A method, device and storage medium for identifying the name of a drug to be identified

Abstract

The present invention discloses a method, device and storage medium for identifying the name of a drug to be identified. The method for identifying the name of a drug to be identified includes the following steps: obtaining a preliminary identified drug name plus specifications; preliminarily retrieving and extracting similar drug names plus specifications; normalizing word segmentation processing; processing the normalized word segmentation results by a set difference quantization algorithm to obtain name similarity; selecting a similar drug name with the highest name similarity as the drug with the same name as the drug to be identified, and outputting it as a recommended drug name. The present invention utilizes word segmentation technology and semantic similarity technology, combined with a quantization method of the symmetric difference of two sets, to obtain the similarity between the preliminary identified drug name plus specifications and the standard drug name plus specification name in the local drug catalog database, thereby achieving rapid identification of drug names and recommending the same drug, solving the problem of time-consuming and labor-intensive identification and recommendation of the same drug in the prior art.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and in particular to a method, device and storage medium for identifying the name of a drug to be identified. Background Art

[0002] With the advancement of artificial intelligence and natural language processing technology, more and more research is devoted to using methods such as text mining, machine learning and deep learning to extract information and build models from massive amounts of drug data to accurately identify drugs and give personalized recommendations. Driven by medical informatization and big data technology, it is expected to provide more intelligent and precise solutions for medical and health management in the future.

[0003] At present, the traditional method of identifying whether it is the same drug is manual identification or fuzzy search through computer function interface. It is affected by the given word order and the influence of uppercase and lowercase letters, Chinese and English, and is slow, inefficient, and easy to miss. In order to accurately locate, it is necessary to repeatedly try different word orders or writing methods.

[0004] For example, a patent application with publication number: CN115481623A discloses a specific entity name recognition optimization method, which includes: training a bi l stm-crf model, extracting possible entities from the text input by the user, merging the specific entity name components through syntactic analysis and conflict detection, obtaining all possible candidate names, and then deduplicating the final result after database verification.

[0005] For example, the name recognition and association recommendation method, device, computer equipment and storage medium announced in the invention patent with announcement number: CN112966504B include: obtaining the name of the accessory, identifying the associated name of the accessory name, aggregating the associated names to obtain an associated list, receiving the associated name selected by the user and setting it as the target name; obtaining at least one accessory information from the transaction system according to the target name and aggregating it to form an accessory list, receiving the accessory information selected by the user and setting it as the transaction information; setting the accessory information selected by the user as the target information, obtaining the accessory information having an accessory association relationship with the target information and aggregating it to form a recommendation list, and setting the accessory information selected by the user as the transaction information; aggregating the transaction information to form order information, and sending the order information to the transaction system.

[0006] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems:

[0007] At present, in the current scenario of explosive growth in data volume, traditional technical methods can no longer meet the needs of production and life, and there is a problem that the identification and recommendation of the same drug is time-consuming and labor-intensive. Summary of the invention

[0008] The embodiments of the present application solve the problem of time-consuming and labor-intensive identification and recommendation of the same drug in the prior art by providing a method, device and storage medium for identifying the name of a drug to be identified, and achieve rapid identification of drug names and recommendation of the same drug.

[0009] The embodiment of the present application provides a method for identifying the name of a drug to be identified, comprising the following steps: obtaining image information of the drug to be identified, extracting text from the image information, and obtaining a preliminary identified drug name plus specifications; connecting to a local drug catalog database, preliminarily searching and extracting drug names and specifications that are similar to the preliminary identified drug name plus specifications; performing normalized word segmentation processing on the preliminary identified drug name plus specifications and similar drug names plus specifications; processing the normalized word segmentation results through a set difference quantization algorithm to obtain the name similarity between the preliminary identified drug name plus specifications and similar drug names plus specifications; selecting a similar drug name with the highest name similarity as the drug with the same name as the drug to be identified, and outputting it as a recommended drug name.

[0010] Furthermore, the word segmentation processing includes the following steps: performing word segmentation processing on the drug name plus specification based on the drug name, drug specification information, and drug measurement unit; performing secondary word segmentation processing on the word segmentation results of the drug name, drug specification information, and drug measurement unit; performing standardization and normalization processing on the word segmentation results obtained from the two word segmentation processings; and obtaining a normalized word segmentation result.

[0011] Furthermore, the method for obtaining the name similarity is: extracting the normalized segmentation results corresponding to the preliminary identified drug name plus specification and the similar drug name plus specification; comparing the normalized segmentation results of the two groups, finding the same segmentations and removing them, and obtaining their respective drug segmentation sets; counting the two groups of drug segmentation sets to obtain the number of different segmentations, recorded as set difference; converting the set difference into name similarity.

