Medical service system based on knowledge graph and logistic regression

Abstract

The invention provides a medical service system based on a knowledge graph and logistic regression. The medical service system comprises a data acquisition and preprocessing module; a data storage module; a data processing and analyzing module; a data visualization module; a knowledge graph module; and a disease prediction module. The system focuses on teenagers, middle-aged people and old people, and four system modules of disease analysis, disease science popularization, disease prevention and health maintenance are set according to different susceptibility characteristics of different groups, so that people can have clearer understanding, prevention and coping means for diseases and health in daily life.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of medical service systems, and specifically to a medical service system based on knowledge graph and logistic regression. Background Art

[0002] When the existing technology provides corresponding smart medical services through a medical platform, it first obtains a series of medical data through the platform, then processes the obtained data, and finally returns the corresponding data processing results.

[0003] The current medical service system makes it difficult to conduct targeted analysis and treatment based on different populations, making it difficult to focus on the three groups of teenagers, middle-aged people, and the elderly, and to set up disease analysis, disease science popularization, disease prevention, and health care based on the different susceptibility characteristics of different groups. Summary of the invention

[0004] The present invention mainly provides a medical service system based on knowledge graph and logistic regression to solve the technical problems raised in the above background technology.

[0005] The technical solution adopted by the present invention to solve the above technical problems is:

[0006] A medical service system based on knowledge graph and logistic regression, including:

[0007] Data collection and preprocessing module, used for data information deduplication and denoising, building word segmentation index files and data feature extraction;

[0008] The data storage module imports the processed data set into HDFS and uses the Hive database based on the distributed file system to store the data. Compared with other relational databases, Hive can store larger amounts of data. Hive is based on HDFS and can store data in the entire cluster.

[0009] Data processing and analysis module, using big data framework for data analysis and E charts for data visualization;

[0010] Data visualization module, designed according to three groups and summarized through a main page;

[0011] The knowledge graph module processes the acquired unstructured data, extracts entities, attributes, and relationships from the data source and forms triples, forms ontological knowledge expressions on this basis, and stores them in the form of graph databases. The knowledge is expressed as a network graph structure through triples to facilitate knowledge expression, extraction, organization, retrieval, and reasoning.

[0012] The disease prediction module uses Python to analyze the data that has been preprocessed and build a logistic regression model. The prediction information is mainly modeled based on the selected data with greater correlation.

[0013] Furthermore, the data acquisition and preprocessing module fills in missing values, smoothes or deletes outliers, and corrects data inconsistencies to obtain standard continuous data for the collected data, and then merges the data involving multiple data sources to generate a new data set, thereby providing a unified data view for subsequent query and analysis processing. In the present invention, the module fills in missing values, smoothes or deletes outliers, and corrects data inconsistencies to obtain standard continuous data for the collected data, and then merges the data involving multiple data sources to generate a new data set, thereby providing a unified data view for subsequent query and analysis processing.

[0014] Furthermore, the data storage module imports the processed data set into HDFS and uses the Hive database based on the distributed file system to store the data. In the present invention, Hive can store more massive data than other relational databases. Hive is based on HDFS and can store data in the entire cluster. The data received by the server is stored in Hive, and the program will read the data through the Hive tool class. Hive itself does not store and calculate data, and Hive can be understood as a client tool, so the data is shared from Hive to MySQL through Sqoop.

[0015] Furthermore, the data processing and analysis module uploads the cleaned data to HDFS, uses Spark RDD to partition, convert, calculate and other operations on the data, uploads the processed data to the Hive table, and uses Hive SQL statements to process and analyze the table data. In the present invention, the visualization of large-scale data in the system is mainly based on the technology of parallel algorithm design, which rationally utilizes limited computing resources and efficiently processes and analyzes the characteristics of specific data sets. The background of this project mainly uses the big data framework for data analysis, and E charts completes data visualization.

[0016] Furthermore, after the data visualization module obtains data from MySQL, the data is displayed in the form of a table by constructing a visualization module using HTML+CSS+JS language.

