A hierarchical classification management method and device based on an intracardiac branch of medicine library of coronary heart disease

Abstract

The present application relates to the technical field of drug hierarchical management, and discloses a hierarchical classification management method based on a cardiology coronary heart disease drug library, which comprises the following steps: performing word vector data cleaning on historical coronary heart disease drug data obtained in advance to obtain a standard coronary heart disease drug word vector set, and splitting the standard coronary heart disease drug word vector set into single drug word vector groups according to drug names; extracting a plurality of drug effect categories and a plurality of drug feedback level categories from the single drug word vector groups; obtaining a drug picture of a coronary heart disease drug to be classified, and extracting standard drug features, standard drug terms and standard drug codes from the drug picture; determining a target drug name according to the standard drug features, the standard drug terms and the standard drug codes, and determining a target drug category and a target drug level corresponding to the target drug name. The present application also provides a hierarchical classification management device based on the cardiology coronary heart disease drug library. The present application can improve the efficiency of drug hierarchical management.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of drug hierarchical management, in particular to a hierarchical classification management method and device based on a coronary heart disease drug library of a cardiology department. BACKGROUND

[0002] Coronary heart disease, namely coronary atherosclerotic heart disease, has become an increasingly non-ignorable disease in modern medicine with the acceleration of the life rhythm and the acceleration of population aging. There are a large number of drugs for treating coronary heart disease in a coronary heart disease drug library. In order to facilitate drug management and use, the coronary heart disease drugs need to be classified.

[0003] The existing coronary heart disease drug classification technology is mostly a drug classification method based on simple drug categories, for example, coronary heart disease drugs are divided into traditional Chinese medicines, western medicines, Chinese patent medicines and biological preparations. In actual application, the drug classification method based on simple drug categories is difficult to show the curative effect of the drugs and the adverse reactions of the drugs. In the process of dispensing drugs by doctors, further retrieval is needed, which may cause low efficiency in drug classification management. SUMMARY

[0004] The application provides a hierarchical classification management method and device based on a coronary heart disease drug library of a cardiology department, which mainly aims to solve the problem of low efficiency in drug classification management.

[0005] To achieve the above-mentioned purpose, the application provides a hierarchical classification management method based on a coronary heart disease drug library of a cardiology department, which comprises the following steps:

[0006] The keyword matching algorithm is used to clean the word vector data of the pre-acquired historical coronary heart disease drug data, so as to obtain a standard coronary heart disease drug word vector set. The standard coronary heart disease drug word vector set is split into single-drug word vector group sets according to drug names.

[0007] Single-drug effect word vector group sets and single-drug medication feedback word vector group sets are extracted from the single-drug word vector group sets. The single-drug effect word vector group sets are subjected to word frequency clustering, so as to obtain a plurality of drug effect classes. The single-drug medication feedback word vector group sets are subjected to word frequency clustering, so as to obtain a plurality of drug feedback level classes.

[0008] A drug picture of a coronary heart disease drug to be classified is acquired, the drug picture is subjected to picture denoising and picture equalization to obtain an equalized drug picture, the equalized drug picture is sequentially subjected to threshold picture segmentation and binarization to obtain a standard drug picture, standard drug features are extracted from the standard drug picture, wherein the sequential threshold picture segmentation and binarization of the equalized drug picture to obtain the standard drug picture comprises: generating a secondary gray scale histogram of the equalized drug picture, sequentially selecting a gray scale level in the secondary gray scale histogram as a target gray scale level, and segmenting the equalized drug picture into a target drug picture and a target background picture according to the target gray scale level; the inter-gray scale variance between the target drug picture and the target background picture is calculated by using the following inter-gray scale variance method:

[0009]

[0010] wherein T refers to the inter-gray scale variance, M refers to the pixel length of the secondary gray scale histogram, N refers to the pixel width of the secondary gray scale histogram, N0 refers to the total number of pixels of the target drug picture, N1 refers to the total number of pixels of the target background picture, n0 refers to the nth0 pixel in the target drug picture, n1 refers to the nth1 pixel in the target background picture, and g(·) is a gray scale value symbol; the target drug picture corresponding to the maximum inter-gray scale variance is selected as a primary drug picture, the primary drug picture is subjected to binarization to obtain a standard drug picture;

[0011] The standard drug picture is subjected to feature picture segmentation to obtain a drug term picture and a drug code picture, standard drug terms are extracted from the drug term picture, and standard drug codes are extracted from the drug code picture;

[0012] The target drug name is determined according to the standard drug features, the standard drug terms and the standard drug codes, the target drug category corresponding to the target drug name is determined by using the drug effect class, the target drug level corresponding to the target drug name is determined by using the drug feedback level class, and the coronary heart disease drug to be classified is subjected to classified management according to the target drug category and the target drug level.

[0013] Optionally, the pre-acquired historical coronary heart disease drug data is subjected to word vector data cleaning by using a keyword matching algorithm to obtain a standard coronary heart disease drug word vector set, comprising:

[0014] The pre-acquired primary drug data is subjected to text segmentation to obtain a primary drug word set;

[0015] Filtering out the garbled words and meaningless words from the primary drug word set to obtain a secondary drug word set, and vectorizing the secondary drug word set into a secondary drug word vector set;

[0016] Selecting the secondary drug word vectors in the secondary drug word vector set one by one as target secondary drug word vectors, and calculating the matching similarity of each target secondary drug word vector with each drug word vector in the preset drug word vector library:

[0017] Selecting the drug word vector corresponding to the maximum matching similarity from the drug word vector library as a target standard drug word vector, and replacing the target secondary drug word vector with the target standard drug word vector to obtain a standard coronary heart disease drug word vector set.

[0018] Optionally, the word frequency clustering of the single-drug effect word vector group set is performed to obtain a plurality of drug effect classes, including:

[0019] Selecting the single-drug effect word vector groups in the single-drug effect word vector group set one by one as target single-drug effect word vector groups, and performing vector weighted merging on the target single-drug effect word vector groups to obtain weighted single-drug effect word vector groups;

[0020] Vector splicing the weighted single-drug effect word vector groups to obtain a standard single-drug effect sentence vector, and collecting all standard single-drug effect sentence vectors into a standard single-drug effect sentence vector set;

[0021] Performing word frequency clustering on the standard single-drug effect sentence vector set by using a preset word frequency distance algorithm to obtain a plurality of drug effect classes.

