Intelligent medical record recommendation method and system

Through the matching rule algorithm based on basic feature data and disease tree, the similarity score of each medical record and patient medical record in the case library is calculated, which solves the problems of incompleteness of medical record matching and high labor costs in the existing technology, and realizes intelligent recommendation Medical records that meet the patient's condition will be assisted in diagnosis and treatment.

CN113889280BActive Publication Date: 2024-11-19SHAN DONG MSUN HEALTH TECH GRP CO LTD +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111255283.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-27
Publication Date
2024-11-19
Estimated Expiration
2041-10-27

AI Technical Summary

Technical Problem

The existing technology has problems of incompleteness and high labor costs in the matching of medical records, and the lack of a unified medical record similarity calculation method, which leads to the matching medical records requiring more labor costs to screen, which is not conducive to expanding application and promotion.

Method used

The matching rule algorithm based on basic feature data and disease tree is used to calculate the similarity scores of each medical record and patient medical record in the case library through the structure and deep structure of the medical record, and the medical record with the highest similarity is recommended.

Benefits of technology

It has realized intelligent recommendations of medical records that meet the patient's condition, reduced labor costs, and improved diagnosis and treatment efficiency and medical quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113889280B_ABST
    Figure CN113889280B_ABST
Patent Text Reader

Abstract

The present invention belongs to the technical field of text processing, and provides an intelligent medical record recommendation method and system, which first obtains the patient's medical record, and adopts a medical record structuring algorithm to structure the patient's medical record to obtain multiple basic nodes and composite nodes; then the node content of the basic node is used as the basic feature of the medical record; a medical record deep structuring algorithm is adopted to split the node content of the composite node into multiple paragraphs, and the detail points belonging to each paragraph are matched, and when the detail points belonging to the paragraph are disease detail points, a disease tree is constructed based on the content of the paragraph; finally, based on the basic features and the disease tree, a matching rule algorithm is adopted to calculate the similarity score of each medical record in the case library and the patient's medical record, and the medical record with the highest similarity score in the case library is the recommended medical record, thereby realizing intelligent recommendation of medical records that meet the patient's condition and assisting doctors in diagnosis and treatment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of text processing, and in particular relates to an intelligent medical record recommendation method and system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Analyzing the patient's condition and choosing the appropriate treatment plan to treat the patient is the main work of the doctor. For doctors, especially primary care doctors, due to inertia of thinking or lack of certain medical knowledge, the doctor's choice of treatment plan sometimes has problems. With the application of information technology in the medical field, assisting doctors in diagnosis and treatment by recommending similar medical records is of great significance to improving doctors' work efficiency and improving medical quality.

[0004] At present, most similar medical record matching in the market is still limited to matching through individual field correspondence, which leads to incomplete matching, and there may be many similar medical records that cannot be matched. Secondly, there is no unified medical record similarity calculation method, which leads to more manpower costs to screen the matched medical records, which is not conducive to expanding application and promotion. Summary of the invention

[0005] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides an intelligent medical record recommendation method and system, which, based on basic feature data and disease tree, adopts a matching rule algorithm to realize intelligent recommendation of medical records that meet the patient's condition, thereby assisting doctors in diagnosis and treatment.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A first aspect of the present invention provides an intelligent medical record recommendation method, which includes:

[0008] Obtain the patient's medical records, and use a medical record structuring algorithm to structure the patient's medical records to obtain multiple basic nodes and composite nodes;

[0009] The node content of the basic node is used as the basic feature of the medical record;

[0010] Using a deep structural algorithm of medical records, the node content of the composite node is split into multiple paragraphs, and the detail points belonging to each paragraph are matched. When the detail point belonging to a paragraph is a disease detail point, a disease tree is constructed based on the content of the paragraph;

[0011] Based on the basic features and disease tree, a matching rule algorithm is used to calculate the similarity score between each medical record in the case database and the patient's medical record. The medical record with the highest similarity score in the case database is the recommended medical record.

