Electronic medicine specification processing method based on knowledge graph and management system thereof

Abstract

The invention provides an electronic medicine instruction processing method based on a knowledge graph and a management system thereof, which solve the problem that the existing electronic medicine instruction cannot be subjected to semantic analysis and the like, and comprises the following steps: S1, data extraction and data cleaning: a user submits medicine instruction data, and the system extracts the text content of the medicine instruction; preprocessing and cleaning the data; s2, carrying out standardization and structural processing on the data, carrying out intelligent association on the data, carrying out semantic understanding and keyword extraction by adopting a natural language processing NLP, and constructing a knowledge graph; and S3, storing the data in a database, and establishing a back-end service. The method has the advantages of stable operation, good semantic understanding effect and the like.

Description

Technical Field

[0001] The present invention belongs to the field of information technology technology, and specifically relates to a method for processing electronic drug instructions based on a knowledge graph and a management system thereof. Background Art

[0002] With the rapid development of information technology, electronic drug instructions have gradually replaced traditional paper instructions and become an important part of the pharmaceutical industry. Electronic drug instructions can not only improve the efficiency of information dissemination, but also realize personalized recommendations and automatic analysis through intelligent processing. In recent years, the progress of natural language processing (NLP), knowledge graphs and artificial intelligence technologies has made intelligent processing of drug instructions possible. Through data extraction, structured processing, semantic understanding and knowledge association, electronic drug instructions can provide more accurate information services for the medical industry and improve the safety and convenience of patients' medication. At present, the existing electronic drug instructions on the market mainly rely on OCR technology for text recognition, combined with database storage and retrieval technology to realize information query and display. Most of these systems have basic data extraction and display functions, and some high-end systems have introduced natural language processing technology to improve the accuracy of user queries. However, the existing technologies mainly focus on traditional text storage and retrieval, and cannot effectively perform semantic analysis, knowledge association and intelligent content generation.

[0003] In order to solve the shortcomings of the existing technology, people have conducted long-term exploration and proposed various solutions. For example, a Chinese patent document discloses a method for constructing a knowledge graph of rational drug use based on drug instructions [201910593831.7], which includes the following steps: S10, extracting drug instructions, summarizing the entities and relationships therein through expert annotation method, and forming an entity and relationship indexing rule base; S20, using a semi-supervised learning method, training a machine learning model based on expert-annotated data and machine learning rules; S30, using the trained machine learning model to predict and annotate unannotated drug instructions to form a knowledge graph of drug relationships.

[0004] The above solution has solved the problem of electronic drug instructions to a certain extent, but the solution still has many shortcomings, such as the inability to perform semantic analysis. Summary of the invention

[0005] The purpose of the present invention is to provide an electronic drug instructions processing method based on a knowledge graph that has a reasonable design and realizes semantic parsing of drug instructions in response to the above problems.

[0006] Another object of the present invention is to provide an electronic drug instructions management system based on a knowledge graph to realize knowledge association in response to the above-mentioned problems.

[0007] To achieve the above object, the present invention adopts the following technical solution: a method for processing electronic drug instructions based on knowledge graph, comprising the following steps: S1: Data extraction and data cleaning: users submit drug insert data, the system extracts the text content of the drug insert, and preprocesses and cleans the data; S2: Standardize and structure the data, intelligently associate the data, use natural language processing (NLP) for semantic understanding and keyword extraction, and build a knowledge graph; S3: Store data in a database and establish backend services.

[0008] In the above-mentioned method for processing electronic drug instructions based on knowledge graph, step S1 includes the following steps: S11: The user submits the drug instructions for format conversion; S12: perform data cleaning and denoising; S13: Separation processing according to the part of speech of the text; S14: Filter meaningless text; S15: Perform feature engineering and keyword extraction; S16: Set text retrieval strategy.

[0009] In the above-mentioned method for processing electronic drug instructions based on knowledge graph, step S2 includes the following steps: S21: Perform data integration, clean unstructured data, perform SQL schema mapping and parse JSON / XML semi-structured data; S22: Standardize the data, unify the naming and establish a synonym library, discretize the data and use the k-means algorithm for cluster analysis to obtain the classified data set; S23: Set association rules and confidence thresholds, and use the Apriori algorithm to perform data association; S24: Perform NLP deep semantic processing, use the RNN model to understand the semantics of drug instructions and capture core semantic units; S25: Build a knowledge graph and perform visualization.

[0010] In the above-mentioned method for processing electronic drug instructions based on knowledge graph, step S25 includes the following steps: S251: Perform document triage, imitate the dynamic routing mechanism of the MoE architecture, build an expert module selector, synthesize drug instructions based on templates, and use GAN to generate complex cases; S252: Perform knowledge extraction, capture structured information, use lightweight BiLSTM-CRF to identify complex entities, manually review the low-confidence results of dynamic routing, and extract relationship paths based on dynamic routing; S253: Align multi-source data and make automatic decisions based on authority weighting; S254: Imitate the on-demand activation strategy of the MoE architecture to perform user portrait analysis and assemble knowledge units.