[0022] Optionally, the word frequency clustering of the standard single-drug effect sentence vector set by using a preset word frequency distance algorithm to obtain a plurality of drug effect classes includes:

[0023] Splitting the standard single-drug effect sentence vector set into a plurality of single-drug effect sentence vector groups, and randomly selecting a primary effect center sentence vector of each single-drug effect sentence vector group;

[0024] Calculating the word frequency similarity between each standard single-drug effect sentence vector in the standard single-drug effect sentence vector set and each primary effect center sentence vector by using a word frequency similarity algorithm:

[0025]

[0026] wherein C refers to the word frequency similarity, p refers to the serial number of the word vector in the standard single-drug effect sentence vector, P refers to the total number of word vectors in the standard single-drug effect sentence vector, r refers to the serial number of the word vector in the primary effect center sentence vector, R refers to the total number of word vectors in the primary effect center sentence vector, f p refers to the weight of the pth word vector in the standard single-drug effect sentence vector, F p refers to the pth word vector in the standard single-drug effect sentence vector, E r refers to the rth word vector in the primary effect center sentence vector, e r refers to the weight of the rth word vector in the primary effect center sentence vector;

[0027] According to the word frequency similarity, each single-drug effect sentence vector group is updated into a corresponding secondary single-drug effect sentence vector group;

[0028] The secondary effect center sentence vector of each secondary single-drug effect sentence vector group is calculated, the center word frequency similarity between each secondary effect center sentence vector and the corresponding primary effect center sentence vector is calculated, and the sum of all center word frequency similarities is taken as a target center word frequency similarity;

[0029] According to the target center word frequency similarity, each secondary single-drug effect sentence vector group is iteratively updated into a corresponding drug effect class.

[0030] Optionally, the picture denoising and picture equalization are performed on the drug picture to obtain an equalized drug picture, including:

[0031] The drug picture is subjected to a gray-scale operation to obtain a gray-scale drug picture;

[0032] The gray-scale drug picture is sequentially subjected to a median filter operation and a bilateral filter operation to obtain a filtered drug picture;

[0033] A primary gray-scale histogram of the filtered drug picture is generated, and a probability equalization operation is performed on the primary gray-scale histogram to obtain an equalized primary gray-scale histogram;

[0034] The filtered drug picture is updated using the equalized primary gray-scale histogram to obtain an equalized drug picture.

[0035] Optionally, the feature picture segmentation operation is performed on the standard drug picture to obtain a drug term picture and a drug code picture, including:

[0036] A primary drug edge is detected from the standard drug picture, and an edge fitting operation is performed on the primary drug edge to obtain a secondary drug edge;

[0037] correct the standard drug picture by using the secondary drug edge to obtain a corrected drug picture;

[0038] extract a corrected drug texture from the corrected drug picture, perform texture matching classification on the corrected drug texture to obtain a drug entry texture and a drug code texture;

[0039] extract a drug entry picture from the corrected drug picture according to the drug entry texture, and extract a drug code picture from the corrected drug picture according to the drug code texture.

[0040] Optionally, the extracting a standard drug entry from the drug entry picture comprises:

[0041] extracting a layout edge from the drug entry picture, performing layout segmentation on the drug entry picture according to the layout edge to obtain a layout entry picture;

[0042] performing character cutting on the layout entry picture to obtain a layout character atlas;

[0043] performing character recognition on the layout character pictures in the layout character atlas one by one to obtain a standard drug entry.

[0044] Optionally, the extracting a standard drug code from the drug code picture comprises:

[0045] performing vertical correction on the drug code picture to obtain a standard drug code picture;

[0046] performing barcode partitioning on the standard drug code picture to obtain a barcode interval set;

[0047] performing character recognition on the barcode intervals in the barcode interval set one by one to obtain a primary drug code;

[0048] performing binary conversion on the primary drug code to obtain a standard drug code.

[0049] Optionally, the determining a target drug name according to the standard drug feature, the standard drug entry, and the standard drug code comprises:

[0050] matching a picture drug name set according to the standard drug feature in a preset drug feature library,

[0051] matching an entry drug name set corresponding to the standard drug entry from the single-drug effect word vector group set;

[0052] querying a code drug name set corresponding to the standard drug code from a preset drug code library;

[0053] An intersection of the picture drug name set, the entry drug name set and the coded drug name set is taken as a target drug name.

[0054] To solve the above problems, the application further provides a hierarchical classification management device based on a coronary heart disease drug library in a cardiology department, which comprises:

[0055] A data cleaning module is configured to perform word vector data cleaning on historical coronary heart disease drug data obtained in advance by using a keyword matching algorithm to obtain a standard coronary heart disease drug word vector set, and split the standard coronary heart disease drug word vector set into single-drug word vector groups according to drug names.

[0056] A word frequency clustering module is configured to extract a single-drug effect word vector group set and a single-drug medication feedback word vector group set from the single-drug word vector group set, perform word frequency clustering on the single-drug effect word vector group set to obtain a plurality of drug effect categories, and perform word frequency clustering on the single-drug medication feedback word vector group set to obtain a plurality of drug feedback level categories.

[0057] A primary feature extraction module is configured to obtain a drug picture of a coronary heart disease drug to be classified, perform picture denoising and picture equalization on the drug picture to obtain an equalized drug picture, sequentially perform threshold picture segmentation and binarization on the equalized drug picture to obtain a standard drug picture, and extract standard drug features from the standard drug picture, wherein the sequentially performing threshold picture segmentation and binarization on the equalized drug picture to obtain a standard drug picture comprises: generating a secondary gray level histogram of the equalized drug picture, selecting a gray level in the secondary gray level histogram as a target gray level one by one, and segmenting the equalized drug picture into a target drug picture and a target background picture according to the target gray level; and calculating an inter-gray level variance between the target drug picture and the target background picture by using the following inter-gray level variance method:

[0058]

[0059] wherein T represents the inter-gray level variance, M represents a pixel length of the secondary gray level histogram, N represents a pixel width of the secondary gray level histogram, N0 represents a total number of pixels of the target drug picture, N1 represents a total number of pixels of the target background picture, n0 represents an nth pixel in the target drug picture, n1 represents an nth pixel in the target background picture, and g(·) is a gray value symbol; selecting a target drug picture corresponding to the maximum inter-gray level variance as a primary drug picture, performing binarization on the primary drug picture to obtain a standard drug picture.

[0060] A secondary feature extraction module is configured to perform feature picture segmentation on the standard medicine picture to obtain a medicine entry picture and a medicine code picture, extract a standard medicine entry from the medicine entry picture, and extract a standard medicine code from the medicine code picture;

[0061] A medicine classification module is configured to determine a target medicine name according to the standard medicine feature, the standard medicine entry, and the standard medicine code, determine a target medicine category corresponding to the target medicine name by using the medicine effect class, determine a target medicine level corresponding to the target medicine name by using the medicine feedback level class, and perform hierarchical management on the coronary heart disease medicine to be classified according to the target medicine category and the target medicine level.

