A medical record data quality monitoring method and system based on a knowledge base

Abstract

The present invention relates to a medical record data quality monitoring method and system based on a knowledge base. The method includes: dividing medical record data into N medical record components, performing a graph representation on the medical record components to obtain a medical record graph; assigning values to nodes in the medical record graph based on the knowledge base, which are called node vectors; and obtaining a medical record quality type corresponding to the medical record based on the node vectors and the medical record graph. The present invention performs a quantitative calculation of the medical record quality type based on a graph with the support of the knowledge base, greatly improving the efficiency of medical record quality monitoring.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent medical treatment, and particularly relates to a method and system for monitoring the quality of medical record data based on a knowledge base. Background Art

[0002] With the development of information technology, the medical and health industry has gradually achieved informatization. As the core of the hospital information system (HIS), the electronic medical record organizes data "centered around the patient", provides electronic storage, query, statistics, data exchange, and can timely integrate the scattered patient information from all aspects into electronic data and assist doctors in making diagnoses. The electronic medical record is the recording carrier of medical treatment behaviors. Therefore, by monitoring and controlling the quality of electronic medical records, the implementation and execution of the core systems of medical quality and safety management can be grasped. For real-time monitoring of electronic medical records, it is first necessary to dock with the medical record system. Based on the big data of medical records, the quality control rules library of the core systems of medical quality and safety is used for analysis and quality control. Hidden dangers and defects are discovered in a timely manner to prevent the occurrence of medical quality and safety incidents. The construction scope of intelligent hospitals in China mainly includes three major fields: "intelligent medical treatment", "intelligent service" and "intelligent management". The intelligentization of medical quality management is an important feature of realizing intelligent hospitals. However, there are currently common problems in hospitals such as low standardization of medical data, mostly unstructured expressions of medical record data, and scattered medical quality control points, resulting in the vast amounts of data accumulated in existing information systems such as HIS, electronic medical records, anesthesia records, and inspection and examination records being unable to play their due roles and values in medical quality management. According to the existing diagnosis and treatment norms and requirements for medical record quality control, it is impossible to achieve refined management requirements for the control of the connotation quality of medical records only by manual means. Since there are many control nodes for medical record quality control, it is difficult for existing technologies to control each department, each doctor, and each medical record, and it is even more difficult to achieve full traceability. Therefore, the level of information management of medical records, the efficiency and quality of medical record supervision in existing technologies are relatively low. Especially for traditional Chinese medicine medical records, the means of standardizing medical data are limited, and the means of monitoring the quality of medical record data using a knowledge base are still limited to semantic analysis, and the availability of the monitoring results provided is not strong. Due to the large workload of current medical record quality control, insufficient quality control levels of quality control personnel, and the scarcity of highly educated full-time quality control personnel with a clinical background, the medical record quality control is not ideal. Moreover, traditional quality control is limited to end-of-term quality control and cannot monitor and remind the clinical diagnosis and treatment process in real time, resulting in the medical record quality control being limited to form and lagging behind the medical process. Only by establishing and improving the medical record quality control system from the source can the fundamental problems of medical record quality be systematically solved.

[0003] Apply the knowledge base and artificial intelligence technology to the monitoring of medical record quality data. Analyze complex medical data through algorithms and software to achieve the purpose of approximating human cognition. Through automatic scanning technology, quality problems in clinical trial data can be discovered for risk and quality control. It can also be used to build an intelligent screening system to improve the efficiency of patient recruitment in clinical trials and data quality control. Build a warning and prediction model through machine learning technology. Predict the trend of analysis indicators in the future for a period of time through the model, and be able to give early warnings for analysis indicators. In addition, it can also be used to analyze the reasons for medical quality problems, realize all-round tracking of the improvement process and multi-dimensional analysis of improvement effects. Utilize the large amount of medical data stored in the knowledge base, which can be used for multi-dimensional analysis and query of medical records, providing the basis and samples for monitoring and analyzing the reasons for abnormal data. The knowledge graph technology can use the knowledge points formed by the structuring of medical records to build a knowledge graph, complete the in-depth semantic understanding of electronic medical record data, and help improve the accuracy and efficiency of medical record quality data monitoring. However, how to specifically apply the comprehensive use of these technologies to make the monitoring of medical record quality data more efficient and accurate, and contribute to improving the quality of medical services and patient safety, is a technical problem to be solved. To solve the problems in the existing technology, the present invention performs quantitative calculation of the medical record quality type based on a graph under the support of a knowledge base, greatly improving the efficiency of medical record quality monitoring. Summary of the Invention