[0011] In the above-mentioned method for processing electronic drug instructions based on knowledge graph, step S252 includes the following steps: S2521: Input the text into the BiLSTM model to obtain the vector of the fused contextual semantic information corresponding to each unit word , the vector Divided into word vectors as shown below , bit vector and segment vector : ; S2521: Vector Extract information features and input the BiLSTM model results into the CRF model for annotation, as shown below: ; ; in, is the normalization factor, Refers to the drug instructions text sequence variable, Annotate variables for entities, is given Output sequence under the condition The conditional probability distribution of is a state feature function and the entity category is associated with the drug instruction text, is a local feature function and the entity category is associated with the external feature.

[0012] In the above-mentioned method for processing electronic drug instructions based on knowledge graph, the multi-source data in step S253 includes drug administration data, pharmaceutical company data and literature data, and the regulatory hierarchy priority is that drug administration data is greater than pharmaceutical company data, and pharmaceutical company data is greater than literature data. The authority weighted automatic decision used is as follows: ; in, is the number of data sources, for data sources, For data source The discrete value of For data source The authority value weight and 0≤ ≤1, is the magnification factor.

[0013] In the above-mentioned method for processing electronic drug instructions based on knowledge graph, the steps S251 and S254 use credibility factors to process the uncertainty of entity relationships. The credibility factors are generated as follows: ; in, is the superordinate entity of the entity relationship, is the subordinate entity of the entity relationship, is the credibility factor of the entity relationship, and the entity relationship credibility is calculated as follows: ; in, For uncertain entity relationships, is the credibility factor of the entity relationship.

[0014] In the above-mentioned method for processing electronic drug instructions based on knowledge graph, after the knowledge graph is constructed in step S2, redundant nodes are merged and invalid information is deleted.

[0015] In the above-mentioned method for processing electronic drug instructions based on knowledge graph, step S3 includes the following steps: S31: Convert the contents of the drug instructions into speech and perform segment optimization; S32: Call the WeChat mini program port and generate the WeChat mini program code.

[0016] An electronic drug instructions management system based on knowledge graph adopts the above-mentioned electronic drug instructions processing method based on knowledge graph.

[0017] Compared with the existing technology, the advantages of the present invention are: improving the accuracy and consistency of drug instruction data through multi-level data cleaning and standardization; introducing semantic analysis and knowledge graph technology to realize intelligent association and recommendation of drug information and improve user query efficiency; supporting QR code scanning access, voice broadcast and voice interaction to make information acquisition more convenient; improving the reusability of drug instruction data through structured storage and intelligent retrieval, and providing more efficient information services for the medical industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a drug instructions processing flow chart of the present invention; Figure 2 is a data processing flow chart of the present invention; Figure 3 It is a flow chart of knowledge graph construction of the present invention; Figure 4 It is a schematic diagram of the management system of the present invention. DETAILED DESCRIPTION

[0019] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1-4 As shown, a method for processing electronic drug instructions based on a knowledge graph includes the following steps: S1: Data extraction and data cleaning. Users submit drug instructions data, and the system extracts the text content of the drug instructions, preprocesses and cleans the data. The drug instructions are the core data source. After obtaining the drug instructions text by crawling the medical website, entity recognition and relationship extraction are required. For example, entities include drug names, ingredients, indications, contraindications, etc., and relationships cover drug interactions, dosage restrictions, etc. The construction process usually combines expert annotation methods with semi-supervised learning to form a rule base and train machine learning models, which are finally stored in a structured manner in the database. S2: Standardize and structure data, intelligently associate data, use natural language processing (NLP) for semantic understanding and keyword extraction, and build a knowledge graph; use deep learning models to extract entities and relationships from unstructured text to improve data annotation accuracy; S3: Store the data in the database and establish backend services. Users can enable the backend services and obtain electronic drug instructions by scanning the code.

[0021] Specifically, step S1 includes the following steps: S11: The user scans the existing document, submits the drug instructions, determines the document type and performs format conversion; S12: Perform data cleaning and denoising, with a focus on cleaning data on traditional Chinese medicine; S13: Separate the text according to its part of speech and use NLTK / Spacy for part of speech tagging; S14: Filter meaningless text and remove stop words and low-frequency words; S15: Use TF-IDF algorithm for feature engineering and keyword extraction; S16: Set text retrieval strategy.