[0062] The embodiment of the present application can reduce the types of words in the historical coronary heart disease medicine data and standardize the expressions of various synonymous words by using a keyword matching algorithm to perform word vector data cleaning on the pre-acquired historical coronary heart disease medicine data to obtain a standard coronary heart disease medicine word vector set, thereby reducing the data dimension and improving the computing efficiency, can facilitate subsequent classification and hierarchical management of medicines for each medicine name by splitting the standard coronary heart disease medicine word vector set into single medicine word vector group sets according to the medicine names, can classify each medicine in the coronary heart disease medicine library according to the effect of the medicine and perform hierarchical management according to the medicine feedback by performing word frequency clustering on the single medicine effect word vector group set to obtain a plurality of medicine effect classes and performing word frequency clustering on the single medicine use feedback word vector group set to obtain a plurality of medicine feedback level classes.

[0063] By carrying out picture denoising and picture equalization on the medicine picture, an equalized medicine picture is obtained, which can remove noise pixels and shooting light spots caused by brightness reflection and the like in the medicine picture, thereby retaining more picture details, by sequentially carrying out threshold picture segmentation and binarization operation on the equalized medicine picture, a standard medicine picture is obtained, which can extract the relevant area of the medicine packaging box from the medicine picture, facilitating subsequent feature extraction, by extracting standard medicine features from the standard medicine picture, the name of the medicine can be recognized according to the packaging features of the medicine, thereby facilitating medicine classification, by extracting standard medicine terms from the medicine term picture and extracting standard medicine codes from the medicine code picture, more medicine features of the standard medicine picture can be extracted, thereby improving the accuracy of medicine name recognition, by classifying and managing the medicine to be classified according to the target medicine category and the target medicine level, the medicine for coronary heart disease can be classified according to the curative effect of the medicine and graded according to the harm degree of the side effect of the medicine, thereby facilitating the selection of doctors and improving the grading efficiency of the medicine. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 A flowchart of the method for grading and classifying the coronary heart disease medicine library based on the cardiology department is provided for an embodiment of the present application.

[0065] Figure 2 A flowchart of generating a standard coronary heart disease medicine term vector set is provided for an embodiment of the present application.

[0066] Figure 3 A flowchart of carrying out term frequency clustering on a single medicine effect term vector group set is provided for an embodiment of the present application.

[0067] Figure 4 A functional module diagram of the device for grading and classifying the coronary heart disease medicine library based on the cardiology department is provided for an embodiment of the present application.

[0068] The implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0069] It should be understood that the specific embodiments described herein are merely intended to explain the present application, and are not intended to limit the present application.

[0070] The embodiment of the present application provides a hierarchical classification management method based on a coronary heart disease drug library of cardiology. The execution subject of the hierarchical classification management method based on the coronary heart disease drug library of cardiology includes but is not limited to at least one of electronic devices such as a server, a terminal and the like which can be configured to execute the method provided by the embodiment of the present application. In other words, the hierarchical classification management method based on the coronary heart disease drug library of cardiology can be executed by software or hardware installed in a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to a single server, a server cluster, a cloud server or a cloud server cluster and the like. The server can be a stand-alone server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks (CDN), and big data and artificial intelligence platforms and the like basic cloud computing services.

[0071] Referring to Figure 1 Fig. 1 is a flowchart of a hierarchical classification management method based on a coronary heart disease drug library of cardiology provided by an embodiment of the present application. In the embodiment, the hierarchical classification management method based on the coronary heart disease drug library of cardiology includes:

[0072] S1, using a keyword matching algorithm to clean the word vector data of the pre-acquired historical coronary heart disease drug data, obtaining a standard coronary heart disease drug word vector set, and splitting the standard coronary heart disease drug word vector set into a single drug word vector group set according to the drug name.

[0073] In the embodiment of the present application, the historical coronary heart disease drug data includes the drug name, drug effect introduction and retrieved drug feedback report of each drug in the coronary heart disease drug library of cardiology, wherein the drug name is, for example, “aspirin enteric-coated tablets”, the drug effect introduction is, for example, “anti-platelet aggregation, adjusting blood lipids, stabilizing arterial plaque and controlling heart rate to reduce myocardial oxygen consumption”, and the drug feedback report includes “allergy, ulcer, vascular damage, ventilation and asthma, etc.”. The standard coronary heart disease drug word vector set refers to a vector set composed of word vectors of all standardized inertial disease drug words.

[0074] In the embodiment of the present application, referring to Figure 2 Fig. 2, the use of a keyword matching algorithm to clean the word vector data of the pre-acquired historical coronary heart disease drug data to obtain a standard coronary heart disease drug word vector set includes:

[0075] S21, performing text segmentation on the pre-acquired primary drug data to obtain a primary drug word set;

[0076] S23, screening out the garbled words and meaningless words from the primary drug word set to obtain a secondary drug word set, and vectorizing the secondary drug word set into a secondary drug word vector set;

[0077] S23, selecting a secondary drug word vector in the secondary drug word vector set as a target secondary drug word vector, and calculating a matching similarity between the target secondary drug word vector and each drug word vector in a preset drug word vector library:

[0078] S24, selecting a drug word vector corresponding to the maximum matching similarity from the drug word vector library as a target standard drug word vector, replacing the target secondary drug word vector with the target standard drug word vector to obtain a standard coronary heart disease drug word vector set.

[0079] In detail, the primary drug data obtained in advance can be subjected to text segmentation by using a bidirectional maximum matching algorithm to obtain a primary drug word set, and the secondary drug word set can be vectorized into a secondary drug word vector set by using a Word2vec model.

[0080] Specifically, the garbled words refer to, for example, garbled characters "&", "#" and the like, and the meaningless words refer to, for example, auxiliary words "de", "di" and the like.

[0081] In detail, the reciprocal of the Euclidean distance between the target secondary drug word vector and each drug word vector can be used as the matching similarity, so as to improve the representation of the matching similarity.

[0082] Specifically, the standard coronary heart disease drug word vector set is split into a single drug word vector group set according to the drug names, which means that a drug name is selected as a target drug name, the word vectors corresponding to the target drug name are filtered from the standard coronary heart disease drug word vector set to form a single drug word vector group, and all the single drug word vector groups are collected into a single drug word vector group set.

[0083] In the embodiment of the application, the historical coronary heart disease drug data obtained in advance is subjected to word vector data cleaning by using a keyword matching algorithm to obtain a standard coronary heart disease drug word vector set, which can reduce the types of words in the historical coronary heart disease drug data, standardize the representation of various synonymous words, reduce the data dimension, improve the calculation efficiency, and split the standard coronary heart disease drug word vector set into a single drug word vector group set according to the drug names, so as to facilitate subsequent classification and hierarchical management of each drug.

[0084] S2, extract a single drug effect word vector set and a single drug use feedback word vector set from the single drug word vector set, perform word frequency clustering on the single drug effect word vector set to obtain a plurality of drug effect classes, and perform word frequency clustering on the single drug use feedback word vector set to obtain a plurality of drug feedback level classes.