[0012] Furthermore, the drug segmentation set also includes the following processing steps: in the drug segmentation set, the name segmentation is arranged in the front position, and the other segmentations are arranged in the back position, so as to obtain the drug segmentation set elements arranged in an orderly manner; if the difference is caused by the drug segmentation set element Key, the difference is corrected and adjusted; the formula for correcting the difference is:

[0013] S diff =[ABS(AB)]+1*M+[ABS(BA)]+1*N

[0014] In the formula, S diffis the difference degree, A is the set of drug segmentations of the preliminary identified drug names and specifications, B is the set of drug segmentations of similar drug names and specifications, the ABS function is used to obtain the absolute value of the difference in the number of elements in set A and set B, M is the correction factor A, and N is the correction factor B; when ABS(AB)>0, then M=1, otherwise M=0, ABS(AB) represents the absolute difference between set A and set B, when ABS(BA)>0, then N=1, otherwise N=0, ABS(BA) represents the absolute difference between set B and set A.

[0015] Furthermore, the formula for converting the set difference into the name similarity is:

[0016]

[0017] In the formula, S i (AB) is the name similarity between the preliminarily identified drug name plus specification and similar drug name plus specification, A is the drug word set corresponding to the preliminarily identified drug name plus specification, B is the drug word set corresponding to the similar drug name plus specification, and the count(.) function is used to count the number of elements in the set.

[0018] Furthermore, the drug information in the local drug catalog database will be marked after being segmented, and the segmentation results corresponding to the drug information will be stored, and the marked drug information will be automatically split again during the next polling.

[0019] Furthermore, after obtaining the name similarity between the preliminary identified drug name plus specification and the similar drug name plus specification, the method also includes the following steps: extracting drug ingredient information based on the image information of the drug to be identified, and normalizing the ingredient information to obtain a set of drug ingredient segmentations to be identified; constructing a first knowledge graph based on the set of drug ingredient segmentations to be identified; extracting the drug ingredient information corresponding to the similar drug names plus specifications that have been preliminarily retrieved and extracted through a natural language algorithm, and normalizing the extracted drug ingredient information to obtain a set of similar drug ingredient segmentations; constructing a second knowledge graph based on the set of similar drug ingredient segmentations; comparing the sub-graphs of the first knowledge graph and the second knowledge graph through a sub-graph matching algorithm to obtain the similarity of the drug ingredient knowledge graph; when there are multiple similar drugs with the same name similarity, comparing the similarity of the drug ingredient knowledge graphs, and selecting the similar drug name corresponding to the highest value of the drug ingredient knowledge graph similarity as the recommended drug name output.

[0020] Furthermore, the drug ingredient knowledge graph similarity can also be used to derive other drug names based on ingredient information and specification information when there are multiple drug names for the same drug. The specific method is: obtain drug specification information 1 of the drug to be identified, and simultaneously obtain drug specification information 2 corresponding to a similar drug name plus specification, and perform word segmentation processing on drug specification information 1 and drug specification information 2 to obtain specification word segmentation result 1 and specification word segmentation result 2; sort the drug ingredient knowledge graph similarities between the drug to be identified and similar drugs from high to low; select similar drugs corresponding to the first K drug ingredient knowledge graph similarities, K is the ranking of the drug ingredient knowledge graph similarity, and K is a natural number, and compare the specification word segmentation result 1 of the drug to be identified and the specification word segmentation result 2 of the similar drug; obtain the similar drug with the highest similarity between specification word segmentation result 1 and specification word segmentation result 2, and record it as the drug alias recommendation output.

[0021] The embodiment of the present application provides a device for identifying the name of a drug to be identified, including: a basic information acquisition module, a similar name retrieval module, a word segmentation processing module, a name similarity analysis module and a recommendation output module: the basic information acquisition module is used to obtain image information of the drug to be identified, extract text from the image information, and obtain a preliminary identified drug name plus specifications; the similar name retrieval module is used to connect to a local drug catalog database, preliminarily retrieve and extract drug names and specifications that are similar to the preliminary identified drug name plus specifications; the word segmentation processing module is used to perform normalized word segmentation processing on the preliminary identified drug name plus specifications and similar drug names and specifications; the name similarity analysis module is used to process the normalized word segmentation results through a set difference quantization algorithm to obtain the name similarity of the preliminary identified drug name plus specifications and similar drug names plus specifications; the recommendation output module is used to select a similar drug name with the highest name similarity as the drug with the same name as the drug to be identified, and output it as a recommended drug name.

[0022] An embodiment of the present application provides a computer-readable storage medium for storing a program, which, when executed by a processor, implements a method for identifying the name of a drug to be identified.

[0023] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0024] 1. By utilizing word segmentation technology and combining it with the quantitative method of the symmetric difference of two sets, we can obtain the preliminary identification of the similarity between the drug name plus specification and the standard drug name plus specification name in the local drug catalog database, and then quickly identify the drug name to recommend the same drug, realize standardized drug recommendation and retrieval, and effectively solve the problem of time-consuming and labor-intensive identification and recommendation of the same drug in the existing technology.

[0025] 2. By analyzing the knowledge graph based on the ingredient information of the drug, similar drugs are found based on the knowledge graph, and then the similarity between each drug and the drug to be identified is compared, thereby achieving secondary verification to identify whether the drugs belong to the same drug, thereby increasing the accuracy of drug recommendations.