[0012] Furthermore, the medical record structuring specifically includes:

[0013] Establish a one-to-many mapping relationship between different medical record types and nodes, and package the nodes using regular expression rules;

[0014] Obtain the patient's medical record, determine the type of the patient's medical record, and use a regular matching algorithm to find the character start position and character end position of the node contained in the medical record type in the medical record;

[0015] All nodes are sorted in ascending order according to the size of the character start position. The text content from the character end position of the previous node to the character start position of the next node is the node content of the previous node. Finally, all matched nodes and node contents are obtained.

[0016] Furthermore, the specific steps of matching and obtaining the detail points of each paragraph are as follows:

[0017] A one-to-many mapping relationship is established between different composite nodes and feature identifiers, and each feature identifier belongs to a detail point;

[0018] Obtain the content of a certain paragraph, and use a regular matching algorithm to match the packaged feature identifiers corresponding to the composite node in sequence;

[0019] If the match fails, determine whether the paragraph is the initial paragraph; if so, specify the detail point to which the paragraph belongs; otherwise, use the detail point to which the previous paragraph belongs as the detail point to which the current paragraph belongs;

[0020] If the match is successful, the detail point belonging to the successfully matched feature identifier is used as the initial detail point of the paragraph, and it is determined whether the number of paragraphs between the previous paragraph with the same detail point and this paragraph exceeds the threshold; if so, the detail point belonging to the previous paragraph is used as the detail point belonging to this paragraph; otherwise, the initial detail point of this paragraph is the detail point belonging to this paragraph.

[0021] Furthermore, the matching rule algorithm includes calculating disease similarity, specifically:

[0022] Calculate the matching value between each disease tree of each medical record in the case database and each disease tree of the patient's medical record, between each disease tree of each medical record in the case database and itself, and between each disease tree of the patient's medical record and itself;

[0023] Based on the matching value, the mean similarity between each disease tree of each medical record in the case library and each disease tree of the patient's medical record is calculated to obtain the disease similarity between each medical record in the case library and the patient's medical record.

[0024] Furthermore, the steps for calculating the matching value are:

[0025] Set the weight of each layer of the disease tree;

[0026] Calculate the number of successful node matches and the number of failed node matches in each layer of the two disease trees;

[0027] The matching value is calculated based on the weight of each layer, the number of nodes that successfully match, and the number of nodes that fail to match.

[0028] Furthermore, the matching rule algorithm also includes calculating the basic information similarity, specifically:

[0029] Set the weights of basic features;

[0030] The basic information similarity is calculated based on the weights of the basic features, the basic features of each medical record in the case database, and the basic features of the patient's medical record.

[0031] Furthermore, the similarity score between each medical record in the case database and the patient's medical record is a weighted sum of the basic information similarity and the disease similarity.

[0032] A second aspect of the present invention provides an intelligent medical record recommendation system, comprising:

[0033] The medical record structuring module is configured to: obtain the patient's medical record, and adopt the medical record structuring algorithm to structure the patient's medical record to obtain a plurality of basic nodes and composite nodes;

[0034] A basic feature acquisition module is configured to: use the node content of the basic node as the basic feature of the medical record;

[0035] A disease tree construction module is configured to: use a deep medical record structuring algorithm to split the node content of the composite node into multiple paragraphs, match the detail points belonging to each paragraph, and when the detail point belonging to the paragraph is a disease detail point, construct a disease tree based on the content of the paragraph;

[0036] The similarity score calculation module is configured as follows: based on basic features and disease trees, a matching rule algorithm is used to calculate the similarity score of each medical record in the case library and the patient's medical record. The medical record with the highest similarity score in the case library is the recommended medical record.

[0037] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the intelligent medical record recommendation method as described above.

[0038] The fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the intelligent medical record recommendation method as described above are implemented.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] The present invention provides an intelligent medical record recommendation method, which is based on basic feature data and a disease tree, and adopts a matching rule algorithm to calculate the similarity score between each medical record in the case library and the patient's medical record. The medical record with the highest similarity score in the case library is the recommended medical record, thereby realizing intelligent recommendation of medical records that meet the patient's condition and assisting doctors in diagnosis and treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0042] Figure 1 is a schematic diagram of a dictionary library of admission records in Embodiment 1 of the present invention;

[0043] Figure 2 It is a schematic diagram of the division of detail points and feature identifiers of the past history according to the first embodiment of the present invention;