[0022] In depth, step S2 includes the following steps: S21: Perform data integration, clean unstructured data, perform SQL schema mapping and parse JSON / XML semi-structured data; S22: Standardize the data, unify the naming and establish a synonym library, for example, map paracetamol to acetaminophen. Use semi-supervised annotation to effectively reduce the amount of manual annotation. Discretize the data and use the k-means algorithm for cluster analysis to obtain the classified data set. S23: Set association rules and confidence thresholds, and use the Apriori algorithm to perform data association; S24: Perform NLP deep semantic processing, use the RNN model to understand the drug instructions and capture the core semantic units, input the drug instructions text and output the core semantic units; S25: Build a knowledge graph and perform visualization.

[0023] Further, step S25 includes the following steps: S251: Perform document triage, imitate the dynamic routing mechanism of the MoE architecture, build an expert module selector, synthesize drug instructions based on templates, use GAN to generate complex cases, simulate rare side effect descriptions, and expand training data; S252: Perform knowledge extraction, capture structured information, use lightweight BiLSTM-CRF to identify complex entities, dynamically route low-confidence results for manual review, and extract relationship paths based on dynamic routing; the extracted entities include drug names, dosages, etc.; S253: Align multi-source data and make automatic decisions based on authority weighting; S254: Imitate the on-demand activation strategy of the MoE architecture to perform user portrait analysis, assemble knowledge units, and use deep semantic understanding to adapt to different scenario requirements.

[0024] Furthermore, step S252 captures structured information such as measurement and frequency by the rule layer, and the model layer uses lightweight BiLSTM-CRF to identify complex entities such as contraindications and interactions. Finally, the verification layer submits the low-confidence results to manual review, which specifically includes the following steps: S2521: Input the text into the BiLSTM model to obtain the vector of the fused contextual semantic information corresponding to each unit word , the vector Divided into word vectors as shown below , bit vector and segment vector : ; S2521: Vector Extract information features and input the BiLSTM model results into the CRF model for annotation, as shown below: ; ; in, is the normalization factor, Refers to the drug instructions text sequence variable, Annotate variables for entities, is given Output sequence under the condition The conditional probability distribution of is a state feature function and the entity category is associated with the drug instruction text, is a local feature function and the entity category is associated with the external feature.

[0025] In addition, the multi-source data in step S253 includes drug administration data, pharmaceutical company data and literature data, and the regulatory hierarchy priority is that drug administration data is greater than pharmaceutical company data, and pharmaceutical company data is greater than literature data. The authority weighted automatic decision used is as follows: ; in, is the number of data sources, for data sources, For data source The discrete value of For data source The authority value weight and 0≤ ≤1, By introducing multiple data sources and taking into account their authority, we can practice the mapping of Chinese and Western medicine knowledge, which is suitable for the medical field that needs to take into account compliance requirements and knowledge diversity. In actual applications, it needs to be adjusted according to the characteristics of the data source. K The value and weight allocation strategy is used, and the decision quality is optimized through supplementary rules (such as timeliness and randomization).

[0026] At the same time, step S251 and step S254 use a credibility factor to process the uncertainty of the entity relationship. The credibility factor is generated as follows: ; in, is the superordinate entity of the entity relationship, is the subordinate entity of the entity relationship, is the credibility factor of the entity relationship, and the entity relationship credibility is calculated as follows: ; in, For uncertain entity relationships, is the credibility factor of the entity relationship.

[0027] Step S251 and step S254 imitate the MoE architecture to build an expert module to analyze the drug ingredients and identify indications, contraindications, etc., and hand it over to the routing network to process the entity relationship, and finally integrate the output results of each expert module to give medication recommendations.

[0028] Step S254 assembles knowledge units according to different user portraits such as doctors, patients or pharmacists, identifies user devices and adjusts the output format, which is suitable for multiple service ports such as mobile terminals and PC terminals.

[0029] It can be seen that after the knowledge graph is constructed in step S2, redundant nodes are merged and invalid information is deleted. At the same time, it is also necessary to monitor the announcements of the Food and Drug Administration in real time and update the knowledge graph through difference comparison.

[0030] Obviously, step S3 includes the following steps: S31: Convert the contents of the drug instructions into speech and perform segment optimization; S32: Call the WeChat mini program port and generate the WeChat mini program code.

[0031] An electronic drug instructions management system based on knowledge graph adopts the above-mentioned electronic drug instructions processing method based on knowledge graph.

[0032] To sum up, the principle of this embodiment is that the user submits the drug instructions data, the system automatically extracts the text content of the drug instructions, uses algorithms to pre-process the extracted data, including format conversion, removal of redundant information, error correction, etc., and performs secondary and tertiary cleaning to improve data quality, perform data standardization and structured processing, improve data consistency, use natural language processing (NLP) for semantic understanding and keyword extraction, build a knowledge graph, establish the relationship between drugs, support intelligent query and recommendation, store the cleaned data in the database, and provide efficient retrieval function.