[0026] 3. By adopting intelligent drug recognition technology, the computer can automatically perform concurrent background tasks and asynchronous data processing tasks, thereby shortening the data processing response time and improving the satisfaction of pharmaceutical companies, which in turn increases customer stickiness and enhances its own advantages in bidding for commercial cooperation. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A flow chart of a method for identifying the name of a drug to be identified provided in an embodiment of the present application;

[0028] Figure 2 A structural diagram of a device for identifying the name of a drug to be identified provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The embodiments of the present application solve the problem of time-consuming and labor-intensive identification and recommendation of the same drug in the prior art by providing a method, device and storage medium for identifying the name of a drug to be identified. By utilizing word segmentation technology and combining it with a quantification method of the symmetric difference of two sets, the similarity between the preliminary identified drug name plus specification and the standard drug name plus specification name in a local drug catalog database is obtained, thereby quickly identifying the drug name and recommending the same drug.

[0030] The technical solution in the embodiment of the present application is to solve the above-mentioned problem that the identification and recommendation of the same drug is time-consuming and labor-intensive. The overall idea is as follows:

[0031] By integrating the circulation information of numerous pharmaceutical distribution companies, using the normalized word segmentation method of pharmaceutical names plus specifications, combining NLP technology, and using a set element difference quantification model for drug name recognition and cleaning recommendations, we can quickly identify drug names and recommend the same drug.

[0032] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0033] like Figure 1As shown, it is a flow chart of the method for identifying the name of a drug to be identified provided in an embodiment of the present application. The method is applied to a device for identifying the name of a drug to be identified, and the method includes the following steps: obtaining image information of the drug to be identified, extracting text from the image information, and obtaining a preliminary identified drug name plus specifications; connecting to a local drug catalog database, and preliminarily searching and extracting drug names plus specifications that are similar to the preliminary identified drug name plus specifications; performing normalized word segmentation processing on the preliminary identified drug name plus specifications and similar drug names plus specifications; processing the normalized word segmentation results through a set difference quantization algorithm, and obtaining the name similarity between the preliminary identified drug name plus specifications and similar drug names plus specifications; selecting a similar drug name with the highest name similarity as the drug with the same name as the drug to be identified, and outputting it as a recommended drug name.

[0034] In this embodiment, optical character recognition (OCR) technology can be used to extract text from image information. Optical character recognition (OCR) technology is a technology that converts text in an image into editable and searchable text. The specific image text extraction process is: first, the image information of the drug to be identified is preprocessed to improve the accuracy of OCR, wherein the preprocessing includes operations such as noise removal, image enhancement and binarization. Text detection is performed on the preprocessed image using an object detection algorithm based on machine learning or deep learning to determine the position and bounding box of the text in the image. For each text area detected, an OCR engine is applied for text recognition, and the text area is converted into editable text and output, wherein the OCR engine can be based on traditional machine learning methods, such as support vector machines (SVM) or deep learning-based methods, such as recurrent neural networks (RNN) or transformers. The OCR output is post-processed to improve recognition accuracy and completeness. The post-processing steps include operations such as text correction, language model correction and text normalization. Through the above steps, the name and specification information of the drug are extracted from the image information of the drug to be identified, and the preliminary identification drug name plus specification is output.

[0035] Furthermore, the word segmentation processing includes the following steps: performing word segmentation processing on the drug name plus specification based on the drug name, drug specification information, and drug measurement unit; performing secondary word segmentation processing on the word segmentation results of the drug name, drug specification information, and drug measurement unit; performing standardization and normalization processing on the word segmentation results obtained from the two word segmentation processings; and obtaining a normalized word segmentation result.

[0036] Specifically, the drug name plus specification is segmented according to the drug name, drug specification information, and drug measurement unit to obtain segmentation results such as [Bai San Ping 10mg*5 tablets], [Bai San Ping tablets 10 mg*5], and [Bai San Ping tablets 10mgX5]; the segmentation results of the drug name, drug specification information, and drug measurement unit are subjected to secondary segmentation processing to obtain the segmentation result of [Bai San Ping, 10mg, 5, tablets].

[0037] In the present embodiment, the drug name plus specification can be regarded as a text language of a special field, which is a text expression with clear meanings combined by Chinese characters, letters, characters and numbers. It is composed of a plurality of words with certain meanings, which can be separated into a plurality of independent words (single-word words or multi-word words). The same drug (including drug name and specification) can have different expressions, such as [Bai San Ping 10mg*5 tablets], [Bai San Ping tablets 10 mg*5], [Bai San Ping tablets 10mgX5], and they all have the same meaning. Milligrams and mg are just the difference between Chinese and English here, and there is no essential difference in the specific drug expression. This expression method without semantic distinction is called semantically identical or similar, which is regarded as common sense in drug identification. To realize the symmetric difference calculation of the set, it is first necessary to realize the normalization of the set elements, and the expression methods with the same meaning use the same set of rules. Contents and categories of the word segmentation dictionary conversion table: Numbers: All places where numbers appear are represented by single-byte numbers, and those that are not single-byte must be converted to single-byte; Letters: All places where English or Greek letters appear are represented by lowercase English or Greek letters, and uppercase or double-byte letters must be converted to lowercase single-byte;

[0038] Other symbols: The single-byte decimal point remains unchanged; other symbols are treated uniformly as cutting characters; Chinese measurement units: All are converted into English or Greek single-byte characters (one character or multiple characters).