[0085] In the embodiment of the application, the single drug effect word vector set is a set of a plurality of single drug effect word vectors, and each single drug effect word vector set contains a word vector of all drug effect introduction words corresponding to a drug name of a coronary heart disease drug; and the single drug use feedback word vector set is a set of a plurality of single drug use feedback word vectors, and each single drug use feedback word vector set contains a word vector of all drug use feedback report words corresponding to a drug name of a coronary heart disease drug.

[0086] In the embodiment of the application, the extracting a single drug effect word vector set and a single drug use feedback word vector set from the single drug word vector set comprises: selecting a single drug word vector set in the single drug word vector set as a target single drug word vector set one by one, extracting a single drug effect word vector set and a single drug use feedback word vector set from the target single drug word vector set respectively, collecting all the single drug effect word vector sets into a single drug effect word vector set, and collecting all the single drug use feedback word vector sets into a single drug use feedback word vector set.

[0087] In the embodiment of the application, referring to Figure 3 The word frequency clustering on the single drug effect word vector set to obtain a plurality of drug effect classes comprises:

[0088] S31, selecting a single drug effect word vector set in the single drug effect word vector set as a target single drug effect word vector set one by one, performing vector weighted merging on the target single drug effect word vector set to obtain a weighted single drug effect word vector set;

[0089] S32, performing vector splicing on the weighted single drug effect word vector set to obtain a standard single drug effect sentence vector, and collecting all the standard single drug effect sentence vectors into a standard single drug effect sentence vector set;

[0090] S33, performing word frequency clustering on the standard single drug effect sentence vector set by using a preset word frequency distance algorithm to obtain a plurality of drug effect classes.

[0091] In detail, the vector-weighted combination of the target single-drug effect word vector set to obtain a weighted single-drug effect word vector set refers to counting the frequency of each single-drug effect word vector in the target single-drug effect word vector set, taking the frequency as the weight of the corresponding single-drug effect word vector, and collecting the weight and the corresponding single-drug effect word vector into the weighted single-drug effect word vector set.

[0092] Specifically, the vector splicing of the weighted single-drug effect word vector set to obtain a standard single-drug effect sentence vector refers to splicing each weighted single-drug effect word vector in the weighted single-drug effect word vector set into a standard single-drug effect sentence vector.

[0093] In the embodiment of the present application, the word frequency clustering of the standard single-drug effect sentence vector set by using a preset word frequency distance algorithm to obtain a plurality of drug effect classes comprises:

[0094] The standard single-drug effect sentence vector set is split into a plurality of single-drug effect sentence vector sets, and a primary effect center sentence vector of each single-drug effect sentence vector set is randomly selected;

[0095] The word frequency similarity algorithm is used to calculate the word frequency similarity between each standard single-drug effect sentence vector in the standard single-drug effect sentence vector set and each primary effect center sentence vector.

[0096]

[0097] Wherein, C refers to the word frequency similarity, p refers to the serial number of the word vector in the standard single-drug effect sentence vector, P refers to the total number of word vectors in the standard single-drug effect sentence vector, r refers to the serial number of the word vector in the primary effect center sentence vector, R refers to the total number of word vectors in the primary effect center sentence vector, f p refers to the weight of the pth word vector in the standard single-drug effect sentence vector, F p refers to the pth word vector in the standard single-drug effect sentence vector, E r refers to the rth word vector in the primary effect center sentence vector, e r refers to the weight of the rth word vector in the primary effect center sentence vector.

[0098] According to the word frequency similarity, each single-drug effect sentence vector set is updated into a corresponding secondary single-drug effect sentence vector set.

[0099] The secondary effect center sentence vector of each secondary single-drug effect sentence vector set is calculated, the center word frequency similarity between each secondary effect center sentence vector and the corresponding primary effect center sentence vector is calculated, and the sum of all center word frequency similarities is taken as a target center word frequency similarity.

[0100] updating each of the secondary single-drug effect sentence vector groups into corresponding drug effect classes according to the target center word frequency similarity.

[0101] Specifically, by calculating the word frequency similarity between each standard single-drug effect sentence vector in the standard single-drug effect sentence vector set and each primary effect center sentence vector by using the word frequency similarity algorithm, the similarity between the two sentence vectors can be determined according to the occurrence frequency of each word vector in the sentence vector and the vector distance.

[0102] In detail, updating each of the single-drug effect sentence vectors into corresponding secondary single-drug effect sentence vector groups according to the word frequency similarity refers to assigning each standard single-drug effect sentence vector in the standard single-drug effect sentence vector set to the single-drug effect sentence vector group corresponding to the primary effect center sentence vector with the maximum word frequency similarity.

[0103] In detail, the method of performing word frequency clustering on the single-drug effect word vector group set to obtain a plurality of drug effect classes is the same as the method of performing word frequency clustering on the single-drug effect word vector group set to obtain a plurality of drug effect classes in step S2 described above, which will not be repeated here.

[0104] In the embodiment of the present application, by performing word frequency clustering on the single-drug effect word vector group set to obtain a plurality of drug effect classes, and performing word frequency clustering on the single-drug feedback word vector group set to obtain a plurality of drug feedback level classes, each drug in the coronary heart disease drug library can be classified according to the effect of the drug and graded according to the drug feedback, thereby realizing the classified and graded management of the drug.

[0105] S3, obtaining a drug picture of a coronary heart disease drug to be graded, performing picture denoising and picture equalization on the drug picture to obtain an equalized drug picture, sequentially performing threshold picture segmentation and binarization operation on the equalized drug picture to obtain a standard drug picture, and extracting standard drug features from the standard drug picture.

[0106] In the embodiment of the present application, the coronary heart disease drug to be graded refers to a coronary heart disease drug in the cardiology department that needs to be managed by drug classification and grading, and the drug picture refers to a picture taken by a classification manager, which includes a tablet packaging of the coronary heart disease drug to be graded.

[0107] In the embodiment of the present application, performing picture denoising and picture equalization on the drug picture to obtain an equalized drug picture comprises:

[0108] performing a gray-scale operation on the drug picture to obtain a gray-scale drug picture;

[0109] performing a median filter operation and a bilateral filter operation on the gray-scale medicine picture in sequence to obtain a filtered medicine picture;

[0110] generating a primary gray-scale histogram of the filtered medicine picture, performing a probability equalization operation on the primary gray-scale histogram to obtain an equalized primary gray-scale histogram;

[0111] updating the filtered medicine picture by using the equalized primary gray-scale histogram to obtain an equalized medicine picture.

