CN110299209A - Similar case history lookup method, device, equipment and readable storage medium storing program for executing - Google Patents

Similar case history lookup method, device, equipment and readable storage medium storing program for executing Download PDF

Info

Publication number
CN110299209A
CN110299209A CN201910557217.5A CN201910557217A CN110299209A CN 110299209 A CN110299209 A CN 110299209A CN 201910557217 A CN201910557217 A CN 201910557217A CN 110299209 A CN110299209 A CN 110299209A
Authority
CN
China
Prior art keywords
subgraph
similarity
data
history
class
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201910557217.5A
Other languages
Chinese (zh)
Other versions
CN110299209B (en
Inventor
魏巍
陈俊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Baidu Netcom Science and Technology Co Ltd
Original Assignee
Beijing Baidu Netcom Science and Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Baidu Netcom Science and Technology Co Ltd filed Critical Beijing Baidu Netcom Science and Technology Co Ltd
Priority to CN201910557217.5A priority Critical patent/CN110299209B/en
Publication of CN110299209A publication Critical patent/CN110299209A/en
Application granted granted Critical
Publication of CN110299209B publication Critical patent/CN110299209B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2455—Query execution
    • G06F16/24564—Applying rules; Deductive queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Abstract

The present invention provides a kind of similar case history lookup method, device, equipment and readable storage medium storing program for executing, passes through and obtains inquiry medical record data and multiple history medical record datas；Obtain the corresponding query graph structured data of inquiry medical record data, and the corresponding history graph structure data of each history medical record data, wherein, query graph structured data and history graph structure data all include first kind subgraph and the second class subgraph, and the intermediate node and leaf node of the second class subgraph are to carry out feature to first kind subgraph to identify；According to root node similarity, first kind subgraph similarity and the second class subgraph similarity, the similarity degree of each history graph structure data and query graph structured data is obtained；According to default selection rule and similarity degree, determine the similar case history lookup result of inquiry medical record data, to extract intrinsic and identifiable subgraph in medical record data, the relevance of corresponding sub-picture content is measured in the comparison, improves the accuracy that similar case history is searched.