[0044] Figure 3 This is a feature identifier matching flow chart of Embodiment 1 of the present invention;

[0045] Figure 4 is a schematic diagram of a past history label according to the first embodiment of the present invention;

[0046] Figure 5 is a modified feature identification matching flow chart of the first embodiment of the present invention;

[0047] Figure 6 This is a schematic diagram of a five-level directory of hypertension according to the first embodiment of the present invention;

[0048] Figure 7 This is a schematic diagram of a disease tree in which the fourth layer is the same and the fifth layer is different in embodiment 1 of the present invention. DETAILED DESCRIPTION

[0049] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0050] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0051] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0052] Embodiment 1

[0053] like Figure 1-7 As shown, this embodiment provides an intelligent medical record recommendation method, which provides an algorithm to analyze the patient's complex condition, intelligently recommend medical records that meet the patient's conditions, and assist doctors in diagnosis and treatment. The specific steps include:

[0054] Step (1): Obtain the patient's medical records, and structure the medical records at a node granularity according to the medical record structuring algorithm P to obtain multiple nodes, including basic nodes and composite nodes, as well as the content of each node.

[0055] As an implementation method, the medical record is an electronic medical record.

[0056] As an implementation method, the medical record types include admission records, outpatient medical records, and inpatient medical records.

[0057] As an implementation method, basic nodes include name, gender, age, etc.; composite nodes include current medical history, past medical history, and physical examination, etc.

[0058] The specific implementation steps of step (1), namely the medical record structuring algorithm P, are as follows:

[0059] Medical records are unstructured texts that need to be structured. According to the 2020 version of the Basic Specifications for Medical Record Writing, different types of medical records have fixed node contents. For example, the admission record includes nodes such as name, gender, age, chief complaint, current medical history, past medical history, personal history, marital history, family history, physical examination, auxiliary examination, preliminary diagnosis, and physician signature. Other types of medical records are similar and also have fixed nodes.

[0060] Step (101): Based on the medical record type, establish a one-to-many mapping relationship between different medical record types and nodes to form a medical record node dictionary. Example: Construction of a dictionary library for admission records, such as Figure 1 shown.

[0061] Step (102): Packing the node name. When doctors write medical records, due to various irregular writing methods, sometimes it is impossible to match only by node name. It is necessary to pack the nodes by regular expression rules, so as to match the nodes with special characters in the medical records.

[0062] For example, "Past history:" is sometimes written as:

[0063] Past history:

[0064] Past history:

[0065] Past History:

[0066] etc.

[0067] It needs to be packaged as:

[0068] [\s| | ]*both[\s| | ]*past[:|:|\s | | ]*history[:|:|\s| | ]+

[0069] Step (103): load the medical record text and determine the type of the patient's medical record. According to the packaged medical record node name, use the regular matching algorithm to find the character start position (character start position value) and character end position (character end position value) of the node contained in the medical record of the medical record type; arrange all nodes in ascending order according to the size of the character start position (node ​​position value size). The text content from the character end position of the previous node to the character start position of the next node is the node content of the previous node, and so on. Finally, all matched nodes and node contents are obtained.

[0070] Step (2): According to the medical record deep structuring algorithm Q, the node content of the composite node (past history, physical examination content) is structured to obtain the patient's disease history, allergy history, physical signs and other data. Specifically: Using the medical record deep structuring algorithm, the node content of the composite node is split into multiple paragraphs, and the detail points belonging to each paragraph are matched.

[0071] Step (2), namely the medical record deep structuring algorithm Q, is specifically implemented as follows:

[0072] According to the result of step (1), the structured content of the medical record at the node granularity is obtained, that is, the structured content of the composite node (past history, physical examination content) is obtained. According to the "2020 Basic Specifications for Medical Record Writing", the content of the past history includes: previous general health status, disease history, infectious disease history, vaccination history, surgical trauma history, blood transfusion history, drug allergy history and other related descriptions; the content of the physical examination includes: general condition, skin mucosa, lymph nodes, etc., which is similar to the past history and also has fixed writing content and mode.