[0112] Specifically, the gray-scale medicine picture can be obtained by performing a gray-scale operation on the medicine picture by using a weighted average algorithm, that is, the gray-scale values of each pixel are obtained by performing a weighted summation on the chroma of the three-color pixels of each pixel in the medicine picture, and the medicine picture is gray-scaled into a gray-scale medicine picture according to the gray-scale values.

[0113] In detail, the gray-scale medicine picture can be filtered by using a median filter, and the gray-scale medicine picture filtered by the median filter can be filtered by using a bilateral filter to obtain a filtered medicine picture. The bilateral filter is a kind of nonlinear filtering method, which is a compromise processing combining the spatial proximity and the pixel value similarity of an image, and simultaneously considers the spatial domain information and the gray-scale similarity to achieve the purpose of edge-preserving denoising. The median filter is a kind of nonlinear digital filter technology, which is often used to remove noise in images or other signals.

[0114] Specifically, the primary gray-scale histogram can be obtained by performing a probability statistics on the distribution of the picture gray scale in the filtered medicine picture. The probability equalization operation on the primary gray-scale histogram to obtain an equalized primary gray-scale histogram refers to calculating the total number of pixels and the gray-scale distribution frequency of the primary gray-scale histogram, and mapping the normalized gray-scale distribution frequency to the filtered medicine picture to obtain an equalized primary gray-scale histogram.

[0115] Specifically, the equalized medicine picture is sequentially subjected to a threshold picture segmentation and a binarization operation to obtain a standard medicine picture, which includes:

[0116] generating a secondary gray-scale histogram of the equalized medicine picture, selecting the gray scales in the secondary gray-scale histogram as target gray scales one by one, and segmenting the equalized medicine picture into a target medicine picture and a target background picture according to the target gray scales;

[0117] The inter-gray variance between the target medicine picture and the target background picture is calculated by using the following inter-gray variance method:

[0118]

[0119] Wherein, T refers to the inter-grayscale variance, M refers to the pixel length of the secondary grayscale histogram, N refers to the pixel width of the secondary grayscale histogram, N0 refers to the total number of pixels of the target drug picture, N1 refers to the total number of pixels of the target background picture, n0 refers to the nth0 pixel in the target drug picture, n1 refers to the nth1 pixel in the target background picture, and g(·) is a grayscale value symbol;

[0120] The target drug picture corresponding to the maximum inter-grayscale variance is selected as a primary drug picture, and a binaryzation operation is performed on the primary drug picture to obtain a standard drug picture.

[0121] In detail, the step of dividing the balanced drug picture into a target drug picture and a target background picture according to the target grayscale level refers to taking the part of the balanced drug picture with a grayscale level less than or equal to the target grayscale level as the target background picture and taking the part of the balanced drug picture with a grayscale level greater than the target grayscale level as the target drug picture.

[0122] In the embodiment of the present application, the inter-grayscale variance between the target drug picture and the target background picture can be calculated by using the inter-grayscale variance method, the grayscale distance between each pixel of the target drug picture and the target background picture can be calculated, and the picture is divided according to the grayscale distance between the two regions, thereby improving the accuracy of picture segmentation.

[0123] Specifically, the step of extracting a standard drug feature from the standard drug picture refers to performing convolution or other operations on the standard drug picture to obtain the picture feature of the standard drug picture.

[0124] In the embodiment of the present application, the noise pixels in the drug picture and the shooting light spots caused by brightness reflection and the like can be removed by performing picture denoising and picture balancing on the drug picture to obtain a balanced drug picture, thereby retaining more picture details, the related region of the drug packaging box can be extracted from the drug picture by sequentially performing threshold picture segmentation and binaryzation operation on the balanced drug picture to obtain a standard drug picture, which facilitates subsequent feature extraction, the name of the drug can be recognized according to the packaging feature of the drug by extracting a standard drug feature from the standard drug picture, thereby facilitating drug classification.

[0125] S4, performing a feature picture segmentation operation on the standard drug picture to obtain a drug term picture and a drug code picture, extracting a standard drug term from the drug term picture, and extracting a standard drug code from the drug code picture.

[0126] The drug term picture refers to a picture part region of a term containing drug introduction in the standard drug picture, for example, a name term of a drug, and the drug code picture refers to a picture part region corresponding to a code of a drug in the standard drug picture.

[0127] In the embodiment of the present application, the feature picture segmentation operation is performed on the standard drug picture to obtain the drug term picture and the drug code picture, including:

[0128] A primary drug edge is detected from the standard drug picture, and a secondary drug edge is obtained by performing edge fitting on the primary drug edge;

[0129] The standard drug picture is rectified by using the secondary drug edge to obtain a rectified drug picture;

[0130] A rectified drug texture is extracted from the rectified drug picture, and drug term texture and drug code texture are obtained by performing texture matching classification on the rectified drug texture;

[0131] The drug term picture is extracted from the rectified drug picture according to the drug term texture, and the drug code picture is extracted from the rectified drug picture according to the drug code texture.

[0132] Specifically, the primary drug edge can be detected from the standard drug picture by using a linear filter, the secondary drug edge can be obtained by performing edge fitting on the primary drug edge by using a least square method, and the standard drug picture can be rectified by using Hough transform or Hough transform to rectify the secondary drug edge.

[0133] In detail, the rectified drug texture can be extracted from the rectified drug picture by using a neural network such as VGG-16, and the drug term texture and the drug code texture can be obtained by performing texture matching classification on the rectified drug texture by using a pooling layer.

[0134] In detail, the standard drug term is extracted from the drug term picture, including:

[0135] A layout edge is extracted from the drug term picture, and a layout term picture is obtained by performing layout segmentation on the drug term picture according to the layout edge;

[0136] A layout character atlas is obtained by performing character cutting on the layout term picture;

[0137] Text recognition is performed on the layout character pictures in the layout character atlas one by one to obtain the standard drug term.

[0138] Specifically, the edges of the publication can be extracted from the drug entry picture by using an edge detection operator such as sobel, the characters in the publication entry picture can be cut by using a vertical projection algorithm to obtain a publication character atlas, and the characters in the publication character atlas can be recognized one by one by using a trained convolutional neural network such as VGG-16 to obtain a standard drug entry.

[0139] In detail, the standard drug code is extracted from the drug code picture, including:

[0140] The drug code picture is vertically corrected to obtain a standard drug code picture;

[0141] The standard drug code picture is divided into barcode intervals to obtain a barcode interval set;

[0142] The barcode intervals in the barcode interval set are recognized one by one to obtain a primary drug code;

[0143] The primary drug code is converted into a binary code to obtain a standard drug code.

[0144] In detail, the standard drug code picture is divided into barcode intervals to obtain a barcode interval set, which means that the standard drug code picture is divided into a start symbol area, a left data symbol area, and a middle separator area according to the positions of the intervals in the standard drug code picture.