[0039] Specifically, there are special normalization methods for drug names and specifications. Most drug names are expressed as product names, but some commercial ones use chemical names, supplemented by specifications and manufacturers. In order to achieve the normalization effect, based on the meaning of words, semantics, and drug characteristics, some indivisible or equivalent or synonymous correspondences are pre-set according to expert knowledge, and the drug names are segmented according to the principle of maximum length priority.

[0040] There are two ways to initialize the word segmentation database. One is to manually review and specify the general initial word segmentation catalog according to the drug catalog released by the National Drug Administration; the other is to add the drug name word segmentation database through front-end manual annotation and machine learning. The specific method is: all query terms that cannot be automatically identified but can be retrieved through manual word segmentation are continuously enriched to the word segmentation database.

[0041] Remove invalid words: For example, "large package", "new package", "joint brand", "collective purchase", etc., common words in non-standard names that are irrelevant to the meaning of the drug itself. Removing these words can reduce the redundancy of word segmentation results and improve the efficiency of subsequent processing. When defining, if the original word = "collective purchase", then the target word = '', empty, and so on. This type of words is continuously accumulated and improved in business operations.

[0042] Stemming: Stemming is the process of reducing different forms of a word to its basic form or root. For example, "medical cotton wool" is defined as "cotton wool" and "medical mask" is defined as "mask". This method helps normalize words with the same or similar meaning to the same form.

[0043] Lemma restoration: Lemma restoration is the process of restoring a word to its standard form in the dictionary. Lemma restoration is to analyze the characteristics of the drug name for definition, which has the particularity of the industry's textual expression. For example, "over-the-counter drugs" is defined as "OTC".

[0044] Equivalent synonym replacement: For words with similar meanings, synonym replacement can be used to normalize them. For example, "desogestrel ethinyl estradiol" and "mammalon" are replaced with a unified representation.

[0045] Part-of-speech tagging and normalization: Normalization is performed based on the word's part-of-speech information. For example, all nouns can be converted into a unified representation, or words can be grouped and normalized based on their part-of-speech. For example, in the normalized segmentation of the name of a pharmaceutical factory, "pharmaceutical factory" and "pharmaceutical company" are unified into "pharmaceutical factory", and "clinic", "outpatient department", and "consulting room" are unified into "consulting room".

[0046] Undefined words: There are many types of drugs, and new products are always released. Therefore, there will always be longer names that are not within the defined range and cannot be split. In this case, the word segmentation rule of "retain 6 characters when truncation is not possible" will be adopted, and will be included in machine learning or manual assisted annotation later. This is data training dependency.

[0047] Rule-based normalization: normalize words according to specific rules or templates. These rules can be based on linguistic knowledge or corpus statistics.

[0048] It should be noted that during the normalization process, it may be necessary to retain some words or features that have an important impact on the meaning of the text to avoid information loss due to over-normalization.

[0049] Furthermore, the method for obtaining name similarity is: extract the normalized segmentation results corresponding to the preliminary identified drug name plus specification and similar drug name plus specification; compare the normalized segmentation results of the two groups, find the same segmentations and remove them, and obtain their respective drug segmentation sets; count the two groups of drug segmentation sets to obtain the number of different segmentations, recorded as set difference; convert the set difference into name similarity.

[0050] In this embodiment, the segmentation structure of the drug standard name plus specification can be regarded as set A, and the segmentation structure of the non-standard drug name plus specification can be regarded as set B. In set A and set B, regardless of the position of the elements, the same elements are removed, and the remaining part [A 差 ]∪[B 差 ], the smaller the union, the higher the similarity between A and B. 差 ]∪[B 差 ] The number of elements is 0, that is, the drugs corresponding to set A and set B are considered to be the same drug

[0051] When the combination of name, specification and measurement unit of the drug information to be identified is not completely consistent with the standard drug information, but the segmentation results are completely consistent, the difference degree = 0. For example:

[0052] Original drug information form:

[0053]

[0054] Drug information word list:

[0055]

[0056] Among them, the standard drug information does not mark the attribute of "manufacturer", so when using the parameter, the "manufacturer" is not involved in the recognition rule. Among them, the drug word set is A = {throat lozenges, 28g, about, 10, grains, paper, box}, B = {throat lozenges, 28g, about, 10, grains, paper, box}; S i (AB)=(A∪B)-(A∩B)=0, the quantified difference is 0, that is, the name similarity is equal to 100%.

[0057] When the combination of name, specification and measurement unit of the drug information to be identified is not completely consistent with the standard drug information, the number and content of the elements in the word segmentation result are exactly the same, but the arrangement order of the elements is different, and the difference degree = 0.