Description

Similar case history lookup method, device, equipment and readable storage medium storing program for executing
Technical field
The present invention relates to technical field of information processing more particularly to a kind of similar case history lookup method, device, equipment and can Read storage medium.
Background technique
In medical field, similar case history retrieval is of great significance in scientific research, clinically.Such as in patient assessment, doctor Life can quickly search case history similar with the patient, and can be made effectively by the diagnosis and treatment path of similar case history and effect in time Judgement；Alternatively, doctor when carrying out analysis of medical record for certain part of case history or writing Case report no, can have one by using for reference The history case history for determining similarity therefrom obtains some diagnostic comments and treatment method that can refer to；Alternatively, in clinical research, It needs to find more similar case histories as starting point from certain part of case history and carry out research discussion in some cases.
Current case history matches retrieval mode, the usually retrieval to case history full text information.For example, to generate heat, breathe not Freely retrieved for keyword, it can be by all case histories in pre-stored case history with fever, unsmooth breath the two keywords All retrieve.
But due to same symptom, its corresponding disease is very different, existing similar case history retrieval mode it is accurate Property is not high.
Summary of the invention
The embodiment of the present invention provides a kind of similar case history lookup method, device, equipment and readable storage medium storing program for executing, improves phase The accuracy and reliability searched like case history.
According to the first aspect of the invention, a kind of similar case history lookup method is provided, comprising:
Obtain inquiry medical record data and multiple history medical record datas；
It obtains the corresponding query graph structured data of the inquiry medical record data and each history medical record data is corresponding History graph structure data, wherein the query graph structured data and the history graph structure data all include first kind subgraph and Second class subgraph, the intermediate node of the first kind subgraph are case history field classification, the intermediate node of the second class subgraph and Leaf node is to carry out feature to the first kind subgraph to identify；
According to root node similarity, first kind subgraph in each history graph structure data and the query graph structured data Similarity and the second class subgraph similarity obtain the similar journey of each the history graph structure data and the query graph structured data Degree；
According to default selection rule and the similarity degree, the inquiry disease is determined in the multiple history medical record data Count the similar case history lookup result of evidence one by one, wherein the corresponding history graph structure data of the similar case history lookup result, tool There is the similarity degree for meeting the default selection rule.
According to the second aspect of the invention, a kind of similar case history lookup device is provided, comprising:
Case history obtains module, for obtaining inquiry medical record data and multiple history medical record datas；
Graph structure module, for obtaining the corresponding query graph structured data of the inquiry medical record data and each described The corresponding history graph structure data of history medical record data, wherein the query graph structured data and the history graph structure data It all include first kind subgraph and the second class subgraph, the intermediate node of the first kind subgraph is case history field classification, described second The intermediate node and leaf node of class subgraph are to carry out feature to the first kind subgraph to identify；
Processing module, for similar to root node in the query graph structured data according to each history graph structure data Degree, first kind subgraph similarity and the second class subgraph similarity obtain each history graph structure data and the query graph knot The similarity degree of structure data；
Selecting module, for selecting rule and the similarity degree according to default, in the multiple history medical record data Determine the similar case history lookup result of the inquiry medical record data, wherein go through described in the similar case history lookup result is corresponding History graph structure data have the similarity degree for meeting the default selection rule.
According to the third aspect of the invention we, a kind of equipment is provided, comprising: memory, processor and computer program, institute State computer program storage in the memory, the processor runs the computer program and executes first aspect present invention And the similar case history lookup method of the various possible designs of first aspect.
According to the fourth aspect of the invention, a kind of readable storage medium storing program for executing is provided, meter is stored in the readable storage medium storing program for executing Calculation machine program, for realizing first aspect present invention and the various possibility of first aspect when the computer program is executed by processor The similar case history lookup method of design.
A kind of similar case history lookup method, device, equipment and readable storage medium storing program for executing provided by the invention are inquired by obtaining Medical record data and multiple history medical record datas；Obtain the corresponding query graph structured data of the inquiry medical record data and each institute State the corresponding history graph structure data of history medical record data, wherein the query graph structured data and the history graph structure number According to all including first kind subgraph and the second class subgraph, the intermediate node of the first kind subgraph is case history field classification, described the The intermediate node and leaf node of two class subgraphs are to carry out feature to the first kind subgraph to identify；It described is gone through according to each History graph structure data and root node similarity, first kind subgraph similarity and the second class subgraph phase in the query graph structured data Like degree, the similarity degree of each the history graph structure data and the query graph structured data is obtained；According to default selection rule With the similarity degree, determine that the similar case history of the inquiry medical record data searches knot in the multiple history medical record data Fruit, to extract subgraph intrinsic in inquiry medical record data and can recognize obtained subgraph, to inquiry medical record data and history The relevance of data is measured in corresponding subgraph in medical record data case history, improves the accuracy that similar case history is searched.
Detailed description of the invention
Fig. 1 is a kind of similar case history lookup method flow diagram provided in an embodiment of the present invention；
Fig. 2 is a kind of schematic diagram of query graph structured data provided in an embodiment of the present invention；
Fig. 3 is the schematic diagram of a kind of first kind subgraph and the second class subgraph provided in an embodiment of the present invention；
Fig. 4 is step S103 optional embodiment flow diagram in a kind of Fig. 1 provided in an embodiment of the present invention；
Fig. 5 is that a kind of similar case history provided in an embodiment of the present invention searches apparatus structure schematic diagram；
Fig. 6 is a kind of hardware structural diagram of equipment provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only It is only a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill Personnel's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Description and claims of this specification and term " first " in above-mentioned attached drawing, " second " etc. are for distinguishing Similar object, without being used to describe a particular order or precedence order.It should be understood that the data used in this way are in appropriate feelings It can be interchanged under condition, so that the embodiment of the present invention described herein can be other than those of illustrating or describing herein Sequence implement.
It should be appreciated that in various embodiments of the present invention, the size of the serial number of each process is not meant to execute sequence It is successive, the execution of each process sequence should be determined by its function and internal logic, the implementation without coping with the embodiment of the present invention Journey constitutes any restriction.
It should be appreciated that in the present invention, " comprising " and " having " and their any deformation, it is intended that covering is not arranged His includes, for example, the process, method, system, product or equipment for containing a series of steps or units are not necessarily limited to clearly Those of list step or unit, but may include be not clearly listed or for these process, methods, product or equipment Intrinsic other step or units.
It should be appreciated that in the present invention, " multiple " refer to two or more.
It should be appreciated that in the present invention, " B corresponding with A ", " B corresponding with A ", " A and B are corresponding " or " B and A It is corresponding ", it indicates that B is associated with A, B can be determined according to A.It determines that B is not meant to determine B only according to A according to A, may be used also To determine B according to A and/or other information.The matching of A and B is that the similarity of A and B is greater than or equal to preset threshold value.
Depending on context, as used in this " if " can be construed to " ... when " or " when ... " or " in response to determination " or " in response to detection ".
Technical solution of the present invention is described in detail with specifically embodiment below.These specific implementations below Example can be combined with each other, and the same or similar concept or process may be repeated no more in some embodiments.
In a medical record data, main diagnostic message, multiple case histories intrinsic field and field contents are generally included.This A little field contents are the content that sufferer or doctor fill in field name in standard specification case history, field name for example, gender, year Age, department, consultation time, diagnosis and treatment type, doctor's advice, main suit, present illness history, physical examination, auxiliary examination.It is matched in current case history In retrieval mode, usually case history full text information or these field contents are retrieved.For example, with " fever ", " breathing is not Freely " retrieved for keyword, it can be by all case histories in pre-stored case history with fever, unsmooth breath the two keywords All retrieve.Also can specify will include the case history of keyword " fever " in main suit as lookup result.However, same main suit Entirely different symptom and disease may be corresponded to, such as fever caused by fever caused by flu and allergy, therapeutic modality and Diagnostic mode differs greatly, the value compared without similitude case history.Therefore existing similar case history retrieval mode is accurate Property is not high.
Accuracy in order to solve the problems, such as existing similar case history retrieval mode is not high, and the embodiment of the present invention provides one kind Similar case history lookup method has query graph structured data and the history of first kind subgraph and the second class subgraph by constructing Graph structure data, the second class subgraph is to carry out feature to the first kind subgraph to identify, finally similar according to root node Degree, first kind subgraph similarity and the second class subgraph similarity determine the phase between inquiry medical record data and history medical record data Like degree, the accuracy that similar case history is searched is improved.
It is a kind of similar case history lookup method flow diagram provided in an embodiment of the present invention referring to Fig. 1, side shown in Fig. 1 The executing subject of method can be software and/or hardware device, such as can be understood as server.Method shown in FIG. 1 includes Step S101 is specific as follows to step S104:
S101 obtains inquiry medical record data and multiple history medical record datas.