[0145] In the embodiment of the application, the standard drug entry is extracted from the drug entry picture, and the standard drug code is extracted from the drug code picture, so that more drug characteristics of the standard drug picture can be extracted, thereby improving the accuracy of drug name recognition.

[0146] S5, according to the standard drug feature, the standard drug entry and the standard drug code to determine the target drug name, using the drug effect class to determine the target drug class corresponding to the target drug name, using the drug feedback level class to determine the target drug level corresponding to the target drug name, according to the target drug class and the target drug level to the classified management of the drug to be classified coronary heart disease.

[0147] In the embodiment of the application, the target drug name is determined according to the standard drug feature, the standard drug entry and the standard drug code, including:

[0148] According to the standard drug feature, a picture drug name set is matched in a preset drug feature library,

[0149] From the single drug effect word vector group set, a word entry drug name set corresponding to the standard drug entry is matched;

[0150] querying the standard drug code corresponding coding drug name set from a preset drug code library;

[0151] The intersection of the picture drug name set, the entry drug name set and the coding drug name set is taken as a target drug name.

[0152] Specifically, the drug feature library refers to a feature library composed of picture features extracted from various pre-recorded drug packaging pictures, and the drug code library can be a national drug code base code.

[0153] In detail, determining the target drug category corresponding to the target drug name by using the drug effect class refers to taking the drug effect class corresponding to the target drug name as the target drug category, and determining the target drug level corresponding to the target drug name by using the drug feedback level class refers to taking the drug feedback level class corresponding to the target drug name as the target drug level.

[0154] In the embodiment of the application, by classifying and managing the coronary heart disease drugs to be classified according to the target drug category and the target drug level, the coronary heart disease drugs can be classified according to the efficacy of the drugs and classified according to the harm degree of the side effects of the drugs, thereby facilitating the selection of physicians and improving the classification efficiency of the drugs.

[0155] The embodiment of the application can reduce the types of words in the historical coronary heart disease drug data and standardize the expressions of various synonymous words by using a keyword matching algorithm to clean the word vector data of the pre-acquired historical coronary heart disease drug data, thereby reducing the data dimension and improving the calculation efficiency. By splitting the standard coronary heart disease drug word vector set into single drug word vector group sets according to the drug names, the subsequent classification and classification management of the drugs for each drug name can be facilitated. By performing word frequency clustering on the single drug effect word vector group set, a plurality of drug effect classes are obtained, and by performing word frequency clustering on the single drug use feedback word vector group set, a plurality of drug feedback level classes are obtained. The drugs in the coronary heart disease drug library can be classified according to the effects of the drugs and classified according to the drug feedback, thereby realizing the classification and classification management of the drugs.

[0156] By carrying out picture denoising and picture equalization on the medicine picture, an equalized medicine picture is obtained, which can remove noise pixels and shooting light spots caused by brightness reflection and the like in the medicine picture, thereby retaining more picture details, by sequentially carrying out threshold picture segmentation and binarization operation on the equalized medicine picture, a standard medicine picture is obtained, which can extract the relevant area of the medicine packaging box from the medicine picture, facilitating subsequent feature extraction, by extracting standard medicine features from the standard medicine picture, the name of the medicine can be recognized according to the packaging features of the medicine, thereby facilitating medicine classification, by extracting standard medicine terms from the medicine term picture and extracting standard medicine codes from the medicine code picture, more medicine features of the standard medicine picture can be extracted, thereby improving the accuracy of medicine name recognition, by classifying and managing the coronary heart disease medicine to be classified according to the target medicine category and the target medicine level, the coronary heart disease medicine can be classified according to the curative effect of the medicine and classified according to the harm degree of the side effect of the medicine, thereby facilitating the selection of physicians and improving the classification efficiency of the medicine. Therefore, the classification and management method based on the coronary heart disease medicine library of the cardiology department can solve the problem of low efficiency in medicine classification and management.

[0157] As Figure 4 shown, it is a functional module diagram of the classification and management device based on the coronary heart disease medicine library of the cardiology department provided by an embodiment of the present application.

[0158] The classification and management device based on the coronary heart disease medicine library of the cardiology department 100 can be installed in an electronic device. According to the functions realized, the classification and management device based on the coronary heart disease medicine library of the cardiology department 100 can include a data cleaning module 101, a term frequency clustering module 102, a primary feature extraction module 103, a secondary feature extraction module 104 and a medicine classification module 105. The modules of the present application can also be referred to as units, which refer to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0159] In this embodiment, the functions of each module / unit are as follows:

[0160] The data cleaning module 101 is used to clean the word vector data of the pre-acquired historical coronary heart disease medicine data by using a keyword matching algorithm, to obtain a standard coronary heart disease medicine word vector set, and to split the standard coronary heart disease medicine word vector set into single-medicine word vector group sets according to the names of the medicines;

[0161] The word frequency clustering module 102 is configured to extract a single-drug effect word vector set and a single-drug medication feedback word vector set from the single-drug word vector set, perform word frequency clustering on the single-drug effect word vector set to obtain a plurality of drug effect classes, and perform word frequency clustering on the single-drug medication feedback word vector set to obtain a plurality of drug feedback level classes.

[0162] The primary feature extraction module 103 is configured to obtain a drug picture of a coronary heart disease drug to be classified, perform picture denoising and picture equalization on the drug picture to obtain an equalized drug picture, sequentially perform threshold picture segmentation and binarization on the equalized drug picture to obtain a standard drug picture, and extract a standard drug feature from the standard drug picture. The sequentially performing threshold picture segmentation and binarization on the equalized drug picture to obtain a standard drug picture includes: generating a secondary gray level histogram of the equalized drug picture, selecting a gray level in the secondary gray level histogram as a target gray level one by one, and segmenting the equalized drug picture into a target drug picture and a target background picture according to the target gray level. The gray level interval variance method is used to calculate the gray level interval variance between the target drug picture and the target background picture as follows:

[0163]

[0164] wherein T represents the gray level interval variance, M represents the pixel length of the secondary gray level histogram, N represents the pixel width of the secondary gray level histogram, N0 represents the total number of pixels of the target drug picture, N1 represents the total number of pixels of the target background picture, n0 represents the nth pixel in the target drug picture, n1 represents the nth pixel in the target background picture, and g(·) is a gray value symbol; the target drug picture corresponding to the maximum gray level interval variance is selected as a primary drug picture, and the primary drug picture is subjected to binarization to obtain a standard drug picture.

[0165] The secondary feature extraction module 104 is configured to perform feature picture segmentation on the standard drug picture to obtain a drug term picture and a drug code picture, extract a standard drug term from the drug term picture, and extract a standard drug code from the drug code picture.

