is the synonym set of the disease result keyword _xi , and one disease result keyword is allowed to have multiple synonym sets; otherwise, it means that the synonym set _Br is not the synonym set of the disease result keyword _xi .

The synonym updating module is used to determine whether the disease result keyword extracted by the keyword extraction module is in the synonym set collected by the synonym collection module. If not, the synonym set is updated. In detail, the synonym updating module limits the value range of the correlation coefficient C _r to 0≤C _r ≤1 through the extreme learning machine classifier, and the threshold The value range is , if C _r = 1, it means that the disease result keyword x _i is exactly in the synonym set B _r ; if C _r = 0 and/or , it means that the disease result keyword _xi has no synonym set in the synonym library Base, then we can create a new synonym set Br ₊₁ , and add _xi to the synonym set Br ₊₁ , so _Br+1 ={ _xi }; if , then query whether the synonym _xi is in the synonym set _Br . If not, add _xi to the synonym set _Br to realize the dynamic update of the synonym base Base.

The material crawling module is used to crawl medical materials suitable for popularizing science for patients. In detail, the material crawling module first calculates the recommendation value of a medical material:

in, It represents the recommendation value of the medical material; G represents the number of times the medical material has been read; H represents the expert score (unit: percentage); I represents the reading behavior, if the medical material has been shared, then I=1, otherwise I=0; J represents the collection behavior, if the medical material has been collected, then J=1, otherwise J=0.

Then calculate the interest value of the medical material:

in, Indicates the interest value of medical materials; O indicates the reading frequency (times/hour).

Finally, calculate the applicable value of the medical material:

in, It represents the applicable value of medical material; e represents the Euler number; Q represents the error coefficient, and its value range is -1≤Q≤1.

Setting Thresholds ,like , then the medical material will be used as medical material suitable for popular science for patients; or the applicable value of all medical materials Sort them from large to small, and select the first S as medical materials suitable for popularizing knowledge to patients.

The material popularization module is used to crawl medical materials to be popularized based on the synonym set of disease result keywords. In detail, the material popularization module obtains medical materials suitable for popularizing science for patients. , where D represents the medical material collection, is the vector form of D, _ds represents the sth medical material in the medical material set, is the vector form of _ds , S is the number of medical materials in the medical material set, s=1,2,...,S. The medical material set is constructed as a graph structure data, each medical material is a node, and the edge between nodes indicates that there is a reference relationship between the two nodes.

Extracting medical materials based on graph attention mechanism Global features:

in, For medical materials Softplus is the activation function; U is the number of attention heads, u represents the u-th attention head, u=1,2,...,U; _Wu represents the weight matrix of the u-th attention head; _Zs is the set of neighbor nodes with edges to node s, z is the neighbor node of node s, _z∈Zs ; Indicates the medical material corresponding to the neighbor node z; Represents the association parameters of neighbor node z to node s, which are:

Among them, exp represents the exponential function with e as the base, e represents the Euler number; SoftShrink is the activation function; is the attention coefficient matrix, T represents the matrix transpose; p is the neighbor node of node s, p∈Z _s and p≠z; Represents the medical material corresponding to the neighbor node p; the symbol || represents the concatenation of tensors.

Compute synonym set B _r relative to medical material Related metrics:

in, is the vector form of _Br ; Represents the synonym set B _r relative to the medical material ; Represents the sigmoid activation function; Indicates the measurement coefficient. Set the threshold ,like , then select the medical material d _s as the medical material to be popularized.

The material translation module is used to translate the medical material to be popularized into a language set by the user and recommend it to the user for reading. In detail, the material translation module provides the user with the function of setting a customary language, such as Chinese, English, minority languages, etc., and translates the medical material to be popularized and recommends it to the user for reading.

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

Claims (5)