[0004] To solve the above problems in the existing technology, the present invention proposes a method and system for monitoring the quality of medical record data based on a knowledge base. The method includes:

[0005] Step S1: Divide the medical record data into N medical record components , ; construct a medical record graph with the medical record components as nodes ; for any two different components , if the components are independent of each other, then set ; if the components are related to each other, then set ;

[0006] Step S2: Assign values to the nodes in the medical record graph based on the knowledge base, called node vectors; specifically including the following steps:

[0007] Step S21: Set the node vector as an attribute representation for each node; obtain an unprocessed node in the medical record graph and initialize the node vector of this node , where: is the kth element in the node vector; k = 1~K; k is the medical record quality type number, and K is the total number of medical record quality types;

[0008] Step S22: Take the component corresponding to the unprocessed node as the current component, and compare it with the medical record samples in the knowledge base. If there is a medical record in the knowledge base that has another component of the same type as the current component, calculate the similarity between the current component and the other component; if the similarity between the two is greater than the similarity threshold, obtain the medical record quality type k of the said medical record, and set the value of element k in the node vector ; otherwise, continue to process the next medical record; repeat this step until all medical records have been processed;

[0009] Step S23: If there are still unprocessed nodes, return to Step S21; otherwise, proceed to the next step;

[0010] Step S24: Perform normalization processing on the node vector; set the node vector ;

[0011] Step S3: Obtain the medical record quality type corresponding to the medical record based on the node vector and the medical record graph; specifically: propagate through the node vector in the medical record graph to obtain the medical record quality type corresponding to the medical record; specifically, it includes the following steps:

[0012] Step S3A1: Set the current medical record quality type to be processed as k, referred to as the current type k;

[0013] Step S3A2: Arrange all the nodes in the medical record graph in ascending order of the values of the node vectors corresponding to the current type k to obtain the propagation queue; to obtain the propagation queue;

[0014] Step S3A3: Obtain an unprocessed node from the propagation queue, referred to as the first node ; perform propagation of the node vector centered on the first node;

[0015] Step S3A4: If there is another node in the medical record graph, referred to as the second node ; the length of the shortest path between the second node and the first node is ; then update the value in the node vector of the second node based on the following formula (1); where: is the propagation attenuation coefficient; is the propagation attenuation coefficient;

[0016] (1);

[0017] Step S3A5: If there are still unprocessed nodes in the propagation queue, return to Step S3A3; otherwise, proceed to the next step;

[0018] Step S3A6: If there is an unprocessed medical record quality type, set , and return to Step S3A1;

[0019] Step S3A7: Renormalize the node vector of each node and set the node vector ; Calculate the graph vector ; Normalize the graph vector and set the graph vector ; Take The medical record quality type corresponding to the type k value as the determined medical record quality type.

[0020] Furthermore, each medical record component relates to different parts of the medical record.

[0021] Furthermore, the initial values of the node vectors before assignment are all 0.

[0022] Furthermore, the similarity is semantic similarity.

[0023] Furthermore, the similarity threshold is a preset value.

[0024] Furthermore, update the medical record quality type of the medical record samples in the knowledge base according to the medical record writing standardization, the accuracy of diagnosis and treatment, the integrity of medical record content, medical record timeliness, legal compliance, data quality, medical record filing quality, and / or traditional Chinese medicine characteristics evaluation.