[0033] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

[0034] Although this article uses more terms such as knowledge graph, it does not exclude the possibility of using other terms. These terms are used only to more conveniently describe and explain the essence of the present invention; interpreting them as any additional restrictions is contrary to the spirit of the present invention.

Claims

1. A method for processing electronic drug instructions based on knowledge graph, characterized in that: The steps include: S1: Data extraction and data cleaning: users submit drug insert data, the system extracts the text content of the drug insert, and preprocesses and cleans the data; S2: Standardize and structure the data, intelligently associate the data, use natural language processing (NLP) for semantic understanding and keyword extraction, and build a knowledge graph; S3: Store data in a database and establish backend services.

2. The method for processing electronic drug instructions based on knowledge graph according to claim 1, characterized in that: The step S1 comprises the following steps: S11: The user submits the drug instructions for format conversion; S12: perform data cleaning and denoising; S13: Separation processing according to the part of speech of the text; S14: Filter meaningless text; S15: Perform feature engineering and keyword extraction; S16: Set text retrieval strategy.

3. The method for processing electronic drug instructions based on knowledge graph according to claim 1, characterized in that: The step S2 comprises the following steps: S21: Perform data integration, clean unstructured data, perform SQL schema mapping and parse JSON / XML semi-structured data; S22: Standardize the data, unify the naming and establish a synonym library, discretize the data and use the k-means algorithm for cluster analysis to obtain the classified data set; S23: Set association rules and confidence thresholds, and use the Apriori algorithm to perform data association; S24: Perform NLP deep semantic processing, use the RNN model to understand the semantics of drug instructions and capture core semantic units; S25: Build a knowledge graph and perform visualization.

4. The method for processing electronic drug instructions based on knowledge graph according to claim 3 is characterized in that: The step S25 comprises the following steps: S251: Perform document triage, imitate the dynamic routing mechanism of the MoE architecture, build an expert module selector, synthesize drug instructions based on templates, and use GAN to generate complex cases; S252: Perform knowledge extraction, capture structured information, use lightweight BiLSTM-CRF to identify complex entities, manually review the low-confidence results of dynamic routing, and extract relationship paths based on dynamic routing; S253: Align multi-source data and make automatic decisions based on authority weighting; S254: Imitate the on-demand activation strategy of the MoE architecture to perform user portrait analysis and assemble knowledge units.

5. The method for processing electronic drug instructions based on knowledge graph according to claim 4 is characterized in that: The step S252 includes the following steps: S2521: Input the text into the BiLSTM model to obtain the vector of the fused contextual semantic information corresponding to each unit word , the vector Divided into word vectors as shown below , bit vector and segment vector : ； S2521: Vector Extract information features and input the BiLSTM model results into the CRF model for annotation, as shown below: ； ； in, is the normalization factor, Refers to the drug instructions text sequence variable, Annotate variables for entities, is given Output sequence under the condition The conditional probability distribution of is a state feature function and the entity category is associated with the drug instruction text, is a local feature function and the entity category is associated with the external feature.

6. The method for processing electronic drug instructions based on knowledge graph according to claim 5 is characterized in that: The multi-source data in step S253 includes drug administration data, pharmaceutical company data and literature data, and the regulatory hierarchy priority is that drug administration data is greater than pharmaceutical company data, and pharmaceutical company data is greater than literature data. The authority weighted automatic decision used is as follows: ； in, is the number of data sources, for data sources, For data source The discrete value of For data source The authority value weight and 0≤ ≤1, is the magnification factor.

7. The method for processing electronic drug instructions based on knowledge graph according to claim 3 is characterized in that: The steps S251 and S254 use a credibility factor to process the uncertainty of the entity relationship. The credibility factor is generated as follows: ； in, is the superordinate entity of the entity relationship, is the subordinate entity of the entity relationship, is the credibility factor of the entity relationship, and the entity relationship credibility is calculated as follows: ； in, For uncertain entity relationships, is the credibility factor of the entity relationship.

8. The method for processing electronic drug instructions based on knowledge graph according to claim 3 is characterized in that: After the knowledge graph is constructed in step S2, redundant nodes are merged and invalid information is deleted.

9. The method for processing electronic drug instructions based on knowledge graph according to claim 1, characterized in that: The step S3 comprises the following steps: S31: Convert the contents of the drug instructions into speech and perform segment optimization; S32: Call the WeChat mini program port and generate the WeChat mini program code.

10. An electronic drug instructions management system based on knowledge graph, characterized in that: The electronic drug instructions processing method based on knowledge graph described in any one of claims 1 to 9 is adopted.

Images