[0073] For example, the contents of the past medical history in the routine medical record are as follows: The patient has a history of "coronary heart disease, hypertension, heart failure and gastrointestinal bleeding". He denies a history of other chronic diseases such as cerebral infarction, diabetes, chronic bronchitis, etc. He denies a history of hepatitis, tuberculosis and close contact, no history of drug and food allergies, no history of trauma and surgery, no history of blood transfusion, and the vaccination history is unknown.

[0074] Step (201): Establish a one-to-many mapping relationship between different composite nodes and feature identifiers, and each feature identifier belongs to a detail point; Figure 2 Shown are the characteristic identifiers included in the past history and the detail points to which the characteristic identifiers belong.

[0075] Step (202): Split the node content of the composite node (eg, the content of the past history) into multiple sections using separators.

[0076] As an implementation method, the separator may be a comma, a period, or a semicolon punctuation mark (note: Chinese and English symbols are not distinguished).

[0077] For example (separated by "|"): The patient has a history of "coronary heart disease, hypertension, heart failure and gastrointestinal bleeding". | Denies a history of other chronic diseases such as cerebral infarction, diabetes, chronic bronchitis, etc. | Denies a history of hepatitis, tuberculosis and close contact, | No history of drug and food allergies, | No history of trauma and surgery, | No history of blood transfusion, | Vaccination history is unknown.

[0078] Step (203): Figure 3 As shown, according to the same principle as step (102), the feature identifiers are packaged, and for each paragraph of content, the packaged feature identifier names are matched in sequence through the regular matching algorithm. If a match is found, the paragraph is labeled with a standard node (that is, the detail point to which the successfully matched feature identifier belongs is used as the initial detail point to which the paragraph belongs). If a match is not found, it is determined whether it is the initial paragraph. If so, the detail point to which the paragraph belongs is specified (for example, it is classified as general health status). If not, it is classified as the label corresponding to the previous paragraph (that is, the detail point to which it belongs).

[0079] For example, for the description of this past history, according to step (203), the label of each section is as follows: Figure 4 shown.

[0080] Step (204): For special cases, if there are paragraphs with other labels in the middle of multiple feature identifiers for the same label, according to the format requirements of the medical record standard, the patient conditions with the same parts or features should be described together. Therefore, the above algorithm is modified. When a paragraph with the same label (belonging to the same detail point) is interspersed with two or more paragraphs with other labels (i.e., belonging to other detail points), the following label is changed to the label of the previous paragraph immediately adjacent.

[0081] In summary, if Figure 5 As shown, the specific steps for matching the detail points belonging to each paragraph are: establishing a one-to-many mapping relationship between different composite nodes and feature identifiers, and each feature identifier belongs to a detail point; obtaining the content of a certain paragraph, and using a regular matching algorithm to match it with the packaged feature identifiers corresponding to the composite node in turn; if the match is not successful, then determine whether the paragraph is the initial paragraph; if so, specify the detail point to which the paragraph belongs; otherwise, use the detail point to which the previous paragraph belongs as the detail point to which the paragraph belongs; if the match is successful, then use the detail point to which the successfully matched feature identifier belongs as the initial detail point to which the paragraph belongs, and determine whether the number of paragraphs between the previous paragraph belonging to the same detail point and the paragraph exceeds a threshold; if so, use the detail point to which the previous paragraph belongs as the detail point to which the paragraph belongs; otherwise, the initial detail point to which the paragraph belongs is the detail point to which the paragraph belongs.

[0082] Step (205): The node contents of other compound nodes (such as the contents of physical examination) are similarly structured using the same algorithm to obtain the symptom data of other compound nodes (i.e., the details of each paragraph in the node contents of the compound nodes and the contents of each paragraph).

[0083] Step (3): Construct a feature set N[...] for each medical record in the structured original medical record and the medical record database, and calculate the medical record similarity based on the matching rule algorithm S.

[0084] The specific implementation steps of step (3) are as follows:

[0085] Step (301): construct a patient feature data set based on the medical record data structured in steps (1)-(2); construct a feature data set for the medical records in the medical record library in the same way.

[0086] Among them, the feature data includes basic feature data and disease detail point data. The node content of the basic node constitutes the basic feature data of the medical record, including the patient's unique identifier, age, and gender; the disease detail point includes medical history, allergy history, and physical signs, etc. The disease detail point data is the name of the disease detail point and the content of the corresponding paragraph of the disease detail point.