For example, server obtains multiple history when receiving the inquiry medical record data for needing to carry out similar case history lookup Medical record data.History medical record data can be pre-stored in case history library, be also possible to obtain from distributed storage unit 's.In some embodiments, it can be using medical record data relevant to inquiry medical record data as history medical record data.For example, Server obtains inquiry medical record data；Then according to the inquiry medical record data, full-text search keyword is determined.Full-text search is closed Keyword is, for example, to extract the words such as obtained such as " flu ", " cough ", " XX drug allergy " from inquiry medical record data.Finally Case history full text is carried out in case history library according to the full-text search key to search, is obtained more comprising the full-text search keyword A history medical record data.
S102 obtains the corresponding query graph structured data of inquiry medical record data and each history medical record data Corresponding history graph structure data, wherein the query graph structured data and the history graph structure data all include the first kind Subgraph and the second class subgraph, the intermediate node of the first kind subgraph are case history field classification, the centre of the second class subgraph Node and leaf node are to carry out feature to the first kind subgraph to identify.
Such as using Feature Engineering structured patient record data, figure is generated respectively to inquiry medical record data and history medical record data Structured data body obtains query graph structured data and history graph structure data.Wherein, history graph structure data can be to obtain The same mode structuring of query graph structured data obtains, be also possible in advance from outside obtain and with history medical record data pair It should store, it is not limited here.The present embodiment is illustrated by taking the acquisition process of query graph structured data as an example.
It referring to fig. 2, is a kind of schematic diagram of query graph structured data provided in an embodiment of the present invention.The present embodiment is to inquire For the corresponding query graph structured data of medical record data, arrow is directed toward respective father node from child node in Fig. 2, based on root node Diagnosis.
Usually in inquiry medical record data, including main diagnosis, essential information and basic patient's condition three categories data.Main diagnosis For the data of text type.Essential information for example may include the project for needing doctor to insert: age and gender.Basic patient's condition example It such as may include the project for needing doctor to insert: doctor's advice, physical examination, diagnosis and treatment type, main suit, auxiliary examination, department, present illness history And consultation time.
For example, can be using the main diagnosis of the inquiry case history as the description information of root node.And the centre under root node Node can be divided into two classes, first kind intermediate node can be the field classification in inquiry medical record data, such as essential information with And the basic patient's condition, referring to fig. 2.Second class intermediate node can be preset feature, such as be intended to the raw project inserted of taking up a job as a doctor Knowledge another characteristic is carried out in content.As shown in Fig. 2, the second class intermediate node for example may include: allergy, operation, drug, crowd Attribute, symptom, disease, sign, inspection, inspection.
For first kind intermediate node, the case history field classification for including using the inquiry case history can be as in the first kind Intermediate node, and using the field that the field classification includes as the first kind leaf node of the field classification, it will be described The corresponding description information of field, the description information as the first kind leaf node.Such as field classification is " essential information ", Then by " essential information " be used as first kind intermediate node, and will " age " and " gender " respectively as the first kind intermediate node it Under 2 first kind leaf nodes.And the leaf node of first kind intermediate node " the basic patient's condition " then includes " doctor's advice ", " physique Inspection ", " diagnosis and treatment type ", " main suit ", " auxiliary examination ", " department ", " present illness history " and " consultation time " 8 first kind leaves Node, graph structure data shown in Figure 2.
It, first can be using preset feature as the second class intermediate node for the second class intermediate node.Such as shown in Fig. 2 Example in, by preset feature " symptom " be used as a second class intermediate node.Second class intermediate node can be to be set in advance Fixed, it needs to carry out knowledge another characteristic.Server is with the feature to each field and each word in inquiry medical record data The description information of section carries out the identification of natural language understanding NLU, obtains the description information of the feature, or obtain the feature Description information and the description information plus factor attribute information or multiply factor attribute information.Such as according to present illness history and master In the description information for fields and the field such as telling, NLU processing is carried out to feature " symptom ", has obtained cough, cough degree (slight) Duration (1 month), fever and the fever time (current 24 hours) of cough.It is possible to by description information " cough " " fever " is as 2 the second class leaf nodes under the second class intermediate node " symptom ".It is retouched and it is possible to " will have " and be used as That states information " cough " multiplies factor attribute information, by " degree is slight " and " continuing 1 month " as description information " cough " Add factor attribute information.Likewise it is possible to which " will continue 24 hours " as description information " fever " adds factor attribute information.Again Such as in the description information according to the fields such as present illness history and main suit and field, NLU processing is carried out to feature " crowd's attribute ", is obtained To the elderly.It so can be by description information " the elderly " as the second class leaf under the second class intermediate node " crowd's attribute " Child node.After the description information of feature has been determined, server can be using the description information of the feature as the feature Second class leaf node, wherein there is described plus factor attribute information or multiply the second class leaf node of factor attribute information Data type is complex data type.
Finally, according to the root node obtained above, the first kind intermediate node, the first kind leaf node, The second class intermediate node and the second class leaf node, the corresponding query graph of the available inquiry medical record data Structured data, wherein the root node, the first kind intermediate node and the first kind leaf node form first kind Figure；The root node, the second class intermediate node and the second class leaf node form the second class subgraph.Referring to Fig. 3, It is the schematic diagram of a kind of first kind subgraph and the second class subgraph provided in an embodiment of the present invention.2 first kind are illustrated in Fig. 3 Figure and 2 the second class subgraphs.The root node of first kind subgraph and the second class subgraph is all main diagnosis, but first kind subgraph First kind intermediate node is field classification, such as essential information or the basic patient's condition, and the second class intermediate node of the second class subgraph It is characterized, such as symptom or sign.First kind subgraph is using leaf node as cutting, and each first kind subgraph includes single A kind of leaf node.And the second class subgraph is then using intermediate node as cutting, a second class subgraph includes in single second class Intermediate node, but may include multiple second class leaf nodes.
After obtaining above-mentioned query graph structured data, for the ease of subsequent using subgraph as the similarity measurement of comparing unit, Query graph structured data first can be split as multiple first kind subgraphs and multiple second class subgraphs.
S103, according to root node similarity, first in each history graph structure data and the query graph structured data Class subgraph similarity and the second class subgraph similarity obtain each history graph structure data and the query graph structured data Similarity degree.
In case history similarity-rough set, the main diagnosis that node is added is entirely different, then is usually not belonging to similar case history, therefore Using root node similarity as history graph structure data compared with the similarity degree of query graph structured data according to one of, Neng Gouti The accuracy that high similar case history is searched.
Specifically, root node, first kind subgraph and the second class subgraph for obtaining history graph structure data, Yi Jicha be can be Root node, first kind subgraph and the second class subgraph for asking graph structure data, then relatively obtain root node with the root node of the two Similarity, the first kind subgraph of the two relatively obtain first kind subgraph similarity, and the second class subgraph of the two relatively obtains second Class subgraph similarity.
In order to clearly illustrate above-mentioned steps S103 (according to each history graph structure data and the query graph knot Root node similarity, first kind subgraph similarity and the second class subgraph similarity in structure data obtain each history graph structure The similarity degree of data and the query graph structured data), it is illustrated in the following with reference to the drawings and specific embodiments.
It referring to fig. 4, is step S103 optional embodiment flow diagram in a kind of Fig. 1 provided in an embodiment of the present invention. In method shown in Fig. 4, including step S201 is to step S204, specific as follows:
S201 is determined according to the root node similarity in the history graph structure data and the query graph structured data First similarity magnitude of the history graph structure data and the query graph structured data.
For example, can be directly by root node similarity, as history graph structure data and the first of query graph structured data Similarity magnitude.
Since the main diagnosis of root node is usually text type data, before determining the first similarity magnitude, either Before step 103 shown in Fig. 1, retouching for model and the root node first can also be determined according to preset text-type similarity Information is stated, determines root node similarity in each history graph structure data and the query graph structured data.Text-type is similar It spends and determines that model for example can be similarity corresponding with text type data shown in following table one and determine model.
S202, according to the conspicuousness classification of leaf node, in the first kind subgraph, determine the significant subgraph of the first kind and The non-significant subgraph of the first kind.
Design application demand when the conspicuousness classification of leaf node can be according to practical similarity mode is set. Assuming that design application demand are as follows: " age ", " gender ", " diagnosis and treatment type ", the screening item that " consultation time " is similar case history result Part.For example, can will include " age ", " gender ", " consultation time ", " diagnosis and treatment type " first kind subgraph, as the first kind Significant subgraph.Wherein, " gender ", the similar case history that " diagnosis and treatment type " is Boolean data type match screening conditions, such as: it is similar Case history screening conditions are " gender: male ", then the result output of similar case history retrieval matching " gender: male ".Wherein, " age ", " consultation time " is that the similar case history of numeric data type screens matching condition, and does similar case history using numerical reduction function With screening, such as: similar case history screening conditions are " age: 20 years old ", and similar case history does decaying matching screening for " age ", false If similar other occurrences of case history search result are almost the same, then " age " and 20 years old are closer, similarity degree is higher.At this It can will include the first kind of " doctor's advice ", " physical examination ", " main suit ", " auxiliary examination ", " department ", " present illness history " in embodiment Subgraph, as the non-significant subgraph of the first kind.
S203, it is similar to the significant subgraph of the first kind in the query graph structured data according to the history graph structure data Degree, determines the second similarity magnitude of the history graph structure data Yu the query graph structured data.