[0017] Furthermore, the disease prediction module divides the data into 75% training set and 25% test set, the training set is used to train the model, and the test set is used to test the accuracy of the model. The selection of the training set and the test set is random, and then the data is standardized to ensure that the variance of the feature data of each dimension is 1 and the mean is 0. In the present invention, the prediction results will not be dominated by the excessive feature values ​​of some dimensions. Since the only two results that need to be predicted are whether or not the disease is likely to occur, after testing the model through various algorithm training, it is found that the accuracy of the logistic regression algorithm is the highest, and logistic regression is the most traditional algorithm with higher accuracy in disease prediction. The PMML model is then used to export the trained model, thereby achieving cross-system use of the model. The project uses the model established by the logistic regression algorithm based on the data entered by the user for prediction, and finally displays the prediction results on the page.

[0018] Furthermore, it also includes a medical service device, which includes a base, the upper surface of the base is connected to a chassis, the interior of the chassis is connected to an electric push rod, the top end of the electric push rod extends to the outside and is connected to a spherical universal joint, and the top end of the spherical universal joint is connected to a data acquisition instrument. In the present invention, the chassis is supported by the base, the electric push rod is protected by the chassis, the spherical universal joint and the acquisition instrument connected to the spherical universal joint are driven by the electric push rod, so as to adjust the height of the acquisition instrument. In this way, it is convenient to adapt to the height of the user. By setting the spherical universal joint, the angle of the acquisition instrument can be adjusted, which is convenient for the use of the acquisition instrument.

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] This invention is mainly based on the background of the flood of informatization. By focusing on three groups of teenagers, middle-aged people and the elderly, according to the different susceptibility characteristics of different groups, four system modules of disease analysis, disease popularization, disease prevention and health preservation are established. By using intelligent knowledge graphs, data analysis and logistic regression algorithm models, the disease status, susceptible symptoms, treatment methods, knowledge popularization, daily health preservation and other issues of common susceptible diseases are explained from different angles, so that we can have a clearer understanding of diseases and health in daily life, and prevention and coping methods.

[0021] The present invention will be explained in detail below with reference to the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a diagram of the system framework of the present invention;

[0023] Figure 2 This is a flow chart of data collection and preprocessing of the present invention;

[0024] Figure 3It is a data storage flow chart of the present invention;

[0025] Figure 4 The database is successfully queried for part of the data migration of the present invention. Figure 1 ;

[0026] Figure 5 The database is successfully queried for part of the data migration of the present invention. Figure 2 ;

[0027] Figure 6 It is a triple structure diagram of the present invention;

[0028] Figure 7 Constructing a flow chart for the knowledge graph of the present invention;

[0029] Figure 8 This is a flow chart of the prediction module of the present invention;

[0030] Fig. 9 The model diagram after saving the PMML function of the present invention;

[0031] Fig.10 It is a schematic diagram of the structure of the medical service device of the present invention. DETAILED DESCRIPTION

[0033] To facilitate understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings, but the present invention can be implemented in different forms and is not limited to the embodiments described in the text. On the contrary, these embodiments are provided to make the content disclosed in the present invention more thorough and comprehensive.

[0034] It should be noted that when an element is referred to as being "fixed on" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used in this article are for illustrative purposes only.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which the present invention belongs. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0036] Embodiment 1, a medical service system based on knowledge graph and logistic regression mainly uses Python language to clean data, then uploads the data to HDFS, and then uses sqoop to upload the data on HDFS to the Hive table, uses Hive SQL statements to analyze table data or uses Spark RDD to analyze table data, stores the analyzed data in MySQL, uses Java SSM framework to obtain data in the database, uses E charts to visualize the data, and uses Python to build relevant regression, classification, recommendation and other models, and displays the relevant prediction data to the front end through the JavaSSM framework. The corresponding relevant data generated by the front end and the function overview are shown in the following table.

[0037]

[0038]

[0039] Embodiment 2: In the life cycle of system development, data collection is in the first link. For different data sets, there may be different structures and modes, such as files, XML trees, relational tables, etc., which are manifested as data heterogeneity. For multiple heterogeneous data sets, further integrated processing or integrated preprocessing is required. The data set of this system comes from the UCI data set, the Tianchi data set and the Kaggle website download. The system construction mainly includes three parts: deduplication and denoising of data information, construction of word segmentation index files and data feature extraction. The collected data is filled with missing values, smoothed or deleted outliers, and corrected for data inconsistencies to obtain standard continuous data. Then the data involving multiple data sources are merged to generate a new data set, which provides a unified data view for subsequent query and analysis.