[0087] Step (302): Obtain similar medical records from the medical record database based on the patient information using a similarity algorithm.

[0088] Among them, the similarity algorithm, that is, the matching rule algorithm S, has the following overall steps:

[0089] ① Calculate patient similarity based on medical history;

[0090] ② Calculate similarity based on basic information, age, and gender;

[0091] ③The similarity between the two medical records is S = ∑S i ·w i , where S i are the similarities of medical records in basic information, medical history, physical signs, etc., w i It is the weight of basic information, medical history, physical signs and other aspects in the total similarity.

[0092] Note: The data input into the medical record is X, and a piece of data in the medical record database is Y.

[0093] The first step is to calculate the disease similarity: when the detail point to which the paragraph belongs is a disease detail point, a disease tree is constructed based on the content of the paragraph; the matching values ​​between each disease tree of each medical record in the case library and each disease tree of the patient's medical record, between each disease tree of each medical record in the case library and itself, and between each disease tree of the patient's medical record and itself are calculated; based on the matching values, the mean similarity between each disease tree of each medical record in the case library and each disease tree of the patient's medical record is calculated to obtain the disease similarity between each medical record in the case library and the patient's medical record. Specifically, the steps of the disease similarity algorithm are as follows:

[0094] (a) The diseases corresponding to each paragraph of the input medical record X and the comparison medical record Y are constructed into two disease trees according to the provincial coding standard. When the patient has a certain disease, the corresponding node value in the disease tree is 1, otherwise it is 0. When the leaf node is 1, its ancestor nodes are all 1.

[0095] Note: According to the provincial coding standard, each diagnosis has five levels of directory. Take hypertension as an example. Figure 6 shown.

[0096] (b) Set the weight of each layer of the disease tree. When X and Y have the same diagnosis in the first layer (i.e. (1, 1)), the weight is w1_1 (greater than 1), the second layer is w2_1...the fifth layer is w5_1; when the first layer has different diagnoses (i.e. (1, 0) or (0, 1)), the weight is w1_2 (less than 1), the second layer is w2_2...the fifth layer is w5_2; when neither of them has a certain diagnosis, that is, the corresponding node is (0, 0), the weight is 1.

[0097] (c) Calculate the number of successful node matches and the number of failed node matches in each layer of the two disease trees. Specifically: match the two tree structure data, and let the number of nodes in the i-th layer that match (1, 1) be di_1, and the number of nodes that match (1, 0) or (0, 1) be di_2.

[0098] (d) Calculate the matching value based on the weight of each layer, the number of nodes that successfully match, and the number of nodes that fail to match:

[0099] S 12 =w1_1 d1_1 ×w1_2 d1_2 ×w2_1 d2_1 ×w2_2 d2_2 ×w3_1 d3_1 ×w3_2 d3_2 ×w4_1 d4_1 ×w4_2d 4_2 ×w5_1 d5_1 ×w5_2 d5_2

[0100] (e) Obtain similarity:

[0101]

[0102] Among them, S 11 is the matching value between medical record X and itself, S 22 is the matching value between medical record Y and itself;

[0103] (f) Repeat the above steps to obtain the similarity between each disease tree of each medical record in the case library and each disease tree of the patient's medical record, calculate the mean similarity, obtain the disease similarity between each medical record in the case library and the patient's medical record, and find the patient who is most similar to it.

[0104] Note: The weight w1_1=1.316 when the first layer has the same diagnosis and all are 1; the weight w1_2=0.76 when the first layer has different diagnoses; the weight w2_1=1.147 when the second layer has the same diagnosis and all are 1; the weight w2_2=0.872 when the second layer has different diagnoses; the weight w3_1=1.071 when the third layer has the same diagnosis and all are 1; the weight w3_2=0.934 when the third layer has different diagnoses; the weight w4_1=1.035 when the fourth layer has the same diagnosis and all are 1; the weight w4_2=0.966 when the fourth layer has different diagnoses; the weight w5_1=1.017 when the fifth layer has the same diagnosis and all are 1; the weight w5_2=0.983 when the fifth layer has different diagnoses.