[0058] Original drug information form:

[0059]

[0060] Drug information word list:

[0061]

[0062] Among them, the standard drug information does not mark the attribute of "manufacturer", so when using the parameter, the "manufacturer" is not involved in the recognition rule. The word segmentation set is A = {75, ethanol, disinfectant, 500ml, bottle}, B = {ethanol, disinfectant, 75, 500ml, bottle}; Si (AB) = (A ∪ B) - (A ∩ B) = 0, the quantified difference = 0, that is, the name similarity is equal to 100%.

[0063] When the combination of the name, specification, and unit of measurement of the drug information to be identified is not completely consistent with the standard drug information, the number and content of the elements in the word segmentation results are also not completely consistent, and the difference is >0.

[0064]

[0065] Among them, standard drug information does not consider the attribute of "manufacturer", and adopts the rule that "manufacturer" does not participate in recognition. The word segmentation set is A = {American ginseng, branch, 250g, bag, 3g, round, grain, g}, a total of 8 elements; B1 = {American ginseng, branch, 250g, bag, 5g, round, grain}, a total of 7 elements; B2 = {American ginseng, branch, 250g, bag, 3g, round, grain}, a total of 7 elements.

[0066] Comparison between drugs A and B1 to be identified:

[0067] SI(AB1)=ABS(A-B1)+ABS(B1-A)=ABS(2)+ABS(-1)=3, and there are 3 difference amounts.

[0068] Comparison between drugs A and B2 to be identified:

[0069] SI(AB2)=ABS(A-B2)+ABS(B2-A)=ABS(1)+ABS(0)=1, and there is one difference.

[0070] The ABS(A-B1) function is used to obtain the absolute value of the difference in the number of elements in set A and set B1, and the ABS(A-B2) function is used to obtain the absolute value of the difference in the number of elements in set A and set B2. SI(AB1) is the sum of the number of elements in the two drug segmentation sets of set A and set B1 that are not in each other's segmentation element set, that is, the difference between set A and set B1. SI(AB2) is the sum of the number of elements in the two drug segmentation sets of set A and set B2 that are not in each other's segmentation element set, that is, the difference between set A and set B2. The difference range is an integer between 0 and n. When n>5, it is basically unreliable.

[0071] According to the difference quantification rule "the closer the value is to 0, the higher the similarity is, and the larger the value is, the lower the similarity is", then there are recognition results:

[0072]

[0073] Among them, when there is no identification result with a quantitative difference of 0, the record of standard drug B2 (code 100017) is the optimal recommendation result of this identification.

[0074] Furthermore, the drug segmentation set also includes the following processing steps: in the drug segmentation set, the name segmentation is arranged in the front position, and the other segmentations are arranged in the back position, so as to obtain the drug segmentation set elements arranged in an orderly manner; if the difference is caused by the drug segmentation set element Key, the difference is corrected and adjusted; the formula for correcting the difference is:

[0075] S diff =[ABS(AB)]+1*M+[ABS(BA)]+1*N

[0076] In the formula, S diff is the degree of difference, A is the set of drug segmentations that preliminarily identify drug names and specifications, B is the set of drug segmentations that are similar to drug names and specifications, the ABS function is used to obtain the absolute value of the difference in the number of elements in set A and set B, M is the correction factor A, and N is the correction factor B.

[0077] When ABS(AB)>0, then M=1, otherwise M=0, ABS(AB) represents the absolute difference between set A and set B. When ABS(BA)>0, then N=1, otherwise N=0, ABS(BA) represents the absolute difference between set B and set A.

[0078] In this embodiment, (AB) represents a set of elements that belong to A but not to B, and (BA) represents a set of elements that belong to B but not to A.

[0079] The difference degree is corrected and adjusted, that is, the difference quantization algorithm is corrected. By correcting the difference quantization algorithm, the final stable intelligent identifier recommendation result is stored in a two-dimensional table in the memory (disk).

[0080] Furthermore, the formula for converting set difference into name similarity is:

[0081]

[0082] In the formula, S i(AB) is the name similarity between the preliminarily identified drug name plus specification and similar drug name plus specification, A is the drug word set corresponding to the preliminarily identified drug name plus specification, B is the drug word set corresponding to the similar drug name plus specification, and the count(.) function is used to count the number of elements in the set.

[0083] Furthermore, the drug information in the local drug catalog database will be marked after being segmented, and the segmentation results corresponding to the drug information will be stored, and the marked drug information will be automatically split again during the next polling.

[0084] In this embodiment, the word library and the standard drug catalog word segmentation identification linkage automatic maintenance mechanism in the local drug catalog database: in the word segmentation dictionary conversion table, drug name word segmentation word library, and drug manufacturer word library, the change of each word library (deletion / addition of word library entries) will affect the subsequent word segmentation results of the drugs to be identified. Therefore, it is required that when each word library is changed, the [word segmentation] mark = N involving these changed entries in the standard drug list should be automatically changed to facilitate the automatic re-splitting of these entries during the next polling. The text in the joint name of drug name + specification + measurement unit can be divided into meaningful vocabulary units by a word segmenter. These vocabulary units can be words, phrases or other meaningful text fragments (even single characters, such as g in drug information expression). The performance of the word segmenter is very important for many NLP computing tasks, such as part-of-speech tagging, information extraction, and text classification. Therefore, the accuracy and efficiency of the word segmenter directly affect the processing capacity and computing performance of big data. By making full use of Oracle's memory computing technology, computer IO is reduced as much as possible to improve computing efficiency.