[0025] Furthermore, .

[0026] Furthermore, the medical record is a traditional Chinese medicine or Western medicine medical record.

[0027] A medical record data quality monitoring system based on a knowledge base, the medical record data quality monitoring system based on a knowledge base is used to implement the above-mentioned medical record data quality monitoring method based on a knowledge base.

[0028] A medical record data quality monitoring platform based on a knowledge base, the medical record data quality monitoring platform based on a knowledge base is used to implement the above-mentioned medical record data quality monitoring method based on a knowledge base.

[0029] A medical record data quality monitoring system based on a knowledge base, the medical record data quality monitoring system based on a knowledge base is used to implement the above-mentioned medical record data quality monitoring method based on a knowledge base.

[0030] A medical record data quality monitoring device based on a knowledge base, the medical record data quality monitoring device based on a knowledge base is used to implement the above-mentioned medical record data quality monitoring method based on a knowledge base.

[0031] A medical record data quality monitoring module based on a knowledge base, the medical record data quality monitoring module based on a knowledge base is used to implement the above-mentioned medical record data quality monitoring method based on a knowledge base.

[0032] A medical record data quality monitoring server based on a knowledge base, which is used to implement the above-mentioned medical record data quality monitoring method based on the knowledge base.

[0033] The beneficial effects of the present invention include:

[0034] (1) By graphically representing the medical record data in a componentized manner, mapping the content relevance of the medical record data to the relevance of the nodes in the graph, expanding based on the dimension of the medical record quality type when assigning values to the nodes, and expressing the flow relationship of the medical record quality data between different components through the propagation of the node vectors in the medical record graph, so that the quantitative calculation of the medical record quality type based on the graph can be carried out with the support of the knowledge base, greatly improving the efficiency of medical record quality monitoring;

[0035] (2) Based on the independent cascade model for updating the node vectors, the basic assumption is that each node makes an independent decision when deciding whether to spread information, making the model easy to implement and operate, and being able to effectively solve the reasoning and analysis under the complex data relationships of the medical record data, providing a quantitative method for understanding and predicting the effect of the spread of the medical record quality type, and more scientifically using the big data information in the knowledge base to guide the progress of medical record quality monitoring;

[0036] (3) Through the expression of the medical record graph and the node vectors, docking with the neural network model, and performing the cross and mutation of the node vectors through the relationships of the nodes in the graph, the descent speed of parameter optimization can be greatly improved, and the stability and robustness of the model can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The drawings described herein are used to provide a further understanding of the present invention, form a part of this application, but do not constitute an improper limitation to the present invention. In the drawings:

[0038] Figure 1 is a schematic diagram of the medical record data quality monitoring method based on the knowledge base provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] The present invention will be described in detail below in conjunction with the drawings and specific embodiments, in which the illustrative embodiments and descriptions are only used to explain the present invention, but do not constitute a limitation to the present invention.

[0040] The present invention proposes a medical record data quality monitoring method and system based on a knowledge base. As shown in the appendix Figure 1 The method includes the following steps:

[0041] Step S1: Divide the medical record data into N medical record components , ; construct a medical record graph with the medical record components as nodes ; for any two different components , if the components are independent of each other, then set ; if the components are related to each other, then set ; it can be known that is the edge between node and node ;

[0042] Preferably: Each medical record component relates to different parts of the medical record; the data content in the medical record component is relevant or irrelevant; then when there is an association between two medical record components, an edge is set between them; of course, it can also be used in a way of the strength of the association to reduce the number of edges in the medical record graph to reduce the computational complexity; when the association between the two is weak, the corresponding edge is not set;

[0043] Preferably: Divide the medical record components according to the organizational structure of the medical record data, the medical record file structure, and the relevance of the medical record data area; after the division, set the type for each component according to the division method;