[0105] Weight source (conditions that weights meet):

[0106] ①w1_1×w1_2=1, w2_1×w2_2=1,…, w5_1×w5_2=1;

[0107] ②w5_2×w5_2=w4_2, w5_1×w5_1=w4_1,…;

[0108] ③If Figure 7 As shown in , when the fourth layer is the same and the fifth layer is different, their similarity is considered to be 95%, that is,

[0109] The second step is to calculate the basic information similarity, specifically: set the weight of the basic features; calculate the basic information similarity based on the weight of the basic features, the basic feature data of each medical record in the case database and the basic feature data of the patient's medical record. The basic information similarity algorithm steps are as follows:

[0110] (a) Gender: same as 1, different as 0, recorded as basic1;

[0111] (b) Age: If the age is the same, the value is 1; if the age is different, the value is 0.02 for every year difference. If the age difference is 50 years, the value is 0.02.

[0112] The above considers that the ages are completely dissimilar, recorded as basic2;

[0113] (c) Basic score basic = basic1×0.17+basic2×0.83.

[0114] Note: Based on the similarity of a 10-year age difference and gender differences, the weight of gender is 0.17 and the weight of age is 0.83.

[0115] In the third step, the similarity score of each medical record in the case database and the patient's medical record is the weighted sum of the basic information similarity and the disease similarity. Specifically: according to the scores and weights of the medical records and basic information, the similarity of the two medical records is calculated as S = ∑S i ·w i .

[0116] Step (4): The medical record with the highest similarity score is the medical record that best matches the patient's condition, that is, the medical record with the highest similarity score in the case database is the recommended medical record.

[0117] The present invention analyzes the patient's complex condition and intelligently recommends medical records that are suitable for the patient to assist doctors in diagnosis and treatment.

[0118] Embodiment 2

[0119] This embodiment provides an intelligent medical record recommendation system, which specifically includes the following modules:

[0120] The medical record structuring module is configured to: obtain the patient's medical record, and adopt the medical record structuring algorithm to structure the patient's medical record to obtain a plurality of basic nodes and composite nodes;

[0121] A basic feature acquisition module is configured to: use the node content of the basic node as the basic feature of the medical record;

[0122] A disease tree construction module is configured to: use a deep medical record structuring algorithm to split the node content of the composite node into multiple paragraphs, match the detail points belonging to each paragraph, and when the detail point belonging to the paragraph is a disease detail point, construct a disease tree based on the content of the paragraph;

[0123] The similarity score calculation module is configured as follows: based on basic features and disease trees, a matching rule algorithm is used to calculate the similarity score of each medical record in the case library and the patient's medical record. The medical record with the highest similarity score in the case library is the recommended medical record.

[0124] It should be noted here that each module in this embodiment corresponds to each step in Example 1 one by one, and the specific implementation process is the same, which will not be repeated here.

[0125] Embodiment 3

[0126] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the intelligent medical record recommendation method described in the above-mentioned embodiment 1 are implemented.

[0127] Embodiment 4

[0128] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of an intelligent medical record recommendation method as described in the first embodiment are implemented.

[0129] 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 hardware embodiments, software embodiments, or embodiments combining software and hardware. Furthermore, 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 and optical storage, etc.) containing computer-usable program code.

[0130] 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.

[0131] 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.

[0132] 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.

[0133] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0134] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent medical record recommendation method, characterized in that: include: Obtain the patient's medical records, and use a medical record structuring algorithm to structure the patient's medical records to obtain multiple basic nodes and composite nodes; The node content of the basic node is used as the basic feature of the medical record; Using a deep structural algorithm of medical records, the node content of the composite node is split into multiple paragraphs, and the detail points belonging to each paragraph are matched. When the detail point belonging to a paragraph is a disease detail point, a disease tree is constructed based on the content of the paragraph; Based on the basic features and disease tree, a matching rule algorithm is used to calculate the similarity score between each medical record in the case database and the patient's medical record. The medical record with the highest similarity score in the case database is the recommended medical record. The matching rule algorithm includes calculating disease similarity, specifically: Calculate the matching value between each disease tree of each medical record in the case database and each disease tree of the patient's medical record, between each disease tree of each medical record in the case database and itself, and between each disease tree of the patient's medical record and itself; Based on the matching value, the mean similarity between each disease tree of each medical record in the case database and each disease tree of the patient's medical record is calculated to obtain the disease similarity between each medical record in the case database and the patient's medical record; The steps to calculate the matching value are: Set the weight of each layer of the disease tree; Calculate the number of successful node matches and the number of failed node matches in each layer of the two disease trees; The matching value is calculated based on the weight of each layer, the number of nodes that successfully match, and the number of nodes that fail to match.