History graph structure data and query graph structured data all may include the significant subgraph of multiple first kind, history figure knot The significant subgraph of each first kind of structure data can be compared with the significant subgraph of the first kind corresponding in query graph structured data Compared with obtaining the significant subgraph similarity of the first kind.So, server can be the history graph structure data and the query graph The product of the significant subgraph similarity of each first kind of structured data, as the history graph structure data and the inquiry graph structure Second similarity magnitude of data.By way of product, it is similar to second that the significant subgraph similarity of each first kind can be improved The influence degree of metric.As long as such as the significant subgraph similarity of any first kind is 0, i.e., the first kind of two graph structure data Significant subgraph is completely uncorrelated, then the second similarity measure is really set to 0 by directly value.
S204, according to the history graph structure data and the non-significant subgraph phase of the first kind in the query graph structured data Like degree and the second class subgraph similarity, the third similarity of the history graph structure data Yu the query graph structured data is determined Magnitude.
Getting the non-significant subgraph similarity of the first kind in history graph structure data and query graph structured data and the When two class subgraph similarities, third similarity measure value can be obtained accordingly.
Since the non-significant subgraph of the first kind and the second class subgraph usually correspond to the specific patient's condition of case history without playing similarity Decisive role, during determining third similarity measure value, for example, can by the history graph structure data with it is described The sum of the non-significant subgraph similarity of each first kind, each second class subgraph similarity in query graph structured data, make For the third similarity measure value of the history graph structure data and the query graph structured data.It is to be understood that third is similar Metric is presented as the summation of the non-significant subgraph similarity of each first kind and each second class subgraph similarity.
Above-mentioned steps S201, step S203, step S204 in the present embodiment are walked not by sequentially limiting described in Fig. 4 Rapid S201, step S203, step S204 can be performed in other orders or simultaneously, herein with no restrictions.
S205, according to first similarity measure of each history graph structure data and the query graph structured data Value, the second similarity magnitude and the third similarity measure value, determine each history graph structure data and the inquiry The similarity degree of graph structure data.
Specifically, it can be and history graph structure data journey similar to query graph structured data determined with following equation one Degree.
Wherein, A=I-M；
The similarity degree of R (d, q) expression history graph structure data d and query graph structured data q；
Sim is similarity operator, Sim (vD, root,vQ, root) it is the first similarity magnitude, vD, rootFor the root of history graph structure data Node, vQ, rootFor the root node of query graph structured data；
For the second similarity magnitude, ui∈M,uj∈ M indicates the significant subgraph of the first kind Leaf node set, thus define the S in the second similarity magnituded,iIt is aobvious for i-th of first kind of history graph structure data Write subgraph, the S in the second similarity magnitudeq,For the significant subgraph of j-th of first kind of query graph structured data；
For third similarity measure value, I is the set of all leaf nodes, ui∈A,uj ∈ A indicates the leaf node in addition to the leaf node set of the significant subgraph of the first kind, thus defines third similarity measure value In Sd,For the non-significant subgraph of i-th of first kind or the second class subgraph of history graph structure data, in third similarity measure value Sq,For the non-significant subgraph of j-th of first kind or the second class subgraph of query graph structured data.
In above-mentioned formula one, root node is, for example, main diagnosis shown in Fig. 2, and M gathers the corresponding affiliated subgraph of leaf node, example As can be the significant subgraph of the first kind belonging to " age " shown in Fig. 2, " gender ", " consultation time " and " diagnosis and treatment type "；A Gather corresponding subgraph for example and can be and is shown in Fig. 2 in addition to " age ", " gender ", " consultation time " and " diagnosis and treatment type " institute State other subgraphs other than subgraph, including the second class subgraph and the non-significant subgraph of the first kind.
S104, according to default selection rule and the similarity degree, in the multiple history medical record data described in determination Inquire the similar case history lookup result of medical record data, wherein the corresponding history graph structure of the similar case history lookup result Data have the similarity degree for meeting the default selection rule.
Such as it can be the number of history medical record data in the similar case history lookup result based on the output of decision-making device model cootrol Amount.Such as the sequence high to Low by similarity degree is carried out to history medical record data with similarity degree, then with preset truncation ratio Example (such as 50%) will meet several history medical record datas of the sequence of truncation ratio to rank first, and searches and ties as similar case history Fruit.It is also possible to according to preset truncation number, number (such as 5) sequence history case history number is truncated in the satisfaction to rank first According to as similar case history lookup result.Default selection rule can be based on number or based on ratio, it is not limited here.
A kind of similar case history lookup method provided in this embodiment, by obtaining inquiry medical record data and multiple history case histories Data；It obtains the corresponding query graph structured data of the inquiry medical record data and each history medical record data is corresponding goes through History graph structure data, wherein the query graph structured data and the history graph structure data all include first kind subgraph and Two class subgraphs, the intermediate node of the first kind subgraph are case history field classification, the intermediate node and leaf of the second class subgraph Child node is to carry out feature to the first kind subgraph to identify；According to each history graph structure data and the inquiry Root node similarity, first kind subgraph similarity and the second class subgraph similarity in graph structure data obtain each history figure The similarity degree of structured data and the query graph structured data；According to default selection rule and the similarity degree, described The similar case history lookup result of the inquiry medical record data is determined in multiple history medical record datas, to extract inquiry case history number Intrinsic subgraph and recognizable obtained subgraph in, to the corresponding subgraph in inquiry medical record data and history medical record data case history The relevance of middle data is measured, and the accuracy that similar case history is searched is improved.
Since the content type of the generation type and embodiment of first kind subgraph and the second class subgraph is different, the embodiment of the present invention It can determine that the similarity of first kind subgraph is similar with the second class subgraph respectively with different two kinds of similarity methods of determination Degree.
For determining the similarity of first kind subgraph, wherein again including determining the significant subgraph similarity of the first kind and the first kind Non-significant subgraph similarity.
In above-described embodiment, it is possible to understand that, in step S203 (according to the history graph structure data and the query graph The significant subgraph similarity of the first kind in structured data determines the history graph structure data and the query graph structured data Second similarity magnitude) and step S204 (according to the in the history graph structure data and the query graph structured data A kind of non-significant subgraph similarity and the second class subgraph similarity, determine the history graph structure data and the inquiry graph structure The third similarity measure value of data) before, it can also include calculating the significant subgraph similarity of the first kind and the non-significant son of the first kind The process of figure similarity.
For example, it may be in the history graph structure data with the query graph structured data with identical intermediate node The first kind subgraph in, the side according to leaf node similarity, root node and leaf node in the first kind subgraph is pre- If weight and root node similarity, first kind subgraph in the history graph structure data and the query graph structured data is obtained Similarity.Wherein, the similarity of the first kind subgraph includes the significant subgraph similarity of the first kind and the first kind Non-significant subgraph similarity, the i.e. calculation method of the similarity of first kind subgraph are exactly the significant subgraph similarity of the first kind and The calculation method of a kind of non-significant subgraph similarity.
In some embodiments, it can be and obtained in history graph structure data and query graph structured data with following formula two The similarity of first kind subgraph.
Sim(Sd,Sq)=Sim (ud,uq)*weight(uv)*Sim(vd,vq) formula two
Wherein, Sim indicates similarity operator；SdIndicate the first kind subgraph of history graph structure data, SqIndicate query graph knot The first kind subgraph of structure data；udIndicate the first kind subgraph S of history graph structure datadLeaf node, uqIndicate query graph knot The first kind subgraph S of structure dataqLeaf node；Weight (uv) indicates the side of root node and leaf node in first kind subgraph Default weight；vdIndicate the first kind subgraph S of history graph structure datadRoot node, vqIndicate the first of query graph structured data Class subgraph SqRoot node.
First kind subgraph in above-mentioned formula two, can be in graph structure data shown in Fig. 2, respectively with age, gender, doctor It advises, the subgraph that physical examination, diagnosis and treatment type, main suit, auxiliary examination, department, present illness history and consultation time are leaf node.Its In, it is the significant subgraph of the first kind using age, gender, diagnosis and treatment type, consultation time as the subgraph of leaf node, remaining is first The non-significant subgraph of class.It presets weight and can be according to the pre-set power of medical expert's experience in the side of root node and leaf node Weight.
It should be understood that in above-mentioned steps S204 (according to the history graph structure data and the query graph structured data In the non-significant subgraph similarity of the first kind and the second class subgraph similarity, determine the history graph structure data and the inquiry The third similarity measure value of graph structure data) before, it can also first obtain in history graph structure data and query graph structured data Second class subgraph similarity.Specifically, it can be that first to obtain in the second class subgraph the side of root node and intermediate node default The side statistical weight of leaf node and intermediate node in weight and the second class subgraph.It should be understood that ground, in history figure knot In structure data, with ud,iIndicate the leaf node of the second class subgraph, cdIndicate intermediate node, vdRoot node is indicated, then the pass on side System are as follows: ud,cd+cdvd=ud,vd.The side c of root node and intermediate nodedvdWeighted value, for belonging to an intermediate node cdNo With leaf node ud,With ud,For, belong to the leaf node of same type, i.e. cdvd=ud,vd-ud,cdWith cdvd=ud,vd-ud,cdIt is It is identical.So the side u of intermediate node and leaf noded,cdThe side u of weight and root node and leaf noded,vdWeight is main track Property is relevant.For example, side ud,ivdThe correlation statistics of the leaf nodes such as diagnosis and symptom A, symptom B, have based on weight expression Medicine interpretation, so according to the correlation statistics of root node and leaf node, to determine leaf node and middle node The side statistical weight of point.Weighted average of the description information based on mutual information Yu chi-square statistics value of e.g. main diagnosis and feature Value.For example, specifically can be at least one leaf node first obtained in the second class subgraph under intermediate node；Obtain institute State the mutual information of root node and at least one leaf node in the second class subgraph.Then root in the second class subgraph is obtained The chi-square statistics value of node and at least one leaf node；It finally will the mutual information corresponding with the intermediate node The weighted sum of value and the chi-square statistics value is counted as the side of leaf node described in the second class subgraph and intermediate node Weight.Then, in the history graph structure data and the query graph structured data with identical intermediate node described the In two class subgraphs, according to the default power in the side of leaf node similarity, the root node and intermediate node in the second class subgraph The side statistical weight and root node similarity of weight, leaf node and intermediate node, obtain the history graph structure data and institute State the second class subgraph similarity in query graph structured data.For example, by the intermediate node of history graph structure data is symptom the Two class subgraphs, the intermediate node with query graph structured data are that the second class subgraph of symptom carries out similarity calculation, obtain centre Node is the second class subgraph similarity of symptom.