[0166] The drug classification module 105 is configured to determine a target drug name according to the standard drug feature, the standard drug term, and the standard drug code, determine a target drug category corresponding to the target drug name by using the drug effect class, determine a target drug level corresponding to the target drug name by using the drug feedback level class, and perform classified management on the coronary heart disease drug to be classified according to the target drug category and the target drug level.

[0167] In detail, each module in the device 100 for hierarchical classification management of the coronary heart disease drug library in the embodiments of the present application adopts the same technical means as the method for hierarchical classification management of the coronary heart disease drug library in the above-mentioned Figures 1 to 3 and can produce the same technical effects, which will not be described here.

[0168] In the several embodiments of the present application, it should be understood that the disclosed device, apparatus, and method can be implemented in other ways. For example, the device embodiments described above are merely schematic. The division of the modules is merely a logical function division. In actual implementation, another division mode can be adopted.

[0169] The modules described as separated components can or can not be physically separated, and the components displayed as modules can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. According to actual needs, some or all of the modules can be selected to achieve the purpose of the embodiments.

[0170] In addition, each functional module in the embodiments of the present application can be integrated in one processing unit, or each unit can exist physically independently, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware, or in the form of hardware plus software function modules.

[0171] It is obvious for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application.

[0172] Therefore, the embodiments should be regarded as exemplary and non-limiting, and the scope of the present application is defined by the appended claims rather than the above description, and all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application. Any reference signs in the claims should not be regarded as limiting the claims to which they relate.

[0173] The embodiments of the present application can acquire and process related data based on artificial intelligence technology. Artificial intelligence (AI) is the theory, method, technology and application system for using digital computers or digital computer-controlled machines to simulate, extend and expand human intelligence, perceive environment, acquire knowledge and use knowledge to obtain optimal results.

[0174] Furthermore, the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural and vice-versa, unless the context clearly requires these exclusions. The mere fact that different features are recited in mutually different dependent claims does not indicate that the

[0175] Finally, it should be noted that the above-mentioned embodiments illustrate rather than limit the application, since the scope of the application is determined by the appended claims.

Claims

1. A hierarchical classification management method based on an intracardiac coronary heart disease drug library, characterized by, The method comprises: S1: using a keyword matching algorithm to clean the word vector data of the pre-acquired historical coronary heart disease drug data, to obtain a standard coronary heart disease drug word vector set, and splitting the standard coronary heart disease drug word vector set into single-drug word vector groups according to drug names; S2: extracting single-drug effect word vector groups and single-drug medication feedback word vector groups from the single-drug word vector groups, performing word frequency clustering on the single-drug effect word vector groups to obtain multiple drug effect categories, and performing word frequency clustering on the single-drug medication feedback word vector groups to obtain multiple drug feedback level categories; S3: acquiring a drug picture of a coronary heart disease drug to be classified, performing picture denoising and picture equalization on the drug picture to obtain an equalized drug picture, sequentially performing threshold picture segmentation and binarization operations on the equalized drug picture to obtain a standard drug picture, and extracting a standard drug feature from the standard drug picture, wherein the sequentially performing threshold picture segmentation and binarization operations on the equalized drug picture to obtain a standard drug picture comprises: S31: generating a secondary gray level histogram of the equalized drug picture, selecting gray levels in the secondary gray level histogram as target gray levels one by one, and segmenting the equalized drug picture into a target drug picture and a target background picture according to the target gray levels; S32: selecting a target drug picture corresponding to the maximum gray level interval variance as a primary drug picture, and performing binarization operation on the primary drug picture to obtain a standard drug picture; S4: performing feature picture segmentation operation on the standard drug picture to obtain a drug term picture and a drug code picture, extracting a standard drug term from the drug term picture, and extracting a standard drug code from the drug code picture; S5: determining a target drug name according to the standard drug feature, the standard drug term, and the standard drug code, determining a target drug category corresponding to the target drug name by using the drug effect categories, determining a target drug level corresponding to the target drug name by using the drug feedback level categories, and performing hierarchical management on the coronary heart disease drug to be classified according to the target drug category and the target drug level.

2. The hierarchical classification management method based on the library of cardiology drugs for coronary heart disease according to claim 1, characterized in that, The method comprises: performing text segmentation on the pre-acquired primary drug data to obtain a primary drug word set; removing random code words and meaningless words from the primary drug word set to obtain a secondary drug word set, and vectorizing the secondary drug word set into a secondary drug word vector set; selecting secondary drug word vectors in the secondary drug word vector set as target secondary drug word vectors one by one, and calculating the matching similarity of the target secondary drug word vectors with each drug word vector in a preset drug word vector library: Select the drug word vector corresponding to the maximum matching similarity from the drug word vector library as a target standard drug word vector, replace the target secondary drug word vector with the target standard drug word vector, and obtain a standard coronary heart disease drug word vector set.

3. The hierarchical classification and management method based on a coronary heart disease drug library in cardiology as described in claim 1, characterized in that, The word frequency clustering on the single-drug effect word vector group set obtains a plurality of drug effect classes, including: Select a single-drug effect word vector group in the single-drug effect word vector group set as a target single-drug effect word vector group, perform vector weighted merging on the target single-drug effect word vector group, and obtain a weighted single-drug effect word vector group; Perform vector splicing on the weighted single-drug effect word vector group to obtain a standard single-drug effect sentence vector, and collect all the standard single-drug effect sentence vectors into a standard single-drug effect sentence vector set; Perform word frequency clustering on the standard single-drug effect sentence vector set by using a preset word frequency distance algorithm to obtain a plurality of drug effect classes.

4. The hierarchical classification management method based on the library of cardiology drugs for coronary heart disease according to claim 3, characterized in that, The word frequency clustering on the standard single-drug effect sentence vector set by using a preset word frequency distance algorithm obtains a plurality of drug effect classes, including: Split the standard single-drug effect sentence vector set into a plurality of single-drug effect sentence vector groups, and randomly select a primary effect center sentence vector of each single-drug effect sentence vector group; Calculate the word frequency similarity between each standard single-drug effect sentence vector in the standard single-drug effect sentence vector set and each primary effect center sentence vector by using the following word frequency similarity algorithm: ； in, This refers to the word frequency similarity. This refers to the index of the word vector in the standard single-drug effect sentence vector. This refers to the total number of word vectors in the standard single-drug effect sentence vector. This refers to the index of the word vector in the central sentence vector of the primary effect. This refers to the total number of word vectors in the central sentence vector of the primary effect. This refers to the first sentence in the standard single-drug effect vector. The weights of each word vector. This refers to the first sentence in the standard single-drug effect vector. Word vectors, This refers to the first sentence in the vector of the primary effect center sentence. Word vectors, This refers to the first sentence in the vector of the primary effect center sentence. The weights of each word vector; Update each single-drug effect sentence vector group into a corresponding secondary single-drug effect sentence vector group according to the word frequency similarity; Calculate a secondary effect center sentence vector of each secondary single-drug effect sentence vector group, calculate the center word frequency similarity between each secondary effect center sentence vector and the corresponding primary effect center sentence vector, and take the sum of all center word frequency similarities as a target center word frequency similarity; Iteratively update each secondary single-drug effect sentence vector group into a corresponding drug effect class according to the target center word frequency similarity.