[0040] In the third embodiment, the processed data set is imported into HDFS. Since the amount of data to be stored is huge, the ordinary relational database is not used directly. Instead, the Hive database based on the distributed file system is used to store the data. Compared with other relational databases, Hive can store larger amounts of data. Hive is based on HDFS and can store data in the entire cluster. The data received by the server is stored in Hive, and the program reads the data through the Hive tool class. Hive itself does not store or calculate data. Hive can be understood as a client tool, so the data is shared from Hive to MySQL through Sqoop.

[0041] Example 4: The visualization of large-scale data in the system is mainly based on the technology of parallel algorithm design, which makes rational use of limited computing resources and efficiently processes and analyzes the characteristics of specific data sets. The background of this project mainly uses the big data framework for data analysis, and E charts completes data visualization. The main processing and analysis process is to upload the cleaned data to HDFS, use Spark RDD to partition, convert, calculate and other operations on the data, upload the processed data to the Hive table, and use Hive SQL statements to process and analyze the table data.

[0042] Example 5: The visualization module is a medium for systematically displaying the data analyzed and processed above. Our visualization module is mainly designed based on three groups and summarized through a main page. The specific technical process can be summarized as obtaining data from MySQL and then displaying the data in a table by constructing a visualization module through HTML+CSS+JS language.

[0043] Embodiment 6: Knowledge graph is one of the semantic web technologies, which describes a graph-based data structure. The construction of knowledge graph is based on a large amount of data, which requires large-scale data collection. The data we collected comes from the Open Kg public knowledge graph data set and related public website data. We roughly process the unstructured data obtained, extract entities, attributes, and relationships from the data source and form triples, and on this basis, form an ontological knowledge expression, and store it in the form of a graph database. The knowledge is expressed as a network graph structure through triples to facilitate the expression, extraction, organization, retrieval and reasoning of knowledge;

[0044] The logical structure of the knowledge graph is divided into two levels: the data layer and the schema layer. In the data layer of the knowledge graph, knowledge is stored in the graph database in units of facts. If the [entity-relationship-entity] or [entity-attribute-value] triples are used as the basic expression of facts, all the data stored in the graph database will constitute a huge entity relationship network and form a knowledge graph. The schema layer is above the data layer and is the core of the knowledge graph. The schema layer stores refined knowledge. An ontology library is usually used to manage the chess-like layer of the knowledge graph. The ontology library's support for axioms, rules, and constraints is used to standardize the connections between entities, relationships, and entity types and attributes. The status of the ontology library in the knowledge graph is equivalent to the chess pieces of the knowledge base. A knowledge base with an ontology library has less redundant knowledge. According to the requirements of our system, we created knowledge graphs for four diseases using the neo4j graph database;

[0045] Knowledge graph technology refers to the technology of establishing and applying knowledge graphs. It is an interdisciplinary research that integrates cognitive computing, knowledge representation and reasoning, information retrieval and extraction, natural language processing and semantic Web, data mining and machine learning. Knowledge graph was proposed by Google in 2012 and successfully applied to search engines. Knowledge graph belongs to the research scope of knowledge engineering, an important research field of artificial intelligence. It is a killer application of using knowledge engineering to establish large-scale knowledge resources. In the era of big data, knowledge engineering is to automatically or semi-automatically obtain knowledge from big data and establish a knowledge-based system to provide Internet intelligent knowledge services. The demand for intelligent services from big data has changed from simply collecting and obtaining information to automated knowledge services. We need to use knowledge engineering to add semantics / knowledge to big data, make data intelligent (smart data), and complete the transformation process from data to information to knowledge, and finally to intelligent applications, so as to achieve insights into big data, provide answers to questions that users care about, provide support for decision-making, and improve user experience. The knowledge graph has shown an increasingly important application value in the following applications. The knowledge graph is a way of reintegrating into human logical thinking, which is convenient for human observation and understanding. We link the front end with the neo4j graph database and display the corresponding relationships of different diseases in the navigation bar. Users can open the navigation bar and click on the corresponding relationships of different diseases to query the information they want to know and the corresponding knowledge graph will be displayed for better understanding.