For example, it may be obtaining in the history graph structure data and the query graph structured data the with following equation three Two class subgraph similarities.
Wherein, the second class subgraph of history graph structure data is Sd={ ud,1,ud,2,…,ud,m,cd,vd},{ud,1cd,ud, 2cd,…,ud,mcd,cdvd), ud,mFor m-th of leaf node of the second class subgraph of history graph structure data, cdFor history figure knot The intermediate node of second class subgraph of structure data, vdFor the root node of the second class subgraph of history graph structure data, ud,cdFor leaf Child node ud,With intermediate node cdSide, cdvdFor intermediate node cdWith root node vdSide；Second class of query graph structured data Subgraph is Sq={ uq,1,uq,2,…,uq,n,cq,vq},{uq,1cq,uq,2cq,…,uq,ncq,cqvq), uq,nFor query graph structure number According to the second class subgraph n-th of leaf node, cqFor the intermediate node of the second class subgraph of query graph structured data, vqTo look into Ask the root node of the second class subgraph of graph structure data, uq,ncqFor leaf node uq,nWith intermediate node cqSide, cqvqFor centre Node cqWith root node vqSide；
weight(ud,icd) indicate the second class subgraph in leaf node and intermediate node side statistical weight；
weight(cdvd) indicate that weight is preset on the side of root node and intermediate node in the second class subgraph；
OperatorWherein, α, β are constant.
Table one
In the embodiment of the similarity of above-mentioned determining first kind subgraph, obtain the history graph structure data with it is described It can also include the data type according to leaf node in query graph structured data before the similarity of first kind subgraph, and select Select the step of corresponding mode determines each leaf node similarity of first kind subgraph.Specifically, it can be and obtain the history In graph structure data and the query graph structured data, the data type of the leaf node of the first kind subgraph.Referring to table one, It is that the optional similarity of four kinds of data types provided in an embodiment of the present invention determines model.Complex data type shown in table one is corresponding Similarity determine in model, multiply factor property set be node x it is all multiply the factor attribute information composition set, add factor category Property integrate the set constituted as node x all plus factor attribute information, wherein node x, y for carrying out similarity-rough set have one by one It is corresponding to multiply factor attribute information, and add factor attribute information correspondingly.The data of the leaf node of first kind subgraph Type is usually all numeric data type, Boolean data type and text data class in simple data type, such as table one Type.It obtains target similarity corresponding with the data type and determines model, corresponding relationship is for example, see Fig. 1.Then basis The target similarity determines of the first kind described in model and the history graph structure data and the query graph structured data The description information of leaf node in figure obtains the first kind described in the history graph structure data and the query graph structured data Each leaf node similarity of subgraph.Determine model to the description information of leaf node in first kind subgraph with target similarity It is calculated, obtains each leaf node similarity of first kind subgraph.Wherein numeric data type is, for example, age, consultation time Etc. fields description information.Boolean data type is, for example, the description information (such as default male is 1, female 0) of gender field.Text Notebook data type is, for example, the description information of the fields such as main suit, present illness history.By selecting corresponding target similarity to determine model Each leaf node similarity is calculated, the accuracy of leaf node similarity is improved, and then improves final lookup structure Accuracy.
In the embodiment of above-mentioned the second class of determination subgraph similarity, history graph structure data and inquiry graph structure are being obtained It can also include the data type according to leaf node in data before the second class subgraph similarity, and select corresponding mode The step of determining each leaf node similarity of the second class subgraph.Specifically, can be obtain the history graph structure data with In the query graph structured data, the data type of the leaf node of the second class subgraph.The leaf node of second class subgraph Data type may be any of four kinds of data types in table one.Wherein, numeric data type, Boolean data type and Text data type is simple data type, is in addition to this complex data type.
If the data type is simple data type, according to the corresponding institute of the leaf node of simple data type Data type is stated, determines that target similarity determines model.Model and simple data type are determined according to the target similarity The description information of the leaf node obtains the second class described in the history graph structure data and the query graph structured data The leaf node similarity of each simple data type of subgraph.Second class subgraph leaf node similarity meter of simple data type The implementation of calculation, similar with the implementation of first kind subgraph leaf node similarity calculation, this will not be repeated here.
If the data type is complex data type, the leaf node for obtaining complex data type corresponding multiplies Factor attribute information and plus factor attribute information.With " symptom " for middle node in complex data type such as query graph structured data In second class subgraph of point, leaf node includes " cough ", and the factor attribute information that multiplies of " cough " is " 1 " (expression), is added Factor attribute information is, for example, " serious ", " 24 hours ".It is possible to which the leaf node by complex data type is corresponding The product of each similarity for multiplying factor attribute information, it is corresponding with the leaf node of complex data type each described plus The product of the weighted sum of the similarity of factor attribute information, as the history graph structure data and the query graph structured data Described in the second class subgraph each complex data type leaf node similarity.Multiply for example, history graph structure data are corresponding Factor attribute information is " 1 " (expression), and adding factor attribute information is, for example, " slight ", " 1 hour ", then with Boolean data The similarity of type determines that model is calculated the factor (1,1) is multiplied, and determines model to adding with the similarity of text data type Factor attribute information (" test serious ", " slight ") is calculated, and determines model to adding the factor with the similarity of numeric data type Attribute information (" 24 hours ", " 1 hour ") is calculated.The description information correspondence of each leaf node multiplies the factor in second class subgraph Attribute information still adds factor attribute information, can preset corresponding relationship.It is each in detecting the second class subgraph The description information of leaf node, according to note of the description information of each leaf node in the second class subgraph in default corresponding relationship Record, obtains that each leaf node in the second class subgraph is corresponding to multiply factor attribute information, plus factor attribute information.
In the embodiment of above-mentioned similar case history lookup method, the matching and rearrangement of similar case history are realized, and realizes disease It goes through degree of a relation amount between feature to calculate, thus the metric calculation etc. of each characteristic attribute of case history improves what similar case history was searched Accuracy.
It is that a kind of similar case history provided in an embodiment of the present invention searches apparatus structure schematic diagram, as shown in Figure 5 referring to Fig. 5 Similar case history search device 50, comprising:
Case history obtains module 51, for obtaining inquiry medical record data and multiple history medical record datas.
Graph structure module 52, for obtaining the corresponding query graph structured data of the inquiry medical record data and each institute State the corresponding history graph structure data of history medical record data, wherein the query graph structured data and the history graph structure number According to all including first kind subgraph and the second class subgraph, the intermediate node of the first kind subgraph is case history field classification, described the The intermediate node and leaf node of two class subgraphs are to carry out feature to the first kind subgraph to identify.
Processing module 53, for according to each history graph structure data and root node phase in the query graph structured data Like degree, first kind subgraph similarity and the second class subgraph similarity, each history graph structure data and the query graph are obtained The similarity degree of structured data.
Selecting module 54, for selecting rule and the similarity degree according to default, in the multiple history medical record data The similar case history lookup result of the middle determination inquiry medical record data, wherein the similar case history lookup result is corresponding described History graph structure data have the similarity degree for meeting the default selection rule.
The similar case history of embodiment illustrated in fig. 5, which searches device, accordingly can be used for executing in embodiment of the method shown in Fig. 1 taking The step of business device executes, it is similar that the realization principle and technical effect are similar, and details are not described herein again.
Optionally, processing module 53, for according in the history graph structure data and the query graph structured data Root node similarity determines the first similarity magnitude of the history graph structure data Yu the query graph structured data；According to The conspicuousness classification of leaf node determines the significant subgraph of the first kind and the non-significant subgraph of the first kind in the first kind subgraph； According to the significant subgraph similarity of the first kind in the history graph structure data and the query graph structured data, gone through described in determination Second similarity magnitude of history graph structure data and the query graph structured data；According to the history graph structure data with it is described The non-significant subgraph similarity of the first kind and the second class subgraph similarity in query graph structured data, determine the history graph structure The third similarity measure value of data and the query graph structured data；According to each history graph structure data and the query graph The first similarity magnitude, the second similarity magnitude and the third similarity measure value of structured data, determine each institute State the similarity degree of history graph structure data Yu the query graph structured data.
Optionally, processing module 53, described according in the history graph structure data and the query graph structured data The significant subgraph similarity of the first kind, determine the second similarity of the history graph structure data Yu the query graph structured data Magnitude；And it is described according to the history graph structure data and the non-significant subgraph phase of the first kind in the query graph structured data Like degree and the second class subgraph similarity, the third similarity of the history graph structure data Yu the query graph structured data is determined Before magnitude, it is also used to the institute with identical intermediate node in the history graph structure data with the query graph structured data It states in first kind subgraph, according to the default power in the side of leaf node similarity, root node and leaf node in the first kind subgraph Weight and root node similarity, obtain the phase of the history graph structure data with first kind subgraph in the query graph structured data Like degree, wherein the similarity of the first kind subgraph includes that the significant subgraph similarity of the first kind and the first kind are non-aobvious Write subgraph similarity.