5. The hierarchical classification management method based on the library of cardiology drugs for coronary heart disease according to claim 1, characterized in that, The picture denoising and picture equalization on the drug picture obtains an equalized drug picture, including: Perform a grayscale operation on the drug picture to obtain a grayscale drug picture; Perform a median filter operation and a bilateral filter operation on the grayscale drug picture in sequence to obtain a filtered drug picture; Generate a primary grayscale histogram of the filtered drug picture, perform a probability equalization operation on the primary grayscale histogram to obtain an equalized primary grayscale histogram, and update the filtered drug picture by using the equalized primary grayscale histogram to obtain an equalized drug picture. The generation of the secondary grayscale histogram of the equalized drug picture, the selection of a grayscale level in the secondary grayscale histogram as a target grayscale level, and the segmentation of the equalized drug picture into a target drug picture and a target background picture according to the target grayscale level, include:

6. The hierarchical classification management method based on the library of cardiology drugs for coronary heart disease according to claim 1, characterized in that, Calculate the inter-grayscale variance between the target drug picture and the target background picture by using the following inter-grayscale variance method: The feature picture segmentation operation on the standard drug picture obtains a drug entry picture and a drug code picture, including: ； wherein, denotes the inter-gray variance, denotes the pixel length of the secondary gray histogram, denotes the pixel width of the secondary gray histogram, denotes the total number of pixels of the target drug picture, denotes the total number of pixels of the target background picture, denotes the i-th pixel in the target drug picture, denotes the i-th pixel in the target drug picture, denotes the i-th pixel in the target background picture, denotes the i-th pixel in the target background picture, is the gray value symbol.

7. The hierarchical classification and management method based on a coronary heart disease drug library in cardiology as described in claim 1, characterized in that, ​ Detecting a primary drug edge from the standard drug picture, performing edge fitting on the primary drug edge to obtain a secondary drug edge; Performing tilt correction on the standard drug picture by using the secondary drug edge to obtain a corrected drug picture; Extracting a corrected drug texture from the corrected drug picture, and performing texture matching classification on the corrected drug texture to obtain a drug entry texture and a drug code texture; Extracting a drug entry picture from the corrected drug picture according to the drug entry texture, and extracting a drug code picture from the corrected drug picture according to the drug code texture.

8. The hierarchical classification management method based on the library of cardiology drugs for coronary heart disease according to claim 7, characterized in that, The extracting a standard drug entry from the drug entry picture comprises: Extracting a layout edge from the drug entry picture, performing layout segmentation on the drug entry picture according to the layout edge to obtain a layout entry picture; Performing character cutting on the layout entry picture to obtain a layout character atlas; Performing character recognition on the layout character pictures in the layout character atlas one by one to obtain a standard drug entry.

9. The hierarchical classification management method based on the library of cardiology drugs for coronary heart disease according to claim 7, characterized in that, The extracting a standard drug code from the drug code picture comprises: Performing vertical correction on the drug code picture to obtain a standard drug code picture; Performing barcode partitioning on the standard drug code picture to obtain a barcode interval set; Performing character recognition on the barcode intervals in the barcode interval set one by one to obtain a primary drug code; Performing binary conversion on the primary drug code to obtain a standard drug code.

10. The method for hierarchical classification management of the library of drugs for coronary heart disease in cardiology department according to claim 1, characterized in that, The determining a target drug name according to the standard drug feature, the standard drug entry, and the standard drug code comprises: Matching a picture drug name set from the standard drug feature in a preset drug feature library, Matching a entry drug name set corresponding to the standard drug entry from the single-drug effect word vector group set; Querying a code drug name set corresponding to the standard drug code from a preset drug code library; Taking the intersection of the picture drug name set, the entry drug name set, and the code drug name set as the target drug name.

11. A device for hierarchical classification management based on a library of drugs for coronary heart disease in cardiology, characterized by, The device comprises: A data cleaning module configured to perform word vector data cleaning on historical coronary heart disease drug data pre-acquired by using a keyword matching algorithm to obtain a standard coronary heart disease drug word vector set, and split the standard coronary heart disease drug word vector set into single-drug word vector groups according to drug names; A word frequency clustering module configured to extract single-drug effect word vector groups and single-drug medication feedback word vector groups from the single-drug word vector groups, perform word frequency clustering on the single-drug effect word vector groups to obtain multiple drug effect categories, and perform word frequency clustering on the single-drug medication feedback word vector groups to obtain multiple drug feedback level categories; The primary feature extraction module is configured to obtain a medicine picture of a medicine to be classified, perform picture denoising and picture equalization on the medicine picture to obtain an equalized medicine picture, perform threshold picture segmentation and binarization on the equalized medicine picture in sequence to obtain a standard medicine picture, and extract standard medicine features from the standard medicine picture. The maximum target medicine picture corresponding to the inter-grayscale variance is taken as a primary medicine picture, and the primary medicine picture is subjected to binarization to obtain a standard medicine picture. The secondary feature extraction module is configured to perform feature picture segmentation on the standard medicine picture to obtain a medicine term picture and a medicine code picture, extract standard medicine terms from the medicine term picture, and extract standard medicine codes from the medicine code picture. The medicine classification module is configured to determine a target medicine name according to the standard medicine features, the standard medicine terms, and the standard medicine codes, determine a target medicine category corresponding to the target medicine name by using the medicine effect class, determine a target medicine level corresponding to the target medicine name by using the medicine feedback level class, and perform classified management on the medicine to be classified according to the target medicine category and the target medicine level.

12. The hierarchical classification management device based on the library of drugs for coronary heart disease in cardiology according to claim 11, characterized in that, The secondary grayscale histogram of the equalized medicine picture is generated, the grayscale levels in the secondary grayscale histogram are selected one by one as target grayscale levels, and the equalized medicine picture is segmented into a target medicine picture and a target background picture according to the target grayscale levels. The inter-grayscale variance between the target medicine picture and the target background picture is calculated by using the following inter-grayscale variance method: ； wherein, denotes the inter-gray variance, denotes the pixel length of the secondary gray histogram, denotes the pixel width of the secondary gray histogram, denotes the total number of pixels of the target drug picture, denotes the total number of pixels of the target background picture, denotes the i-th pixel in the target drug picture, denotes the i-th pixel in the target drug picture, denotes the i-th pixel in the target background picture, denotes the i-th pixel in the target background picture, is the gray value symbol.

Images