1. A disease popular science method based on artificial intelligence, characterized by comprising the following steps: Step 1, input the user's diagnosis report and extract disease result keywords from the diagnosis report; Before step 1, the steps include: Building a disease dictionary , which contains the results of named diseases worldwide, F represents the disease dictionary, is the vector form of F, _fj represents the jth disease result in the disease dictionary, is the vector form of _fj , m represents the number of disease results in the disease dictionary, j=1,2,...,m; The step 1 specifically comprises the following steps: Segment the user's diagnostic report to obtain the segmented diagnostic report , where X represents the diagnostic report after word segmentation, is the vector form of X, _xi represents the i-th word in the diagnostic report after word segmentation, is the vector form of _xi , n represents the number of segmented words in the diagnostic report after segmentation, i=1,2,...,n; Each word vector exist Traverse and calculate and Ratings between: in, express and Ratings between; represents the compensation coefficient corresponding to the mth disease result; represents the compensation coefficient corresponding to the m-1th disease result; Indicates the compensation coefficient corresponding to the first disease result; represents the weight corresponding to the mth disease outcome, and T represents the matrix transpose; Indicates the weight corresponding to the first disease result; Represents the bias corresponding to the first disease result; represents the bias corresponding to the mth disease result; The rating Sort by highest to lowest, and users select disease result keywords based on the ratings; Step 2: Based on the extreme learning machine classifier technology, a synonym set of the disease result keyword is collected; if the disease result keyword is not in the synonym set, the synonym set is updated; Step 3: crawl medical materials to be popularized based on the synonym set of the disease result keywords; Step 4: After translating the medical materials to be popularized into the language set by the user, the materials are recommended to the user for reading. 2. The disease science popularization method based on artificial intelligence according to claim 1 is characterized in that: in step 2, the step of collecting synonym sets of disease result keywords based on extreme learning machine classifier technology includes: Assuming that the disease result keyword is x _i , the context set A of the disease result keyword x _i is constructed based on the diagnosis report X = {x ₁ ,x ₂ ,...,x _n } after word segmentation: Where k represents the context degree, k<i and k<n-i; Pre-build the synonym database Base = {B ₁ ,B ₂ ,...,B _R } of various disease results, where R is the number of synonym sets in the synonym database, B _r is the synonym set corresponding to the rth disease result, r = 1, 2,..., R, and we have: in, represents the lth synonym in the synonym set _Br , L represents the number of synonyms in the synonym set _Br , l=1,2,...,L; The correlation coefficient between the disease result keyword _xi and the synonym set _Br is calculated based on the extreme learning machine classifier: Where C _r represents the correlation coefficient between the disease result keyword _xi and the rth synonym set; sigmoid represents the activation function; ELM represents the extreme learning machine classifier; Preset Threshold ,like , it means that the synonym set _Br is the synonym set of the disease result keyword _xi ; otherwise, it means that the synonym set _Br is not the synonym set of the disease result keyword _xi . 3. The disease science popularization method based on artificial intelligence according to claim 2 is characterized in that: in step 2, if the disease result keyword is not in the synonym set, the step of updating the synonym set includes: The correlation coefficient C _r is limited to 0 ≤ _{C r} ≤ 1 by the extreme learning machine classifier, and the threshold The value range is , if C _r = 1, it means that the disease result keyword x _i is exactly in the synonym set B _r ; if C _r = 0 and/or , it means that the disease result keyword _xi has no synonym set in the synonym library Base, so a new synonym set _Br+1 is created, and _xi is added to the synonym set Br ₊₁ , and _Br+1 ={ _xi }; if , then query whether the synonym _xi is in the synonym set B _r . If not, add _xi to the synonym set B _r . 4. The disease popular science method based on artificial intelligence according to claim 3 is characterized in that: the step 3 specifically comprises the following steps: Medical materials suitable for popularizing knowledge to patients are , where D represents the medical material collection, is the vector form of D, _ds represents the sth medical material in the medical material set, is the vector form of _ds , S is the number of medical materials in the medical material set, s=1,2,...,S; The medical material collection is constructed as graph structure data. is a node, and the edge between nodes indicates that there is a reference relationship between the two nodes; Extracting medical materials based on graph attention mechanism Global features: in, For medical materials Softplus is the activation function; U is the number of attention heads, u represents the u-th attention head, u=1,2,...,U; W _u represents the weight matrix of the u-th attention head; Z _s is the set of neighbor nodes with edges to node s, z is the neighbor node of node s, z∈Z _s ; Indicates the medical material corresponding to the neighbor node z; Represents the association parameters of neighbor node z to node s, which are: Among them, exp represents the exponential function with e as the base, e represents the Euler number; SoftShrink is the activation function; is the attention coefficient matrix, T represents the matrix transpose; p is the neighbor node of node s, p∈Z _s and p≠z; represents the medical material corresponding to the neighbor node p; the symbol || represents the gluing of tensors; Compute synonym set B _r relative to medical material Related metrics: in, is the vector form of _Br ; Represents the synonym set B _r relative to the medical material ; Represents the sigmoid activation function; represents the measurement coefficient; Setting Thresholds ,like , then select the medical material d _s as the medical material to be popularized. 5. An artificial intelligence-based disease science popularization system, applied to the artificial intelligence-based disease science popularization method according to any one of claims 1 to 4, characterized in that it comprises: Dictionary building module, used to build disease dictionaries; Dictionary building module pre-builds disease dictionaries , which contains the results of named diseases worldwide, F represents the disease dictionary, is the vector form of F, _fj represents the jth disease result in the disease dictionary, is the vector form of _fj , m represents the number of disease results in the disease dictionary, j=1,2,...,m; A keyword extraction module is used to extract disease result keywords from the diagnosis report input by the user; The keyword extraction module performs word segmentation on the user's diagnostic report to obtain a diagnostic report after word segmentation. , where X represents the diagnostic report after word segmentation, is the vector form of X, _xi represents the i-th word in the diagnostic report after word segmentation, is the vector form of _xi , n represents the number of segmented words in the diagnostic report after segmentation, i=1,2,...,n; Each word vector exist Traverse and calculate and Ratings between: in, express and Ratings between; represents the compensation coefficient corresponding to the mth disease result; represents the compensation coefficient corresponding to the m-1th disease result; Indicates the compensation coefficient corresponding to the first disease result; represents the weight corresponding to the mth disease outcome, and T represents the matrix transpose; Indicates the weight corresponding to the first disease result; Represents the bias corresponding to the first disease result; represents the bias corresponding to the mth disease result; The rating Sort by highest to lowest, and users select disease result keywords based on the ratings; The synonym collection module is used to collect synonym sets of disease result keywords based on extreme learning machine classifier technology; A synonym updating module, used for judging whether the disease result keyword extracted by the keyword extracting module is in the synonym set collected by the synonym collecting module, and if not, updating the synonym set; The material popular science module is used to crawl medical materials to be popularized based on the synonym set of disease result keywords; The material translation module is used to translate the medical materials to be popularized into the language set by the user and recommend them to the users for reading.