[0085] Furthermore, after obtaining the name similarity between the preliminary identified drug name plus specification and the similar drug name plus specification, the following steps are also included: extracting drug ingredient information based on the image information of the drug to be identified, and normalizing the ingredient information for word segmentation to obtain a set of drug ingredient segmentations to be identified; constructing a first knowledge graph based on the set of drug ingredient segmentations to be identified; extracting the drug ingredient information corresponding to the similar drug names plus specifications that have been preliminarily retrieved and extracted through a natural language algorithm, and normalizing the extracted drug ingredient information for word segmentation to obtain a set of similar drug ingredient segmentations; constructing a second knowledge graph based on the set of similar drug ingredient segmentations; comparing the sub-graphs of the first knowledge graph and the second knowledge graph through a sub-graph matching algorithm to obtain the similarity of the drug ingredient knowledge graph; when there are multiple similar drugs with the same name similarity, comparing the similarity of the drug ingredient knowledge graphs, and selecting the similar drug name corresponding to the highest value of the drug ingredient knowledge graph similarity as the recommended drug name output.

[0086] In this embodiment, the similarity of the drug ingredient knowledge graph can be obtained by the following formula:

[0087]

[0088] Where Ss is the similarity of the drug ingredient knowledge graph, the sim(.) function is used to calculate the similarity between two objects, the diff(.) function is used to calculate the difference between two objects, r is the number of the common subgraph of the first knowledge graph and the second knowledge graph, t is the common number of subgraphs of the first knowledge graph and the second knowledge graph, μ r is the weight of the rth sub-graph in the knowledge graph, GA is the vectorization result of the first knowledge graph, and GB is the vectorization result of the second knowledge graph.

[0089] Furthermore, the drug ingredient knowledge graph similarity can also be used to derive other drug names based on the ingredient information and specification information when there are multiple drug names for the same drug. The specific method is: obtain drug specification information 1 of the drug to be identified, and simultaneously obtain drug specification information 2 corresponding to the similar drug name plus specification, and perform word segmentation processing on drug specification information 1 and drug specification information 2 to obtain specification word segmentation result 1 and specification word segmentation result 2; sort the drug ingredient knowledge graph similarities between the drug to be identified and similar drugs from high to low; select the first K similar drugs corresponding to the drug ingredient knowledge graph similarities, K is the ranking of the drug ingredient knowledge graph similarity, K is a natural number, and compare the specification word segmentation result 1 of the drug to be identified and the specification word segmentation result 2 of the similar drug; obtain the similar drug with the highest similarity between specification word segmentation result 1 and specification word segmentation result 2, and record it as the drug alias recommendation output.

[0090] In this embodiment, K is the ranking of the drug ingredient knowledge graph similarity, which is manually set and selected. Some drugs may have different names, but their ingredients and specifications may be the same. By using the drug ingredient knowledge graph similarity, drugs with similar ingredients can be identified and their other names can be inferred, thereby helping users to identify these drugs more easily.

[0091] like Figure 2As shown, it is a structural diagram of the device for identifying the name of a drug to be identified provided in an embodiment of the present application. The device for identifying the name of a drug to be identified provided in an embodiment of the present application includes: a basic information acquisition module, a similar name retrieval module, a word segmentation processing module, a name similarity analysis module and a recommendation output module: Basic information acquisition module: used to obtain image information of the drug to be identified, extract text from the image information, and obtain a preliminary identified drug name plus specifications; Similar name retrieval module: used to connect to the local drug catalog database, preliminarily retrieve and extract drug names plus specifications that are similar to the preliminary identified drug name plus specifications; Word segmentation processing module: used to perform normalized word segmentation processing on the preliminary identified drug name plus specifications and similar drug names plus specifications; Name similarity analysis module: used to process the normalized word segmentation results through a set difference quantization algorithm to obtain the name similarity of the preliminary identified drug name plus specifications and similar drug names plus specifications; Recommendation output module: used to select a similar drug name with the highest name similarity as the drug with the same name as the drug to be identified, and output it as a recommended drug name.

[0092] Among them, an embodiment of the present application also provides a computer-readable storage medium for storing a program, which, when executed by a processor, implements a method for identifying the name of a drug to be identified.

[0093] The technical scheme in the above-mentioned embodiment of the present application has at least the following technical effects or advantages: relative to a specific entity name recognition optimization method disclosed in the patent application with publication number: CN115481623A, the embodiment of the present application utilizes word segmentation technology and combines the quantification method of the symmetric difference of two sets to obtain a preliminary identification of the similarity between the drug name plus specification and the standard drug name plus specification name in the local drug catalog database, thereby quickly identifying the drug name and recommending the same drug; relative to the name recognition and association recommendation method, device, computer equipment and storage medium announced in the invention patent with announcement number: CN112966504B, the embodiment of the present application analyzes the knowledge graph based on the ingredient information of the drug, finds similar drugs based on the knowledge graph, and then compares the similarity between each drug and the drug to be identified, thereby achieving secondary verification to identify whether the drugs belong to the same drug, thereby increasing the accuracy of drug recommendations.