[0046] Embodiment seven, use Python to analyze the data that has been preprocessed, build a logistic regression model, and predict information is mainly based on the selected data with greater correlation for modeling. The data is divided into 75% training set and 25% test set, the training set is used to train the model, and the test set is used to test the accuracy of the model. The selection of training set and test set is random, and then the data is standardized to ensure that the variance of the feature data of each dimension is 1 and the mean is 0, so that the prediction results will not be dominated by the feature values ​​that are too large in some dimensions. Since the results that need to be predicted are only two types of whether or not the disease is likely to occur, after testing the model through various algorithm training models, it is found that the correct rate of the logistic regression algorithm is the highest, and logistic regression is the most traditional algorithm with higher accuracy in disease prediction, and then the PMML model is used to export the trained model, so as to achieve the cross-system use of the model. The project uses the model established by the logistic regression algorithm according to the data entered by the user for prediction, and finally displays the prediction results on the page.

[0047] The above is an exemplary description of the present invention in combination with the accompanying drawings. It is obvious that the specific implementation of the present invention is not limited to the above-mentioned method. As long as such non-substantial improvements are made by adopting the method concept and technical solution of the present invention, or the concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. A medical service system based on knowledge graph and logistic regression, characterized in that: include: Data collection and preprocessing module, used for data information deduplication and denoising, building word segmentation index files and data feature extraction; The data storage module imports the processed data set into HDFS and uses the Hive database based on the distributed file system to store the data. Compared with other relational databases, Hive can store larger amounts of data. Hive is based on HDFS and can store data in the entire cluster. Data processing and analysis module, using big data framework for data analysis and E charts for data visualization; Data visualization module, designed according to three groups and summarized through a main page; The knowledge graph module processes the acquired unstructured data, extracts entities, attributes, and relationships from the data source and forms triples, forms ontological knowledge expressions on this basis, and stores them in the form of graph databases. The knowledge is expressed as a network graph structure through triples to facilitate knowledge expression, extraction, organization, retrieval, and reasoning. The disease prediction module uses Python to analyze the data that has been preprocessed and build a logistic regression model. The prediction information is mainly modeled based on the selected data with greater correlation.

2. According to claim 1, a medical service system based on knowledge graph and logistic regression is characterized in that: The data collection and preprocessing module obtains standard continuous data by filling missing values, smoothing or deleting outliers, and correcting data inconsistencies, and then merges data involving multiple data sources to generate a new data set, providing a unified data view for subsequent query and analysis processing.

3. According to claim 1, a medical service system based on knowledge graph and logistic regression is characterized in that: The data storage module imports the processed data set into HDFS and uses the Hive database based on the distributed file system to store the data.

4. According to claim 1, a medical service system based on knowledge graph and logistic regression is characterized in that: The data processing and analysis module uploads the cleaned data to HDFS, uses Spark RDD to partition, convert, calculate and other operations on the data, uploads the processed data to the Hive table, and uses Hive SQL statements to process and analyze the table data.

5. According to claim 1, a medical service system based on knowledge graph and logistic regression is characterized in that: After the data visualization module obtains data from MySQL, it displays the data in the form of a table by constructing a visualization module using HTML+CSS+JS language.

6. According to claim 1, a medical service system based on knowledge graph and logistic regression is characterized in that: The disease prediction module divides the data into 75% training set and 25% test set. The training set is used to train the model, and the test set is used to test the accuracy of the model. The selection of the training set and the test set is random, and then the data is standardized to ensure that the variance of the feature data in each dimension is 1 and the mean is 0.

7. According to claim 1, a medical service system based on knowledge graph and logistic regression is characterized in that: The medical service device also comprises a base (10), the upper surface of the base (10) is connected to a chassis (20), the interior of the chassis (20) is connected to an electric push rod (30), the top end of the electric push rod (30) extends to the outside and is connected to a spherical universal joint (40), and the top end of the spherical universal joint (40) is connected to a data acquisition instrument (50).

Images