Optionally, processing module 53, for by each the of the history graph structure data and the query graph structured data The product of a kind of significant subgraph similarity, it is similar to the second of the query graph structured data as the history graph structure data Metric.
Optionally, processing module 53, described according in the history graph structure data and the query graph structured data The non-significant subgraph similarity of the first kind and the second class subgraph similarity, determine the history graph structure data and the query graph Before the third similarity measure value of structured data, it is also used to obtain the side of leaf node and intermediate node in the second class subgraph Statistical weight；In described second with identical intermediate node of the history graph structure data and the query graph structured data In class subgraph, weight, leaf are preset according to the side of leaf node similarity, root node and intermediate node in the second class subgraph The side statistical weight and root node similarity of node and intermediate node, obtain the history graph structure data and the query graph Second class subgraph similarity in structured data.
Optionally, processing module 53, for obtaining the history graph structure data and the query graph with following equation three Second class subgraph similarity in structured data:
Wherein, the second class subgraph of history graph structure data is Sd={ ud,1,ud,2,…,ud,m,cd,vd},{ud,1cd,ud, 2cd,…,ud,mcd,cdvd), ud,mFor m-th of leaf node of the second class subgraph of history graph structure data, cdFor history figure knot The intermediate node of second class subgraph of structure data, vdFor the root node of the second class subgraph of history graph structure data, ud,mcdFor leaf Child node ud,With intermediate node cdSide, cdvdFor intermediate node cdWith root node vdSide；Second class of query graph structured data Subgraph is Sq={ uq,1,uq,2,…,uq,n,cq,vq},{uq,1cq,uq,2cq,…,uq,ncq,cqvq), uq,nFor query graph structure number According to the second class subgraph n-th of leaf node, cqFor the intermediate node of the second class subgraph of query graph structured data, vqTo look into Ask the root node of the second class subgraph of graph structure data, uq,cqFor leaf node uq,With intermediate node cqSide, cqvqFor middle node Point cqWith root node vqSide；
weight(ud,icd) indicate the second class subgraph in leaf node and intermediate node side statistical weight；
weight(cdvd) indicate that weight is preset on the side of root node and intermediate node in the second class subgraph；
OperatorWherein, α, β are constant.
Optionally, processing module 53, for obtaining at least one leaf in the second class subgraph under intermediate node Node；Obtain the mutual information of root node and at least one leaf node in the second class subgraph；Obtain second class The chi-square statistics value of root node and at least one leaf node in subgraph；It will be corresponding with the intermediate node described mutual The weighted sum of the value of information and the chi-square statistics value, the side as leaf node and intermediate node described in the second class subgraph Statistical weight.
Optionally, processing module 53, for by the history graph structure data with it is each in the query graph structured data The sum of the non-significant subgraph similarity of the first kind, each described second class subgraph similarity, as the history graph structure data With the third similarity measure value of the query graph structured data.
Optionally, processing module 53, in the tool in the history graph structure data and the query graph structured data Have in the first kind subgraph of identical intermediate node, according to leaf node similarity in the first kind subgraph, root node with Weight and root node similarity are preset in the side of leaf node, obtain the history graph structure data and the query graph structure number According to the similarity of middle first kind subgraph, wherein the similarity of the first kind subgraph includes that the significant subgraph of the first kind is similar Before degree and the non-significant subgraph similarity of the first kind, it is also used to obtain the history graph structure data and the query graph knot In structure data, the data type of the leaf node of the first kind subgraph；Obtain target phase corresponding with the data type Model is determined like spending；Model and the history graph structure data and the inquiry graph structure are determined according to the target similarity The description information of leaf node in first kind subgraph described in data obtains the history graph structure data and the query graph knot Each leaf node similarity of first kind subgraph described in structure data.
Optionally, processing module 53, in the tool in the history graph structure data and the query graph structured data Have in the second class subgraph of identical intermediate node, according to leaf node similarity in the second class subgraph, root node with The while statistical weight and root node similarity in default weight, leaf node and intermediate node of intermediate node, described in acquisition In history graph structure data and the query graph structured data before the second class subgraph similarity, it is also used to obtain the history figure In structured data and the query graph structured data, the data type of the leaf node of the second class subgraph；If the data Type is simple data type, then according to the corresponding data type of the leaf node of simple data type, determines mesh Mark similarity determines model；Retouching for the leaf node of model and simple data type is determined according to the target similarity Information is stated, each simple data of the second class subgraph described in the history graph structure data and the query graph structured data is obtained The leaf node similarity of type；If the data type is complex data type, the leaf of complex data type is obtained Child node is corresponding to multiply factor attribute information and plus factor attribute information；The leaf node of complex data type is corresponding The product of each similarity for multiplying factor attribute information, it is corresponding with the leaf node of complex data type each described plus The product of the weighted sum of the similarity of factor attribute information, as the history graph structure data and the query graph structured data Described in the second class subgraph each complex data type leaf node similarity.
Optionally, processing module 53, described according to each history graph structure data and the query graph structured data Middle root node similarity, first kind subgraph similarity and the second class subgraph similarity, obtain each history graph structure data with Before the similarity degree of the query graph structured data, it is also used to determine model and described according to preset text-type similarity The description information of root node determines root node similarity in each history graph structure data and the query graph structured data.
Optionally, graph structure module 52, for using the main diagnosis of the inquiry case history as the description information of root node； The case history field classification for including using the inquiry case history is as first kind intermediate node, and the field for including by the field classification Respectively as the first kind leaf node of the field classification；By the corresponding description information of the field, as the first kind The description information of leaf node；Using preset feature as the second class intermediate node, and with the feature to each field and The description information of each field carries out the identification of natural language understanding NLU, obtains the description information of the feature, or obtain The description information of the feature and the description information plus factor attribute information or multiply factor attribute information；By the feature Second class leaf node of the description information as the feature, wherein there is described plus factor attribute information or multiply factor attribute The data type of second class leaf node of information is complex data type；According to the root node, the first kind middle node Point, the first kind leaf node, the second class intermediate node and the second class leaf node obtain the inquiry disease It counts one by one according to corresponding query graph structured data, wherein the root node, the first kind intermediate node and the first kind leaf Child node forms first kind subgraph；The root node, the second class intermediate node and the second class leaf node are formed Second class subgraph.
Optionally, case history obtains module 51, for obtaining inquiry medical record data；According to the inquiry medical record data, determine Full-text search keyword；Case history full text is carried out in case history library according to the full-text search key to search, is obtained comprising described complete Multiple history medical record datas of literary search key.
It is a kind of hardware structural diagram of equipment provided in an embodiment of the present invention referring to Fig. 6, which includes: place Manage device 61, memory 62 and computer program；Wherein
Memory 62, for storing the computer program, which can also be flash memory (flash).The calculating Machine program is, for example, to realize application program, the functional module etc. of the above method.
Processor 61, for executing the computer program of the memory storage, to realize above-mentioned similar case history lookup side Each step that server executes in method.It specifically may refer to the associated description in previous methods embodiment.
Optionally, memory 62 can also be integrated with processor 61 either independent.
When the memory 62 is independently of the device except processor 61, the equipment can also include:
Bus 63, for connecting the memory 62 and processor 61.The equipment of Fig. 6 can further include transmitter (being not drawn into figure), the similar case history lookup result of the inquiry medical record data for being sent out the acquisition of processor 61.
The present invention also provides a kind of readable storage medium storing program for executing, computer program is stored in the readable storage medium storing program for executing, it is described The similar case history lookup method provided when computer program is executed by processor for realizing above-mentioned various embodiments.
Wherein, readable storage medium storing program for executing can be computer storage medium, be also possible to communication media.Communication media includes just In from a place to any medium of another place transmission computer program.Computer storage medium can be general or special Any usable medium enough accessed with computer capacity.For example, readable storage medium storing program for executing is coupled to processor, to enable a processor to Information is read from the readable storage medium storing program for executing, and information can be written to the readable storage medium storing program for executing.Certainly, readable storage medium storing program for executing can also be with It is the component part of processor.Processor and readable storage medium storing program for executing can be located at specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC) in.In addition, the ASIC can be located in user equipment.Certainly, Processor and readable storage medium storing program for executing can also be used as discrete assembly and be present in communication equipment.Readable storage medium storing program for executing can be read-only Memory (ROM), random access memory (RAM), CD-ROM, tape, floppy disk and optical data storage devices etc..
The present invention also provides a kind of program product, the program product include execute instruction, this execute instruction be stored in it is readable In storage medium.At least one processor of equipment can read this from readable storage medium storing program for executing and execute instruction, at least one processing Device executes this and executes instruction so that equipment implements the similar case history lookup method that above-mentioned various embodiments provide.
In the embodiment of above equipment, it should be appreciated that processor can be central processing unit (English: Central Processing Unit, referred to as: CPU), it can also be other general processors, digital signal processor (English: Digital Signal Processor, referred to as: DSP), specific integrated circuit (English: Application Specific Integrated Circuit, referred to as: ASIC) etc..General processor can be microprocessor or the processor is also possible to any conventional place Manage device etc..It can be embodied directly in hardware processor in conjunction with the step of the method disclosed in the present and execute completion or use Hardware and software module combination in reason device execute completion.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution The range of scheme.