[0094] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0096] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0097] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0098] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0099] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for identifying the name of a drug to be identified, characterized in that: The following steps are involved: By obtaining the image information of the drug to be identified, extracting the text from the image information, and obtaining the preliminary identification of the drug name and specification; Connect to the local drug catalog database to preliminarily search and extract drug names and specifications that are similar to the preliminarily identified drug names and specifications; The initially identified drug names plus specifications and similar drug names plus specifications are normalized and segmented; The normalized word segmentation results were processed by the set difference quantification algorithm to obtain the name similarity of the drug name plus specification and similar drug names plus specifications; Select the similar drug name with the highest name similarity as the drug with the same name as the drug to be identified, and output it as the recommended drug name; The method for obtaining the name similarity is: Extract the normalized segmentation results corresponding to the preliminary identified drug name plus specification and similar drug names plus specifications; Compare the normalized segmentation results of the two groups, find the same segmentations and remove them, and obtain their respective drug segmentation sets; The number of different segmented words in the two groups of drug segmented words is obtained by counting, which is recorded as the set difference degree; Convert set differences into name similarities; The drug segmentation set also includes the following processing steps: In the drug segmentation set, the name segmentation is arranged in the front position, and the other segmentations are arranged in the back position, so as to obtain the drug segmentation set elements arranged in an orderly manner; If the difference is caused by the drug segmentation set element Key, the difference is corrected and adjusted; The formula for correcting the difference is: S diff =[ABS(A-B)]+1*M+[ABS(B-A)]+1*N In the formula, S diff is the difference degree, A is the drug segmentation set of the preliminary identified drug name plus specification, B is the drug segmentation set of similar drug names plus specifications, the ABS function is used to obtain the absolute value of the difference in the number of elements in set A and set B, M is the correction factor A, and N is the correction factor B; When ABS(AB)>0, then M=1, otherwise M=0, ABS(AB) represents the absolute difference between set A and set B. When ABS(BA)>0, then N=1, otherwise N=0, ABS(BA) represents the absolute difference between set B and set A. The formula for converting set difference into name similarity is: In the formula, S i (AB) is the name similarity between the preliminarily identified drug name plus specification and similar drug name plus specification, A is the drug segmentation set corresponding to the preliminarily identified drug name plus specification, B is the drug segmentation set corresponding to the similar drug name plus specification, count ( . ) The function is used to count the number of elements in a collection; After obtaining the name similarity of the preliminary identification drug name plus specification and similar drug name plus specification, the following steps are also included: Extracting drug ingredient information based on the image information of the drug to be identified, and performing normalized segmentation processing on the ingredient information to obtain a segmentation set of the drug ingredients to be identified; Constructing a first knowledge graph based on the word segmentation set of the drug components to be identified; The drug ingredient information corresponding to the similar drug names and specifications that were initially retrieved and extracted is extracted through a natural language algorithm, and the extracted drug ingredient information is normalized and segmented to obtain a similar drug ingredient segmentation set; Construct a second knowledge graph based on a set of similar drug ingredient segmentations; Compare the subgraphs of the first knowledge graph and the second knowledge graph through a subgraph matching algorithm to obtain the similarity of the drug ingredient knowledge graphs; When there are multiple similar drugs with the same name similarity, the drug ingredient knowledge graph similarity is compared, and the similar drug name corresponding to the drug ingredient knowledge graph with the highest similarity value is selected as the recommended drug name output; The similarity of the drug ingredient knowledge graph can be obtained by the following formula: In the formula, S s is the similarity of drug ingredient knowledge graph, sim ( . ) The function is used to calculate the similarity between two objects, diff ( . ) The function is used to calculate the difference between two objects, r is the number of the common subgraphs of the first knowledge graph and the second knowledge graph, t is the common number of subgraphs of the first knowledge graph and the second knowledge graph, μ r is the weight of the rth sub-graph in the knowledge graph, GA is the vectorization result of the first knowledge graph, and GB is the vectorization result of the second knowledge graph.

2. The method for identifying the name of a drug to be identified as claimed in claim 1, characterized in that: The word segmentation process comprises the following steps: The drug name plus specification is segmented by drug name, drug specification information, and drug measurement unit; The segmentation results of drug name, drug specification information and drug measurement unit are respectively subjected to secondary segmentation processing; The segmentation results obtained from the two segmentation processes are both normalized and normalized; Get the normalized word segmentation results.

3. The method for identifying the name of a drug to be identified as claimed in claim 1, characterized in that: The drug information in the local drug catalog database will be marked after being segmented, and the segmentation results corresponding to the drug information will be stored, and the marked drug information will be automatically split again during the next polling.