2. The intelligent medical record recommendation method according to claim 1, characterized in that: The medical record structuring specifically includes: Establish a one-to-many mapping relationship between different medical record types and nodes, and package the nodes using regular expression rules; Obtain the patient's medical record, determine the type of the patient's medical record, and use a regular matching algorithm to find the character start position and character end position of the node contained in the medical record type in the medical record; All nodes are sorted in ascending order according to the size of the character start position. The text content from the character end position of the previous node to the character start position of the next node is the node content of the previous node. Finally, all matched nodes and node contents are obtained.

3. The intelligent medical record recommendation method according to claim 1, characterized in that: The specific steps of matching and obtaining the detail points of each paragraph are as follows: A one-to-many mapping relationship is established between different composite nodes and feature identifiers, and each feature identifier belongs to a detail point; Obtain the content of a certain paragraph, and use a regular matching algorithm to match the packaged feature identifiers corresponding to the composite node in sequence; If the match fails, determine whether the paragraph is the initial paragraph; if so, specify the detail point to which the paragraph belongs; otherwise, use the detail point to which the previous paragraph belongs as the detail point to which the current paragraph belongs; If the match is successful, the detail point belonging to the successfully matched feature identifier is used as the initial detail point of the paragraph, and it is determined whether the number of paragraphs between the previous paragraph with the same detail point and this paragraph exceeds the threshold; if so, the detail point belonging to the previous paragraph is used as the detail point belonging to this paragraph; otherwise, the initial detail point of this paragraph is the detail point belonging to this paragraph.

4. The intelligent medical record recommendation method according to claim 1, characterized in that: The matching rule algorithm also includes calculating the basic information similarity, specifically: Set the weights of basic features; The basic information similarity is calculated based on the weights of the basic features, the basic features of each medical record in the case database, and the basic features of the patient's medical record.

5. The intelligent medical record recommendation method according to claim 4, characterized in that: The similarity score between each medical record in the case database and the patient's medical record is the weighted sum of the basic information similarity and the disease similarity.

6. An intelligent medical record recommendation system, characterized in that: include: The medical record structuring module is configured to: obtain the patient's medical record, and adopt the medical record structuring algorithm to structure the patient's medical record to obtain a plurality of basic nodes and composite nodes; A basic feature acquisition module is configured to: use the node content of the basic node as the basic feature of the medical record; A disease tree construction module is configured to: use a deep medical record structuring algorithm to split the node content of the composite node into multiple paragraphs, match the detail points belonging to each paragraph, and when the detail point belonging to the paragraph is a disease detail point, construct a disease tree based on the content of the paragraph; A similarity score calculation module is configured to: calculate the similarity score of each medical record in the case library and the patient's medical record based on the basic features and the disease tree, using a matching rule algorithm, and the medical record with the highest similarity score in the case library is the recommended medical record; the matching rule algorithm includes calculating the disease similarity, specifically: calculating the matching value between each disease tree of each medical record in the case library and each disease tree of the patient's medical record, between each disease tree of each medical record in the case library and itself, and between each disease tree of the patient's medical record and itself; Based on the matching value, the mean similarity between each disease tree of each medical record in the case database and each disease tree of the patient's medical record is calculated to obtain the disease similarity between each medical record in the case database and the patient's medical record; the matching value calculation steps are: setting the weight of each layer of the disease tree; calculating the number of successful node matches and the number of failed node matches of the two disease trees at each layer; The matching value is calculated based on the weight of each layer, the number of nodes that successfully match, and the number of nodes that fail to match.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in an intelligent medical record recommendation method as described in any one of claims 1 to 5 are implemented.

8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the intelligent medical record recommendation method as described in any one of claims 1-5 are implemented.