Claims (16)

1. a kind of similar case history lookup method characterized by comprising
Obtain inquiry medical record data and multiple history medical record datas；
Obtain the corresponding query graph structured data of the inquiry medical record data and the corresponding history of each history medical record data Graph structure data, wherein the query graph structured data and the history graph structure data all include first kind subgraph and second Class subgraph, the intermediate node of the first kind subgraph are case history field classification, the intermediate node and leaf of the second class subgraph Node is to carry out feature to the first kind subgraph to identify；
It is similar to root node similarity, first kind subgraph in the query graph structured data according to each history graph structure data Degree and the second class subgraph similarity obtain the similarity degree of each the history graph structure data and the query graph structured data；
According to default selection rule and the similarity degree, the inquiry case history number is determined in the multiple history medical record data According to similar case history lookup result, wherein the corresponding history graph structure data of the similar case history lookup result have full The similarity degree of the foot default selection rule.
2. the method according to claim 1, wherein described look into according to each history graph structure data with described Root node similarity, first kind subgraph similarity and the second class subgraph similarity in graph structure data are ask, each history is obtained The similarity degree of graph structure data and the query graph structured data, comprising:
According to the root node similarity in the history graph structure data and the query graph structured data, the history figure is determined First similarity magnitude of structured data and the query graph structured data；
According to the conspicuousness classification of leaf node, in the first kind subgraph, determine that the significant subgraph of the first kind and the first kind are non- Significant subgraph；
According to the significant subgraph similarity of the first kind in the history graph structure data and the query graph structured data, institute is determined State the second similarity magnitude of history graph structure data Yu the query graph structured data；
According to the non-significant subgraph similarity of the first kind and the in the history graph structure data and the query graph structured data Two class subgraph similarities determine the third similarity measure value of the history graph structure data Yu the query graph structured data；
According to the first similarity magnitude of each history graph structure data and the query graph structured data, described second Similarity magnitude and the third similarity measure value determine each history graph structure data and the query graph structured data Similarity degree.
3. according to the method described in claim 2, it is characterized in that, being looked into according to the history graph structure data with described described The significant subgraph similarity of the first kind in graph structure data is ask, determines the history graph structure data and the query graph structure number According to the second similarity magnitude；And
The non-significant subgraph similarity of the first kind according in the history graph structure data and the query graph structured data With the second class subgraph similarity, the third similarity measure value of the history graph structure data Yu the query graph structured data is determined Before, further includes:
In the first kind with identical intermediate node of the history graph structure data and the query graph structured data In figure, weight and root section are preset according to the side of leaf node similarity, root node and leaf node in the first kind subgraph Point similarity, obtains the similarity of first kind subgraph in the history graph structure data and the query graph structured data, wherein The similarity of the first kind subgraph includes that the significant subgraph similarity of the first kind is similar with the non-significant subgraph of the first kind Degree.
4. according to the method in claim 2 or 3, which is characterized in that it is described according to the history graph structure data with it is described The significant subgraph similarity of the first kind in query graph structured data, determines the history graph structure data and the inquiry graph structure Second similarity magnitude of data, comprising:
By the product of the history graph structure data and the significant subgraph similarity of each first kind of the query graph structured data, make For the second similarity magnitude of the history graph structure data and the query graph structured data.
5. according to the method described in claim 2, it is characterized in that, being looked into according to the history graph structure data with described described The non-significant subgraph similarity of the first kind and the second class subgraph similarity in graph structure data are ask, determines the history graph structure number Before the third similarity measure value with the query graph structured data, further includes:
Obtain the side statistical weight of leaf node and intermediate node in the second class subgraph；
In second class with identical intermediate node of the history graph structure data and the query graph structured data In figure, weight, leaf node are preset according to the side of leaf node similarity, root node and intermediate node in the second class subgraph With the side statistical weight and root node similarity of intermediate node, the history graph structure data and the inquiry graph structure are obtained Second class subgraph similarity in data.
6. according to the method described in claim 5, it is characterized in that, according to leaf node similarity in the second class subgraph, described The while statistical weight and root node similarity in default weight, leaf node and intermediate node of root node and intermediate node, Obtain the second class subgraph similarity in the history graph structure data and the query graph structured data, comprising:
The second class subgraph similarity in the history graph structure data and the query graph structured data is obtained with following equation:
Wherein, the second class subgraph of history graph structure data is Sd=({ uD, 1, uD, 2..., uD, m, cd, vd, { uD, 1cd, uD, 2cd..., uD, mcd, cdvd), uD, mFor m-th of leaf node of the second class subgraph of history graph structure data, cdFor history figure The intermediate node of second class subgraph of structured data, vdFor the root node of the second class subgraph of history graph structure data, uD, mcdFor Leaf node uD, mWith intermediate node cdSide, cdvdFor intermediate node cdWith root node vdSide；The second of query graph structured data Class subgraph is Sq=({ uQ, 1, uQ, 2..., uQ, n, cq, vq, { uQ, 1cq, uQ, 2cq..., uQ, ncq, cqvq), uQ, nFor query graph knot N-th of leaf node of the second class subgraph of structure data, cqFor the intermediate node of the second class subgraph of query graph structured data, vq For the root node of the second class subgraph of query graph structured data, uQ, ncqFor leaf node uQ, nWith intermediate node cqSide, cqvqFor Intermediate node cqWith root node vqSide；
weight(uD, icd) indicate the second class subgraph in leaf node and intermediate node side statistical weight；
weight(cdvd) indicate that weight is preset on the side of root node and intermediate node in the second class subgraph；
OperatorWherein, α, β are constant.
7. according to the method described in claim 5, it is characterized in that, leaf node is in acquisition the second class subgraph The side statistical weight of intermediate node, comprising:
Obtain at least one leaf node in the second class subgraph under intermediate node；
Obtain the mutual information of root node and at least one leaf node in the second class subgraph；
Obtain the chi-square statistics value of root node and at least one leaf node in the second class subgraph；
By the weighted sum of the association relationship corresponding with the intermediate node and the chi-square statistics value, as described second The side statistical weight of leaf node described in class subgraph and intermediate node.
8. according to method described in claim 2,3,5,6 or 7, which is characterized in that described according to the history graph structure data The non-significant subgraph similarity of the first kind and the second class subgraph similarity with the query graph structured data, determine the history The third similarity measure value of graph structure data and the query graph structured data, comprising:
By in the history graph structure data and the query graph structured data the non-significant subgraph similarity of each first kind, The sum of each described second class subgraph similarity, the third phase as the history graph structure data and the query graph structured data Likelihood metric value.
9. according to the method described in claim 3, it is characterized in that, described in the history graph structure data and the query graph It is similar according to leaf node in the first kind subgraph in the first kind subgraph with identical intermediate node of structured data Degree, root node and leaf node side preset weight and root node similarity, obtain the history graph structure data with it is described The similarity of first kind subgraph in query graph structured data, wherein the similarity of the first kind subgraph includes the first kind Before significant subgraph similarity and the non-significant subgraph similarity of the first kind, further includes:
It obtains in the history graph structure data and the query graph structured data, the number of the leaf node of the first kind subgraph According to type；
It obtains target similarity corresponding with the data type and determines model；
It is determined described in model and the history graph structure data and the query graph structured data according to the target similarity The description information of leaf node in first kind subgraph obtains institute in the history graph structure data and the query graph structured data State each leaf node similarity of first kind subgraph.
10. according to the method described in claim 5, it is characterized in that, described in the history graph structure data and the inquiry In the second class subgraph with identical intermediate node of graph structure data, according to leaf node phase in the second class subgraph Like the while statistical weight and root section in default weight, leaf node and intermediate node of degree, the root node and intermediate node Point similarity obtains in the history graph structure data and the query graph structured data before the second class subgraph similarity, also Include:
It obtains in the history graph structure data and the query graph structured data, the number of the leaf node of the second class subgraph According to type；
If the data type is simple data type, according to the corresponding number of the leaf node of simple data type According to type, determine that target similarity determines model；The described of model and simple data type is determined according to the target similarity The description information of leaf node obtains the second class subgraph described in the history graph structure data and the query graph structured data Each simple data type leaf node similarity；
If the data type is complex data type, the leaf node for obtaining complex data type corresponding multiplies the factor Attribute information and plus factor attribute information；The corresponding each factor attribute that multiplies of the leaf node of complex data type is believed The product of the similarity of breath, it is corresponding with the leaf node of complex data type each described plus factor attribute information similar The product of the weighted sum of degree, as the second class subgraph described in the history graph structure data and the query graph structured data The leaf node similarity of each complex data type.
11. according to claim 1, method described in 9 or 10, which is characterized in that described according to each history graph structure data With root node similarity, first kind subgraph similarity and the second class subgraph similarity in the query graph structured data, obtain each Before the similarity degree of the history graph structure data and the query graph structured data, further includes:
The description information that model and the root node are determined according to preset text-type similarity determines each history figure knot Root node similarity in structure data and the query graph structured data.
12. according to claim 1, method described in 9 or 10, which is characterized in that the acquisition inquiry medical record data is corresponding Query graph structured data, wherein the query graph structured data and the history graph structure data all include first kind subgraph With the second class subgraph, the intermediate node of the first kind subgraph is case history field classification, the intermediate node of the second class subgraph It is to carry out feature to the first kind subgraph to identify with leaf node, comprising:
Using the main diagnosis of the inquiry case history as the description information of root node；
Include as first kind intermediate node, and by the field classification using the case history field classification that the inquiry case history includes First kind leaf node of the field respectively as the field classification；By the corresponding description information of the field, as described The description information of a kind of leaf node；
Using preset feature as the second class intermediate node, and the description with the feature to each field and each field Information carries out the identification of natural language understanding NLU, obtains the description information of the feature, or obtains the description letter of the feature Breath and the description information plus factor attribute information or multiply factor attribute information；Using the description information of the feature as described in Second class leaf node of feature, wherein there is described plus factor attribute information or multiply the second class leaf of factor attribute information The data type of node is complex data type；
According to the root node, the first kind intermediate node, the first kind leaf node, the second class intermediate node with And the second class leaf node, obtain the corresponding query graph structured data of the inquiry medical record data, wherein described section Point, the first kind intermediate node and the first kind leaf node form first kind subgraph；The root node, described second Class intermediate node and the second class leaf node form the second class subgraph.
13. the method according to claim 1, wherein acquisition inquiry medical record data and multiple history case histories Data, comprising:
Obtain inquiry medical record data；
According to the inquiry medical record data, full-text search keyword is determined；
Case history full text is carried out in case history library according to the full-text search key to search, is obtained comprising the full-text search keyword Multiple history medical record datas.
14. a kind of similar case history searches device characterized by comprising
Case history obtains module, for obtaining inquiry medical record data and multiple history medical record datas；
Graph structure module, for obtaining the corresponding query graph structured data of the inquiry medical record data and each history The corresponding history graph structure data of medical record data, wherein the query graph structured data and the history graph structure data are all wrapped First kind subgraph and the second class subgraph are included, the intermediate node of the first kind subgraph is case history field classification, the second class The intermediate node and leaf node of figure are to carry out feature to the first kind subgraph to identify；
Processing module, for according to root node similarity in each history graph structure data and the query graph structured data, First kind subgraph similarity and the second class subgraph similarity obtain each history graph structure data and the query graph structure number According to similarity degree；
Selecting module, for being determined in the multiple history medical record data according to default selection rule and the similarity degree The similar case history lookup result of the inquiry medical record data, wherein the corresponding history figure of the similar case history lookup result Structured data has the similarity degree for meeting the default selection rule.
15. a kind of equipment characterized by comprising memory, processor and computer program, the computer program are deposited In the memory, the processor runs the computer program perform claim and requires 1~13 any similar disease for storage Go through lookup method.
16. a kind of readable storage medium storing program for executing, which is characterized in that be stored with computer program, the meter in the readable storage medium storing program for executing 1~13 any similar case history is required when calculation machine program is executed by processor for realizing the computer program perform claim Lookup method.
CN201910557217.5A 2019-06-25 2019-06-25 Similar medical record searching method, device and equipment and readable storage medium Active CN110299209B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910557217.5A CN110299209B (en) 2019-06-25 2019-06-25 Similar medical record searching method, device and equipment and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910557217.5A CN110299209B (en) 2019-06-25 2019-06-25 Similar medical record searching method, device and equipment and readable storage medium