4. The method for identifying the name of a drug to be identified as claimed in claim 1, characterized in that: The drug ingredient knowledge graph similarity can also be used to obtain other drug names based on ingredient information and specification information when there are multiple drug names for the same drug. The specific method is as follows: Obtain drug specification information 1 of the drug to be identified, and simultaneously obtain drug specification information 2 corresponding to a similar drug name plus specification, and perform word segmentation processing on drug specification information 1 and drug specification information 2 to obtain specification word segmentation result 1 and specification word segmentation result 2; Sort the drug ingredient knowledge graph similarities between the drug to be identified and similar drugs from high to low; Select similar drugs corresponding to the first K drug ingredient knowledge graph similarities, where K is the ranking of the drug ingredient knowledge graph similarity and K is a natural number, and compare the specification word segmentation result 1 of the drug to be identified with the specification word segmentation result 2 of the similar drug; The similar drugs with the highest similarity between specification segmentation result 1 and specification segmentation result 2 are obtained and recorded as the drug alias recommendation output.

5. A device for identifying the name of a drug to be identified, characterized in that: It includes basic information acquisition module, similar name retrieval module, word segmentation processing module, name similarity analysis module and recommendation output module: The basic information acquisition module is used to obtain the image information of the drug to be identified, extract the text from the image information, and obtain the name and specification of the drug to be identified preliminarily; The similar name search module is used to connect to the local drug catalog database, preliminarily search and extract drug names and specifications that are similar to the preliminarily identified drug names and specifications; The word segmentation processing module is used to perform normalized word segmentation processing on the initially identified drug names plus specifications and similar drug names plus specifications; The name similarity analysis module is used to process the normalized word segmentation results through a set difference quantization algorithm to obtain the name similarity of the drug name plus specification and similar drug names plus specifications for preliminary identification; The recommendation output module is used to select the similar drug name with the highest name similarity as the drug with the same name as the drug to be identified, and output it as the recommended drug name; The method for obtaining the name similarity is: Extract the normalized segmentation results corresponding to the preliminary identified drug name plus specification and similar drug names plus specifications; Compare the normalized segmentation results of the two groups, find the same segmentations and remove them, and obtain their respective drug segmentation sets; The number of different segmented words in the two groups of drug segmented words is obtained by counting, which is recorded as the set difference degree; Convert set differences into name similarities; The drug segmentation set also includes the following processing steps: In the drug segmentation set, the name segmentation is arranged in the front position, and the other segmentations are arranged in the back position, so as to obtain the drug segmentation set elements arranged in an orderly manner; If the difference is caused by the drug segmentation set element Key, the difference is corrected and adjusted; The formula for correcting the difference is: S diff =[ABS(A-B)]+1*M+[ABS(B-A)]+1*N In the formula, S diff is the difference degree, A is the drug segmentation set of the preliminary identified drug name plus specification, B is the drug segmentation set of similar drug names plus specifications, the ABS function is used to obtain the absolute value of the difference in the number of elements in set A and set B, M is the correction factor A, and N is the correction factor B; When ABS(AB)>0, then M=1, otherwise M=0, ABS(AB) represents the absolute difference between set A and set B. When ABS(BA)>0, then N=1, otherwise N=0, ABS(BA) represents the absolute difference between set B and set A. The formula for converting set difference into name similarity is: In the formula, S i (AB) is the name similarity between the preliminarily identified drug name plus specification and similar drug name plus specification, A is the drug segmentation set corresponding to the preliminarily identified drug name plus specification, B is the drug segmentation set corresponding to the similar drug name plus specification, count ( . ) The function is used to count the number of elements in a collection; After obtaining the preliminary identification of the similarity between the drug name plus specification and the similar drug name plus specification, the following steps are also included: Extracting drug ingredient information based on the image information of the drug to be identified, and performing normalized segmentation processing on the ingredient information to obtain a segmentation set of the drug ingredients to be identified; Constructing a first knowledge graph based on the word segmentation set of the drug components to be identified; The drug ingredient information corresponding to the similar drug names and specifications that were initially retrieved and extracted is extracted through a natural language algorithm, and the extracted drug ingredient information is normalized and segmented to obtain a similar drug ingredient segmentation set; Construct a second knowledge graph based on a set of similar drug ingredient segmentations; Compare the subgraphs of the first knowledge graph and the second knowledge graph through a subgraph matching algorithm to obtain the similarity of the drug ingredient knowledge graphs; When there are multiple similar drugs with the same name similarity, the drug ingredient knowledge graph similarity is compared, and the similar drug name corresponding to the drug ingredient knowledge graph with the highest similarity value is selected as the recommended drug name output; The similarity of the drug ingredient knowledge graph can be obtained by the following formula: In the formula, S s is the similarity of drug ingredient knowledge graph, sim ) . ) The function is used to calculate the similarity between two objects, diff ( . ) The function is used to calculate the difference between two objects, r is the number of the common subgraphs of the first knowledge graph and the second knowledge graph, t is the number of common subgraphs of the first knowledge graph and the second knowledge graph, μ r is the weight of the rth sub-graph in the knowledge graph, GA is the vectorization result of the first knowledge graph, and GB is the vectorization result of the second knowledge graph.

6. A computer-readable storage medium for storing a program, characterized in that: When the program is executed by a processor, the method for identifying the name of a drug to be identified as described in any one of claims 1 to 4 is implemented.

Images