[0044] Step S2: Assign values to the nodes in the medical record graph based on the knowledge base, which is called the node vector; specifically, it includes the following steps:

[0045] Step S21: Set the node vector as the attribute representation for each node; obtain an unprocessed node in the medical record graph and initialize the node vector of this node , where: is the k-th element in the node vector; k = 1~K; k is the medical record quality type number, and K is the total number of medical record quality types. That is to say, the node vector is corresponding to the medical record quality type; that is to say, when assigning values to the nodes, it is carried out based on the dimension of the medical record quality type;

[0046] Step S22: Take the component corresponding to this node as the current component and compare it with the medical record samples in the knowledge base. If there is a medical record in the knowledge base that has another component of the same type as the current component, then calculate the similarity between the current component and the other component; if the similarity between the two is greater than the similarity threshold, then obtain the medical record quality type k of the said medical record and set the value of the element k in the node vector ; otherwise, continue to process the next medical record; repeat this step until all medical records are processed; where: is the k-th element in the node vector, representing the attribute value of the node in the medical record quality type k;

[0047] Preferably: The initial values of the node vectors before assignment are all 0;

[0048] Preferably: comparing it with the medical record samples in the knowledge base specifically means: comparing it with the same type of medical record samples summarized in the knowledge base; for example, for the same disease, belonging to the same department, using the same template, etc.

[0049] Preferably: the similarity is semantic similarity; the similarity threshold is a preset value.

[0050] Step S23: If there are still unprocessed nodes, return to step S21; otherwise, proceed to the next step.

[0051] Step S24: Perform normalization processing on the node vector; set the node vector ;

[0052] Step S3: Obtain the medical record quality type corresponding to the medical record based on the node vector and the medical record graph; specifically: obtaining the medical record quality type corresponding to the medical record by propagating the node vector in the medical record graph; or obtaining the medical record quality type through a neural network model.

[0053] The process of determining the medical record quality type corresponding to the medical record by propagating the node vector in the medical record graph specifically includes the following steps:

[0054] Step S3A1: Set the current medical record quality type to be processed as k, called the current type k.

[0055] Preferably: Set the initial value of k to 1.

[0056] Step S3A2: Arrange all the nodes in the medical record graph in ascending order of the values of the node vectors corresponding to the current type k to obtain a propagation queue. to obtain a propagation queue.

[0057] Step S3A3: Obtain an unprocessed node from the propagation queue, called the first node ; perform propagation of the node vector with the first node as the center.

[0058] Step S3A4: If there is another node in the medical record graph, called the second node ; the length of the shortest path between the second node and the first node is ; update the value in the node vector of the second node based on the following formula (1); where: is the propagation attenuation coefficient; is the propagation attenuation coefficient;

[0059] (1);

[0060] Preferably: ;

[0061] Alternatively: If the length is Then update the value in the node vector of the second node based on Equation (1), otherwise, do not update; where is the longest propagation path, which is a preset value; is the longest propagation path, which is a preset value;

[0062] Preferably: set ; it can be set according to the computational complexity and the size of the medical record graph;

[0063] Alternatively: starting from the second node corresponding to the shortest path with the largest length to the second node corresponding to the shortest path with the smallest length, then update the value in the node vector of the second node based on Equation (1) in sequence; the update mode adopted here is an iterative update mode; obviously, there are no two identical nodes on the above shortest path; and each node is only updated once; Alternatively: starting from the second node corresponding to the shortest path with the largest length to the second node corresponding to the shortest path with the smallest length, then update the value in the node vector of the second node based on Equation (1) in sequence; the update mode adopted here is an iterative update mode; obviously, there are no two identical nodes on the above shortest path; and each node is only updated once; value; the update mode adopted here is an iterative update mode; obviously, there are no two identical nodes on the above shortest path; and each node is only updated once;

[0064] Step S3A5: If there are still unprocessed nodes in the propagation queue, return to Step S3A3; otherwise, proceed to the next step;

[0065] Alternatively: The above Steps S3A2~S3A5 are specifically to update the node vector based on the independent cascade model; specifically including the following steps:

[0066] Step S3A41: Determine the propagation probability between every two nodes ; specifically: Compare the first node (and its corresponding component) and the second node (and its corresponding component) with the medical record samples in the knowledge base respectively. If there is a medical record in the knowledge base that has another node (and its corresponding component) of the same type as the first node (and its corresponding component) or the second node (and its corresponding component), then calculate the similarity between the first node (and its corresponding component) or the second node (and its corresponding component) and this other node (and its corresponding component) respectively; if the similarities of the two are equal (or close), then set ; otherwise, continue to process the next medical record sample; repeat this step until all medical records are processed; set ; where: is the count value; is the number of medical record samples;

[0067] Preferably: set the initial value of the count value to 0;

[0068] Step S3A42: Obtain a group of two unprocessed nodes with a connection relationship from the medical record graph, which are respectively called the first node and the second node ; perform two-way propagation on the first node and the second node; repeat this step until all two nodes with a connection relationship are processed;

[0069] The two-way propagation between the first node and the second node is specifically as follows: If it is greater than or equal to the preset propagation probability, then set , set ;

[0070] Alternatively, the two-way propagation between the first node and the second node is specifically as follows: perform normalization on the propagation probability, obtain the normalized value of the propagation probability based on the following formula (2); update the node vectors of the first node and the second node based on the following formulas (3) and (4).

[0071] (2);

[0072] (3);

[0073] (4);

[0074] Step S3A6: If there is an unprocessed medical record quality type, then set , and return to Step S3A1;

[0075] Step S3A7: Use the medical record quality type corresponding to the type k value of as the determined medical record quality type; specifically: re-normalize the node vector of each node, and set the node vector ; calculate the graph vector ; normalize the graph vector, and set the graph vector ; use the medical record quality type corresponding to the type k value of as the determined medical record quality type;

[0076] Alternatively, Step S3A7 specifically includes the following steps:

[0077] Step S3A71: Calculate the information entropy of each medical record quality type ; specifically, calculate the information entropy using the following formulas (5) to (7) ; where: is to perform operation under the condition of ; is an intermediate variable;

[0078] (5);

[0079] (6);

[0080] (7);

[0081] Step S3A72: Renormalize the node vectors of each node and set the node vectors ; Calculate the graph vector ; Normalize the graph vector and set the graph vector ; Take the medical record quality type corresponding to the k value of as the determined medical record quality type;

[0082] Preferably: The medical record quality type includes K types, where: The larger the K value, the better the medical record quality; The medical record quality type is obtained from perspectives such as the writing standardization of medical records, the accuracy of diagnosis and treatment, the integrity of medical record content, the timeliness of medical records, legal compliance, data quality, medical record filing quality, and traditional Chinese medicine feature evaluation. Each medical record quality type reflects the comprehensive evaluation results of these perspectives;

[0083] Alternatively: The medical record quality type includes K types, where: The smaller the K value, the better the medical record quality;

[0084] Alternatively: The K types included in the medical record quality type respectively correspond to perspectives such as the writing standardization of medical records, the accuracy of diagnosis and treatment, the integrity of medical record content, the timeliness of medical records, legal compliance, data quality, medical record filing quality, and traditional Chinese medicine feature evaluation to obtain the medical record quality type. Each medical record quality type reflects the independent evaluation results of these perspectives. Each element in the graph vector respectively indicates the independent evaluation result of an independent perspective;

[0085] Preferably: The data quality includes whether the data in the medical record is accurate and error-free, including diagnosis codes, surgical codes, drug usage records, etc.;

[0086] The method for obtaining the medical record quality type through the neural network model is specifically as follows: Construct a neural network model as a medical record data quality prediction model, enrich the knowledge base based on the genetic algorithm, and input the node input vector corresponding to the medical record into the prediction model to obtain the medical record quality type of this medical record;

[0087] Specifically, it includes the following steps:

[0088] Step S3B1: Construct a prediction model based on the BP neural network; Construct samples based on the medical records in the knowledge base; The input of the prediction model is the input vector formed by concatenating the node vectors of all nodes; The output part of the prediction model is the medical record quality type; The number of nodes in the input layer and the hidden layer of the prediction model is both ; The number of output nodes is K; That is, the ( node corresponds to the kth element of the pathology graph node ;

[0089] Preferably, the splicing method of the node vectors is to connect the head and tail in the order of the node numbers to form the input vector of the element;

[0090] Step S3B2: Determine the mutation probability, and set the mutation probability of each node to ;

[0091] Preferably; set =-0.01~0.1;

[0092] Alternatively: Set the mutation probability ; where: is the base mutation probability, is the node degree of; set the crossover probability to a preset value, for example: 0.5~0.9; initialize the weights and biases;

[0093] Preferably: Randomly initialize the weights (W) and biases (b) of all connections in the network; for each sample, calculate the output value of each layer, and for each node in the hidden layer and the output layer, use the activation function sigmoid to calculate the output y=f(Wx + b); where: y is the activation function; x is the input value; since the number of nodes in the input layer is equal to the number of nodes in the hidden layer (convolutional layer), the corresponding activation function is set to ;

[0094] Preferably: The connection layer is set as a fully connected layer;

[0095] Step S3B3: Set the constraint condition as: when there is a connection edge between two nodes in the medical record graph, then they can cross-mutate with each other, otherwise, they cannot cross-mutate;

[0096] Step S3B4: For the node , its node vector is , perform a mutation operation on its k-th element to obtain the mutated node , whose node vector is ; specifically: Use the following formula (8) to calculate the element value of the node vector of the mutated node; where: The node vector of is , is the propagation coefficient, is a random number between; for the nodes and , whose node vectors are respectively , perform a crossover operation on their k-th elements to obtain the crossover node , whose node vector is ; specifically: Use the following formula (9) to calculate the element value of the node vector of the mutated node; where: the node The node vector of , is a random number between;

[0097] (8);

[0098] (9);

[0099] Step S3B5: Set the fitness function as the mean square error function; Repeat the above steps S3B1 to S3B4 until the fitness function is satisfied; At this time, the optimization of the weights and bias parameters of the BP neural network by the genetic algorithm is completed; Based on the connection relationship between nodes, introducing propagation into the node crossover and mutation process can greatly improve the descent speed of parameter optimization and quickly improve the stability and robustness of the model;

[0100] Step S3B6: Preprocess the node vector of the current medical record, set ; After splicing its node vectors, input them into the prediction model to obtain the medical record quality type corresponding to the current medical record;

[0101] Furthermore: Randomly select medical records as the current medical record and obtain the corresponding medical record quality type to achieve the monitoring of the quality of medical record data;

[0102] Based on the same inventive concept, the present invention also provides a medical record data quality monitoring system based on a knowledge base, and the system is used to implement the above-mentioned medical record data quality monitoring method based on a knowledge base;

[0103] Based on the same inventive concept, the present invention also provides a medical record data quality monitoring server based on a knowledge base, and the system is used to implement the above-mentioned medical record data quality monitoring method based on a knowledge base;

[0104] Based on the same inventive concept, the present invention also provides a medical record data quality monitoring client based on a knowledge base, and the system is used to implement the above-mentioned medical record data quality monitoring method based on a knowledge base;

[0105] A computer program (also known as a program, software, software application, script, or code) can be written in any form of programming language, including assembly or interpreted languages, declarative or procedural languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. A computer program may or may not correspond to a file in a file system. The program can be stored as part of a file that holds other programs or data (such as one or more scripts stored in a markup language document), in a single file dedicated to the program, or in multiple cooperating files (such as files that store one or more modules, subroutines, or portions of code). A computer program can be deployed to execute on one computer or on multiple computers located at one site or distributed across multiple sites and interconnected by a communication network.

[0106] Those skilled in the art will appreciate that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

[0107] The present invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart illustrations and / or block diagrams, and combinations of flows and / or blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing device create means for implementing the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or combinations of blocks.

[0108] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or combinations of blocks.

[0109] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are executed on the computer or other programmable apparatus to produce a computer-implemented process, thereby the instructions executed on the computer or other programmable apparatus provide steps for realizing the functions specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 steps for realizing the functions specified in one block or a plurality of blocks.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the specific implementation manners of the present invention can still be modified or equivalently replaced, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A method for monitoring the quality of medical record data based on a knowledge base, characterized in that: The method comprises: Step S1: Divide the medical record data into N medical record components , ; Use medical record components as nodes to build a medical record graph ; For any two different components , if the component are independent of each other, then set ; If the component If they are related to each other, set ; Each medical record component relates to a different part of the medical record; the data contents in the medical record components are related or unrelated; Step S2: Assign values ​​to nodes in the medical history graph based on the knowledge base to represent attributes, which is called node vector ; It is the kth element in the node vector, indicating the attribute value of the node in the medical record quality type k; k=1~K; k is the medical record quality type number, K is the total number of medical record quality types; specifically: Step S21: Set a node vector for each node as an attribute representation; obtain an unprocessed node in the medical history graph and initialize the node vector of the node ; Step S22: Take the component corresponding to the node as the current component, and compare it with the medical record sample in the knowledge base. If there is a medical record in the knowledge base that has another component of the same type as the current component, calculate the similarity between the current component and the other component; if the similarity between the two is greater than the similarity threshold, obtain the medical record quality type k of the medical record, and set the value of element k in the node vector ; Otherwise, continue processing the next medical record; Repeat this step until all medical records have been processed; Step S3: Propagate the node vector in the medical record graph to obtain the medical record quality type corresponding to the medical record; for any first node , if there is another node in the medical history graph, it is called the second node ; The length of the shortest path between the two is ; then set ;in: is the propagation attenuation coefficient; after the propagation is completed, The medical record quality type corresponding to the type k value is taken as the determined medical record quality type.

2. The method for monitoring medical record data quality based on a knowledge base according to claim 1, characterized in that: Each medical record component refers to a different part of the medical record.

3. The method for monitoring medical record data quality based on a knowledge base according to claim 2, characterized in that: The initial value of the node vector is 0 before it is assigned.

4. The method for monitoring medical record data quality based on a knowledge base according to claim 3, characterized in that: The data contents in different medical record components are either related or unrelated.

5. The method for monitoring medical record data quality based on a knowledge base according to claim 4, characterized in that: Divide the medical record components according to the organizational structure of the medical record data.

6. The method for monitoring medical record data quality based on a knowledge base according to claim 5, characterized in that: Update the medical record quality type of the medical record samples in the knowledge base based on the standardization of medical record writing, accuracy of diagnosis and treatment, completeness of medical record content, timeliness of medical records, legal compliance, data quality, quality of medical record archiving and / or evaluation of TCM characteristics.

7. The method for monitoring medical record data quality based on a knowledge base according to claim 6, characterized in that: 。 8. The method for monitoring medical record data quality based on a knowledge base according to claim 7, characterized in that: The medical records are traditional Chinese medicine or western medicine records.

9. A medical record data quality monitoring system based on a knowledge base, characterized in that: The medical record data quality monitoring system based on the knowledge base is used to implement the medical record data quality monitoring method based on the knowledge base described in any one of claims 1 to 8.

10. A medical record data quality monitoring platform based on a knowledge base, characterized in that: The medical record data quality monitoring platform based on the knowledge base is used to implement the medical record data quality monitoring method based on the knowledge base described in any one of claims 1 to 8.

Images