Publications (2)

Publication Number Publication Date
CN110299209A true CN110299209A (en) 2019-10-01
CN110299209B CN110299209B (en) 2022-05-20

Family

ID=68028806

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910557217.5A Active CN110299209B (en) 2019-06-25 2019-06-25 Similar medical record searching method, device and equipment and readable storage medium

Country Status (1)

Country Link
CN (1) CN110299209B (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111370086A (en) * 2020-02-27 2020-07-03 平安国际智慧城市科技股份有限公司 Electronic case detection method, electronic case detection device, computer equipment and storage medium
CN111599427A (en) * 2020-05-14 2020-08-28 郑州大学第一附属医院 Recommendation method and device for unified diagnosis, electronic equipment and storage medium
CN111613339A (en) * 2020-05-15 2020-09-01 山东大学 Similar medical record searching method and system based on deep learning
CN113010746A (en) * 2021-03-19 2021-06-22 厦门大学 Medical record sequence retrieval method and system based on subtree inverted index
WO2021121129A1 (en) * 2020-06-30 2021-06-24 平安科技（深圳）有限公司 Method and apparatus for similar case detection, device, and storage medium
WO2021197491A1 (en) * 2020-04-03 2021-10-07 清华大学 Network representation learning algorithm across medical data sources
CN113590777A (en) * 2021-06-30 2021-11-02 北京百度网讯科技有限公司 Text information processing method and device, electronic equipment and storage medium

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120265788A1 (en) * 2001-11-30 2012-10-18 Intelligent Medical Objects, Inc. Longitudinal Electronic Record System and Method With Problem Determination and Plan Ordering
US20130096944A1 (en) * 2011-10-13 2013-04-18 The Board of Trustees of the Leland Stanford, Junior, University Method and System for Ontology Based Analytics
JP2017174406A (en) * 2016-03-24 2017-09-28 富士通株式会社 Healthcare risk estimation system and method
CN108388580A (en) * 2018-01-24 2018-08-10 平安医疗健康管理股份有限公司 Merge the dynamic knowledge collection of illustrative plates update method of medical knowledge and application case
CN109215754A (en) * 2018-09-10 2019-01-15 平安科技（深圳）有限公司 Medical record data processing method, device, computer equipment and storage medium
CN109785968A (en) * 2018-12-27 2019-05-21 东软集团股份有限公司 A kind of event prediction method, apparatus, equipment and program product

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120265788A1 (en) * 2001-11-30 2012-10-18 Intelligent Medical Objects, Inc. Longitudinal Electronic Record System and Method With Problem Determination and Plan Ordering
US20130096944A1 (en) * 2011-10-13 2013-04-18 The Board of Trustees of the Leland Stanford, Junior, University Method and System for Ontology Based Analytics
JP2017174406A (en) * 2016-03-24 2017-09-28 富士通株式会社 Healthcare risk estimation system and method
CN108388580A (en) * 2018-01-24 2018-08-10 平安医疗健康管理股份有限公司 Merge the dynamic knowledge collection of illustrative plates update method of medical knowledge and application case
CN109215754A (en) * 2018-09-10 2019-01-15 平安科技（深圳）有限公司 Medical record data processing method, device, computer equipment and storage medium
CN109785968A (en) * 2018-12-27 2019-05-21 东软集团股份有限公司 A kind of event prediction method, apparatus, equipment and program product

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王超: "基于用户意图分析的电子病历检索技术研究", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》 *

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111370086A (en) * 2020-02-27 2020-07-03 平安国际智慧城市科技股份有限公司 Electronic case detection method, electronic case detection device, computer equipment and storage medium
WO2021169099A1 (en) * 2020-02-27 2021-09-02 平安国际智慧城市科技股份有限公司 Electronic patient record detection method and apparatus, computer device and storage medium
WO2021197491A1 (en) * 2020-04-03 2021-10-07 清华大学 Network representation learning algorithm across medical data sources
CN111599427A (en) * 2020-05-14 2020-08-28 郑州大学第一附属医院 Recommendation method and device for unified diagnosis, electronic equipment and storage medium
CN111599427B (en) * 2020-05-14 2023-03-31 郑州大学第一附属医院 Recommendation method and device for unified diagnosis, electronic equipment and storage medium
CN111613339A (en) * 2020-05-15 2020-09-01 山东大学 Similar medical record searching method and system based on deep learning
CN111613339B (en) * 2020-05-15 2021-07-09 山东大学 Similar medical record searching method and system based on deep learning
WO2021121129A1 (en) * 2020-06-30 2021-06-24 平安科技（深圳）有限公司 Method and apparatus for similar case detection, device, and storage medium
CN113010746A (en) * 2021-03-19 2021-06-22 厦门大学 Medical record sequence retrieval method and system based on subtree inverted index
CN113010746B (en) * 2021-03-19 2023-08-29 厦门大学 Medical record graph sequence retrieval method and system based on sub-tree inverted index
CN113590777A (en) * 2021-06-30 2021-11-02 北京百度网讯科技有限公司 Text information processing method and device, electronic equipment and storage medium

Also Published As

Publication number Publication date
CN110299209B (en) 2022-05-20

Similar Documents

Publication Publication Date Title
CN110299209A (en) Similar case history lookup method, device, equipment and readable storage medium storing program for executing
Altaf et al. Applications of association rule mining in health informatics: a survey
Du et al. Neural attention with character embeddings for hay fever detection from twitter
US20210233658A1 (en) Identifying Relevant Medical Data for Facilitating Accurate Medical Diagnosis
US20150178386A1 (en) System and Method for Extracting Measurement-Entity Relations
WO2009037615A1 (en) System and method for analyzing electronic data records
US8812506B2 (en) System and method for conducting processor-assisted indexing and searching
US20070282940A1 (en) Thread-ranking apparatus and method
CN112635011A (en) Disease diagnosis method, disease diagnosis system, and readable storage medium
CN109155152B (en) Clinical report retrieval and/or comparison
KR20170133692A (en) Method and Apparatus for generating association rules between medical words in medical record document
JP6567484B2 (en) Estimated model construction system, estimated model construction method and program
Pereira et al. Text mining applied to electronic medical records: a literature review
Ghosh et al. Uncovering hidden semantics of set information in knowledge bases
Nelson et al. Deep forecasting of translational impact in medical research
Sun et al. Interpretable time-aware and co-occurrence-aware network for medical prediction
Carvalho et al. Knowledge Graph Embeddings for ICU readmission prediction
Pokharel et al. Representing EHRs with temporal tree and sequential pattern mining for similarity computing
JP2021536636A (en) How to classify medical records
Wegrzyn-Wolska et al. Social media analysis for e-health and medical purposes
Memarzadeh et al. A study into patient similarity through representation learning from medical records
CN109997201A (en) For the accurate clinical decision support using data-driven method of plurality of medical knowledge module
Kenei et al. Visualizing semantic structure of a clinical text document
CN110033862B (en) Traditional Chinese medicine quantitative diagnosis system based on weighted directed graph and storage medium
CN110033835B (en) Device for patient record identification

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant