CN108461110A - Medical information processing method, device and equipment - Google Patents
Medical information processing method, device and equipment Download PDFInfo
- Publication number
- CN108461110A CN108461110A CN201710093245.7A CN201710093245A CN108461110A CN 108461110 A CN108461110 A CN 108461110A CN 201710093245 A CN201710093245 A CN 201710093245A CN 108461110 A CN108461110 A CN 108461110A
- Authority
- CN
- China
- Prior art keywords
- information
- symptom
- description information
- time
- disease
- 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
Links
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
A kind of medical information processing method of the application offer, device and equipment, the method includes:In advance from least a electronic health record, obtains disease type information, symptom description information and the timing node of symptom occur, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease;Result is obtained to more parts of electronic health records of each disease type to integrate, obtain the symptom description information of each disease type information and the correspondence model of timing node in advance;After obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance, the correspondence to match with the targeted exposure symptoms information and its timing node is searched, and diseases analysis is carried out according to lookup result.Diseases analysis efficiency and accuracy rate can be improved using application scheme.
Description
Technical field
This application involves field of computer technology more particularly to medical information processing method, device and equipment.
Background technology
In medical industry, doctor mainly carries out the diseases analysis such as medical diagnosis on disease, disease prediction of the development trend by experience.Doctor
Raw experience carries subjectivity, it is not easy to quantify, and the long-time clinical practice of doctor, exchange is needed to summarize and could obtain
It arrives.As it can be seen that due to needing to carry out diseases analysis by artificial experience, so causing diseases analysis efficiency low, due to the warp of doctor
Test that obtain difficulty with subjectivity and experience big, so the accuracy rate of diseases analysis can be caused low.
Invention content
The application provides medical information processing method, device and equipment, low, accurate to solve prior art diseases analysis efficiency
The low problem of true rate.
According to the embodiment of the present application in a first aspect, provide a kind of medical information processing method, the method includes:
In advance from least a electronic health record, obtains disease type information, symptom description information and symptom occur
Timing node, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease;
Result is obtained to more parts of electronic health records of each disease type to integrate, obtain each disease type information in advance
Symptom description information and timing node correspondence model;
After obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance, search and the mesh
The correspondence that mark symptom information and its timing node match, and diseases analysis is carried out according to lookup result.
According to the second aspect of the embodiment of the present application, a kind of medical information processing device is provided, described device includes:
Model construction module, for from least a electronic health record, obtaining disease type information, symptom description letter in advance
Cease and occur the timing node of symptom, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease;To each
More parts of electronic health records of disease type obtain result and are integrated, obtain each disease type information symptom description information and when
The correspondence model of intermediate node;
Information analysis module, after obtaining targeted exposure symptoms information and its timing node, in the correspondence built in advance
In model, the correspondence to match with the targeted exposure symptoms information and its timing node is searched, and disease is carried out according to lookup result
Disease analysis.
According to the third aspect of the embodiment of the present application, a kind of electronic equipment is provided, including:
Processor;Memory for storing the processor-executable instruction;
Wherein, the processor is configured as:
After obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance, search and the mesh
The correspondence that mark symptom information and its timing node match, and diseases analysis is carried out according to lookup result;
The building process of the correspondence model includes:
From at least a electronic health record, obtains disease type information, symptom description information and the time of symptom occur
Node, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease;
Result is obtained to more parts of electronic health records of each disease type to integrate, and obtains the disease of each disease type information
The correspondence model of shape description information and timing node.
When using the embodiment of the present application medical information processing method, device and equipment, natural language processing skill can be passed through
Art obtains disease type information, symptom description information and the timing node of symptom occurs from least a electronic health record,
To build the symptom description information of each disease type information and the correspondence model of timing node, and obtaining target disease
After shape information and its timing node, in the correspondence model built in advance, search and the targeted exposure symptoms information and its time
The correspondence that node matches, and diseases analysis is carried out according to lookup result.Due to being directed to each disease all by numerous diseases
The symptom that the electronic health record of people, numerous timing nodes and timing node occur, which is used as, refers to foundation, and data are comprehensive, to improve
Precision of analysis, and due to automated analysis, analysis efficiency can be improved.
It should be understood that above general description and following detailed description is only exemplary and explanatory, not
The application can be limited.
Description of the drawings
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the application
Example, and the principle together with specification for explaining the application.
Fig. 1 is one embodiment flow chart of the application medical information processing method.
Fig. 2 is the normalized one embodiment flow chart of symptom in the application medical information processing method.
Fig. 3 is one embodiment flow chart that sequelae model is built in the application medical information processing method.
Fig. 4 is another embodiment flow chart of the application medical information processing method.
Fig. 5 is a kind of hardware structure diagram of electronic equipment where the application medical information processing device.
Fig. 6 is one embodiment block diagram of the application medical information processing device.
Specific implementation mode
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all embodiments consistent with the application.On the contrary, they be only with it is such as appended
The example of consistent device and method of some aspects be described in detail in claims, the application.
It is the purpose only merely for description specific embodiment in term used in this application, is not intended to be limiting the application.
It is also intended to including majority in the application and "an" of singulative used in the attached claims, " described " and "the"
Form, unless context clearly shows that other meanings.It is also understood that term "and/or" used herein refers to and wraps
Containing one or more associated list items purposes, any or all may be combined.
It will be appreciated that though various information, but this may be described using term first, second, third, etc. in the application
A little information should not necessarily be limited by these terms.These terms are only used for same type of information being distinguished from each other out.For example, not departing from
In the case of the application range, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as
One information.Depending on context, word as used in this " if " can be construed to " ... when " or " when ...
When " or " in response to determination ".
Currently, in medical industry, doctor mainly carries out the diseases such as medical diagnosis on disease, disease prediction of the development trend by artificial experience
Disease analysis.Different doctors are since practice experience, the difference of learning ability cause the experience obtained different, the not abundant doctor of experience
Diseases analysis is often carried out according to the symptom that respective time occurs, data are unilateral and inaccurate.Meanwhile disease is carried out by experience
Disease analysis, causes diseases analysis efficiency low.
In order to avoid the defect that diseases analysis accuracy rate is low, efficiency is low, the application provides a kind of medical information processing method,
This method can be divided into the structure stage of correspondence model and the application stage using correspondence model.In an example
In, structure stage and application stage can be executed by the same electronic equipment.In another example, since the structure stage needs
Big data analysis is carried out with the equipment that higher position manages ability, and the application stage requires relatively not the processing capacity of electronic equipment
Height, and after the success of correspondence model construction, the model can be shared between distinct electronic apparatuses, avoids each electronic equipment
The wasting of resources caused by model construction is all carried out, it therefore, can be by one or a set of electronic equipment for managing ability with higher position
Correspondence model is built, other electronic equipments can directly use the correspondence model built.
As shown in FIG. 1, FIG. 1 is one embodiment flow chart of the application medical information processing method, this method can wrap
Following steps 101 are included to step 103, step 101 and step 102 are the prebuild stages of correspondence model, and step 103 is to answer
The stage of diseases analysis is carried out with correspondence model.
In a step 101, in advance from least a electronic health record, obtain disease type information, symptom description information with
And there is the timing node of symptom, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease.
In a step 102, result is obtained to more parts of electronic health records of each disease type to integrate, obtain each in advance
The symptom description information of disease type information and the correspondence model of timing node.
In step 103, after obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance
In, the correspondence to match with the targeted exposure symptoms information and its timing node is searched, and disease point is carried out according to lookup result
Analysis.
In electronic health record often record have the symptom of patient, the medical treatment letter such as the time of the symptom and diagnostic result occur
Breath.The application carries out big data analysis to the medical information in numerous electronic health records, obtains disease and goes out under different time node
Existing symptom (i.e. the evolution of disease), can be configured to each by the symptom that each disease occurs under different time node
The symptom description information of disease type and the correspondence model of timing node, and then carry out disease point using correspondence model
Analysis.
As seen from the above-described embodiment, can be by natural language processing technique, state of an illness, medical history in electronic health record etc. are non-
In structured text, obtains disease type information, symptom description information and the timing node of symptom occur, to build each
The symptom description information of disease type information and the correspondence model of timing node, and obtain targeted exposure symptoms information and its when
After intermediate node, in the correspondence model built in advance, lookup matches with the targeted exposure symptoms information and its timing node
Correspondence, and diseases analysis is carried out according to lookup result.Due to for each disease all by the electronic health record of numerous patients,
The symptom that numerous timing nodes and timing node occur, which is used as, refers to foundation, and data are comprehensive, to improve the standard of analysis result
True property, and due to automated analysis, analysis efficiency can be improved.
For disease type information, disease type information can be intended to indicate that the identification information of disease type, such as disease
The title of sick type.The division of disease type divides according to demand, can carry out thick division, can also carry out refinement point.For example,
Disease type information may include flu, arrhythmia cordis, coronary heart disease, cerebral hemorrhage, leukaemia, diabetes etc..It is more accurate in order to obtain
Disease type can also be finely divided by true disease type.For example, flu is divided into wind-cold type flu and anemopyretic cold.
In every part of electronic health record, often record has disease type information in diagnostic result, therefore can directly be taken out from diagnostic result
Take disease type information.
For symptom description information, symptom description information is to describe the information of symptom, each disease be likely to occur it is a kind of or
A variety of symptoms.For example, being directed to wind-cold type flu, patient often will appear the symptoms such as watery nasal discharge, sneezing, the thin white, chilly of tongue.Needle
To anemopyretic cold, often there are the symptoms such as perspiration, abscess of throat, mouth parched and tongue scorched, the yellow, nasal obstruction of phlegm in patient.
In one example, if there is the semiograhy area for individually recording symptom description information in electronic health record
Domain, what it is due to the semiograhy regional record is symptom description information, then can directly extract disease from the semiograhy region
Shape description information, to obtain the symptom of the case.
In one example, if there is no the semiograhy areas for individually recording symptom description information in electronic health record
Disease is identified in the multi information that then needs to comform in domain, but symptom description information and other information are recorded in case history simultaneously
Shape description information.
In order to identify that symptom description information, the application can describe the symptom to prestore in pattern in multi information of comforming
Character is matched at least a electronic health record, and the symptom describes pattern, and to include symptom description information context can go out
The position relationship of existing character and symptom description information and character;It is closed according to the position of the symptom description information and character
System, obtains symptom description information from the context of match information.
Wherein, certain normal modes are often used when describing symptom, such as:" symptom occur ... " " goes out
It is existing ... ", " with ... phenomenon " etc., these patterns can be known as symptom and describe pattern.Symptom, which describes record in pattern, to be had
The character that symptom description information context will appear.Such as " appearance ", " symptom ", " with ", " phenomenon " etc., while further including disease
The position relationship of shape description information and character, so as to according to the position of the location determination symptom description information of character.For example, " going out
In existing ... symptom " pattern, the position that symptom description information occurs for another example, " goes out between character " appearance " and character " symptom "
Now ... " in pattern, the position that symptom description information occurs is after character " appearance ".Therefore, mould is described into the symptom to prestore
When character in formula is matched in electronic health record, it may be determined that position of the character in electronic health record, then according to symptom
Position estimating of the position relationship and character of description information and character in electronic health record goes out symptom description letter in electronic health record
The position of breath, so as to obtain symptom description information from the context of match information.
Pattern is described about symptom, can be acquired by way of being manually entered, pattern learning can also be passed through
Mode learns to obtain from big data.
In one example, the determination step that the symptom describes pattern includes:
Using known symptom description information as seed, and kind is extracted from least a electronic health record using matching algorithm
Son;
Based on extracted seed, character is extracted from the context of seed in electronic health record and identify seed with
The position relationship of character;
According to the frequency of occurrences of the character extracted and the position relationship identified, determine that symptom describes pattern.
It is known that symptom description information may include the symptom description information being manually entered, can also be including the use of
The symptom description information that application scheme determines.For example, after determining that symptom describes pattern, schema extraction disease is described using symptom
Shape description information, then the symptom description information extracted can be used as known symptom description information in next round pattern drill.
The present embodiment can be using known symptom description information as seed (sample), and utilizes matching algorithm from least one
Seed is extracted in part electronic health record.The purpose of extraction seed is the position in order to determine seed in electronic health record, to be based on
Extracted seed extracts character from the context of seed in electronic health record and identifies that the position of seed and character is closed
System, and determine whether character and position relationship describing mould as symptom according to the frequency of occurrences of character and position relationship
Formula.
Wherein, the present embodiment matches seed using matching algorithm from electronic health record.In one example, matching algorithm
It can be preceding to maximum matching method, extract seed to maximum matching method using preceding, the accuracy rate of extraction seed can be improved.
About the frequency of occurrences, due to that can extract various characters and identify a variety of position relationships, rather than all carry
The position relationship of the character and identification that go out can constitute symptom and describe pattern, in all information extracted and identified, be carried
The character taken and the position relationship repetitive rate identified are higher, are more likely to be the normal mode used when description symptom,
Therefore, in all information extracted and identified, the frequency of occurrences of extracted character and the position relationship identified is calculated,
So that it is determined that symptom describes pattern.
For example, after extracting seed, character string near extracted seed and character string and seed can be enumerated
Position relationship then symptom is described to the occurrence number of preliminary mode and all diseases so that it is determined that symptom describes preliminary mode
The ratio that shape describes the number of preliminary mode describes the frequency of occurrences of preliminary mode as the symptom, and using the frequency of occurrences as referring to
Mark is given a mark, and pattern is described so that the high symptom of score is described preliminary mode and is determined as symptom.
After obtaining symptom and describing pattern, in one example, if retouched using unified describing mode in electronic health record
Symptom is stated, then the symptom to prestore is described into the character in pattern, is matched at least a electronic health record, according to the disease
The position relationship of shape description information and character extracts symptom description information directly from the context of match information.
The method describes pattern using symptom and extracts symptom description information directly from electronic health record, improves acquisition symptom and retouches
State the efficiency of information.
In another example, it is described since different form may be used in same symptom, for convenience subsequently to every
More parts of electronic health records of kind disease type obtain result and are integrated, and need to be standardized symptom, use the mode of cluster
Carry out the normalizing of symptom.Specifically, the position relationship according to the symptom description information and character, from the upper of match information
Symptom description information is hereinafter obtained, including:According to the position relationship of the symptom description information and character, from match information
Symptom original description information is extracted in context;The symptom original description information of same symptom is normalized to identical symptom to retouch
State information.
The validation of information extracted from match information context is symptom original description information by the present embodiment, and will be same
The symptom original description information of one symptom is normalized to identical symptom description information, so as to improve follow-up integral data
Efficiency.On the one hand, symptom original description information can be normalized to standard disease after obtaining symptom original description information every time
Shape description information.It on the other hand, can also be after obtaining all symptom original description information, by same symptoms original description information
It is normalized to classical symptom description information.
For symptom normalization operation, in one example, classical symptom description information and the standard can be preset with
The symptom original description information bank that symptom description information is likely to occur is retouched symptom original description information and the symptom of acquisition are original
It states information in information bank to be matched, when matching degree reaches preset requirement, which is normalized to symptom
The corresponding classical symptom description information of original description information bank.
In another example, the application also provides a kind of normalized method of symptom, as shown in Fig. 2, Fig. 2 is the application
The normalized one embodiment flow chart of symptom in medical information processing method returns the symptom original description information of same symptom
One to turn to identical symptom description information include step 201 to step 204:
In step 201, for partial symptoms original description information in extraction information, same time in same disease is gone out
Existing each symptom original description information is divided into different clustering clusters.
Wherein, extraction information refers to the symptom original description information extracted from the context of match information, by its middle part
The basis for dividing symptom original description information to be used as partition clustering cluster.The symptom described under same time due to same disease is often
Difference, therefore each symptom original description information that same time in same disease occurs can be divided into different clustering clusters,
To realize Preliminary division clustering cluster, the efficiency and accuracy of clustering cluster division can be improved.
In step 202, calculate current symptomatic original description information in clustering cluster symptom original description information it is similar
Degree, and determine whether the current symptomatic original description information clustering cluster or newly-built clustering cluster is added and will work as according to similarity
Newly-built clustering cluster is added in preceding symptom original description information.
Wherein, the current symptomatic original description information is to extract the symptom original description for not having that clustering cluster is added in information
Information.
The step is determined by the similarity of symptom original description information in current symptomatic original description information and clustering cluster
Whether the clustering cluster is added in current symptomatic original description information, if similarity is higher than predetermined threshold value, by current symptomatic original
The clustering cluster is added in beginning description information, if similarity is less than predetermined threshold value, creates clustering cluster, and retouch current symptomatic is original
It states information and newly-built clustering cluster is added.
In one example, the similarity between symptom original description information may be used identical characters quantity and account for shorter word
The ratio of string length is accorded with to indicate.
As it can be seen that no symptom original description information that clustering cluster is added is added this step has clustering cluster or newly-built cluster
Symptom original description information is added in clustering cluster by cluster, realization.
In step 203, after corresponding clustering cluster is added in all symptom original description information, according to different clustering clusters
Between symptom original description information highest similarity, judge whether to merge clustering cluster, and execute corresponding processing.
After corresponding clustering cluster is added in all symptom original description information, the cluster of symptom original description information is indicated
It completes.It, can be according to symptom original description between different clustering clusters since different clustering clusters may belong to same class symptom
The highest similarity of information judges whether to merge clustering cluster, and executes corresponding processing.Such as with the first clustering cluster and
Second clustering cluster illustrates, and any symptom in any symptom original description information in the first clustering cluster and the second clustering cluster is former
Beginning description information carries out similarity mode, obtains similarity value.When each symptom original description information is equal in the first clustering cluster
After carrying out similarity mode with each symptom original description information in the second clustering cluster, all similarity values can be obtained, and
Then it is pre- to judge whether highest similarity is more than for the highest similarity value for determining symptom original description information in the two classes clustering cluster
If similarity threshold, if so, the first clustering cluster and the second clustering cluster are carried out clustering cluster merging, otherwise closed without clustering cluster
And.
In step 204, it after all clustering clusters merge judgement and processing, retouches the symptom of same clustering cluster is original
It is same symptoms description information to state information unification.
After all clustering clusters merge judgement and processing, each clustering cluster represents different symptoms, therefore can incite somebody to action
The symptom original description information unification of same clustering cluster is same symptoms description information, as a clustering cluster specified standard
Symptom description information.When for the symptom description information of clustering cluster specified value, a mark can be named taking human as the clustering cluster
Accurate symptom description information;The highest symptom description information of occurrence rate in clustering cluster can also be appointed as to the symptom description of standard
Information etc..
Due to containing the different symptoms original description information of same symptom in clustering cluster, when detect symptom original
When beginning description information belongs to a certain clustering cluster, which is converted to the disease of the corresponding standard of the clustering cluster
Shape description information.
As seen from the above-described embodiment, the present embodiment realizes the normalization of symptom by the way of cluster, easy to implement.
It, can will be from match information in order to improve the accuracy of symptom description information in an optional realization method
Context in the information extracted as candidate symptom, using candidate symptom with the similarity of character string of seed, to candidate symptom
It gives a mark, the candidate symptom information that score is met the requirements is unsatisfactory for requirement as symptom original description information, by score
Candidate symptom information is not as symptom original description information.Wherein, similarity of character string includes English character similarity, middle word
Accord with similarity, digital similarity etc..
As it can be seen that the information extracted from match information context is screened, to ensure that the information screened is
Symptom original description information, to improve the accuracy of symptom original description information.
For timing node, timing node is used to describe the duration passed through afterwards since disease hair, in order to will be same
The temporal information for occurring symptom under one disease is normalized to the relative time of the time on the basis of the same time, to convenient follow-up
Result is obtained to more parts of electronic health records of each disease type to integrate, and is improved and is integrated feasibility and integration efficiency.Time
Node is using the disease hair time started as initial time, for example, timing node can be " first day, second day, third day ... ", also
The describing modes such as can be " after one day, after five days, after 30 days ... ".
In one example, if the time for occurring symptom in electronic health record is described by the way of timing node
, then the temporal expression to prestore can be utilized, directly from the context of the symptom description information of at least a electronic health record
Extraction time node, to improve the efficiency for obtaining timing node.
Wherein, the temporal expression is the time description information not comprising the specific time, for example, " the * days ", " * is a
Moon " etc. is used to describe the expression formula of timing node.
For example, electronic health record:There is within first day slight sneezing, there is serious sneezing, symptom snotty in third day,
There is within 4th day serious sneezing, have a running nose, it is then to our hospital with chilly, the thin white symptom of tongue, it is diagnosed as " chill sense
It emits ".
As it can be seen that may be used " occurring ... ", " symptom occurred ... ", " with ... symptom " symptom mould is described
Formula extracts symptom description information;The temporal expression of " the * days " the extraction time node " first directly from case history may be used
It ", " third day ", " the 4th day ".
In another example, if the time for occurring symptom in electronic health record is carried out by the way of timing node
Description, but be described using other modes, then it needs temporal information being normalized to the time on the basis of the disease hair time
Relative time, there is the timing node of symptom to obtain.
Specifically, the timing node for symptom occur is obtained from least a electronic health record in advance, including:
Using the temporal expression to prestore, when being extracted from the context of the symptom description information of at least a electronic health record
Between information;
The temporal information extracted is normalized to the relative time of the time on the basis of the disease hair time, symptom occurs in acquisition
Timing node.
Wherein, the temporal expression is the time description information not comprising the specific time, is disease described in electronic health record
The phrase of shape time of origin.Such as temporal expression can be:Before * days, before the * months, before *, more than * before day etc..Temporal expression
It can be the deterministic expression by way of being manually entered.For example, user directly will be in temporal expression input system.Again
Such as, user is by common time sample (before such as three days, before one month, before more than ten days, before one week) input system, system pair
Chinese figure or Arabic numerals in time sample are replaced using fixed character, to obtain temporal expression etc..
After obtaining temporal expression, traversal search can be carried out in electronic health record, to extracting time information.In electricity
In sub- case history, the sequencing that temporal information occurs generally defers to the time sequencing that symptom actually occurs.Temporal information is by electronics
Case history is cut into description in different time periods, and the symptom extracted in every section represents the detailed illness of current slot.
Using the temporal expression to prestore, when being extracted from the context of the symptom description information of at least a electronic health record
Between information, to obtain the correspondence of temporal information and symptom description information, the temporal information extracted be symptom description letter
The symptom time of origin of breath.
The temporal expression to prestore is being utilized after extracting time information in the context of symptom description information, it can be by institute
The temporal information of extraction is normalized to the relative time of the time on the basis of the disease hair time, obtains the timing node for symptom occur.
Since the sequencing of temporal information appearance in electronic health record generally defers to the time sequencing that symptom actually occurs, because
First temporal information can be determined as the disease hair time by this.It, can be by all temporal information normalizings after determining the disease hair time
The relative time for turning to the time on the basis of the disease hair time, to obtain the timing node for symptom occur.
The application also provides a kind of specific method for normalizing, described to be normalized to send out with disease by the temporal information extracted
The relative time of time on the basis of time obtains the timing node for symptom occur, including:
The number in the temporal information is extracted, regard the number as time absolute value;
The chronomere in the temporal information is extracted, when time absolute value being scaled unified according to the chronomere
Between unit time value;
Extract the information for describing time relativeness in the temporal information;
Determine the disease hair time of disease in the electronic medical records;
According to the time value, the information and disease hair time for describing time relativeness, by the time
Information is normalized to the relative time of the time on the basis of the disease hair time, obtains the timing node for symptom occur.
Since different time unit description temporal information may be used in same a electronic health record, it can first extract
Number in temporal information regard number as time absolute value;Chronomere in extracting time information again, according to chronomere
Time absolute value is scaled to the time value of unified time unit.Same chronomere can preassign, for example, be appointed as day,
Week etc..Different time unit and the conversion relation of unified time unit can pre-establish, for example, being scaled within 1 week 7 days, one
The moon is scaled 30 days, is scaled within 1 year the conversion relations such as 365 days.
In temporal information, in addition to record has number, chronomere, the letter of time relativeness is described toward contact record
Breath, after which tends to occur at chronomere, such as the information such as " rear ", " preceding ".Therefore, it can extract in the temporal information
Information for describing time relativeness.
Since the sequencing of temporal information appearance in electronic health record generally defers to the time sequencing that symptom actually occurs, because
First temporal information can be determined as the disease hair time by this.
Time value, the information for describing time relativeness and disease hair the time determine after, can according to it is described when
Between value, it is described for describe time relativeness information and disease hair the time, by the temporal information be normalized to disease send out
The relative time of time on the basis of time, to obtain the timing node for symptom occur.
For example, electronic health record:There is double lower limb activity in patient before 3 months slightly unfavorable, and weak, nothing occurs in patient before 1 week
It becomes thin, no dizziness headache, no pale complexion, no fever chilly, no speech is ambiguous, to ask further diagnosis and treatment, comes our hospital today just
It examines.
As it can be seen that can be with extracting time information from the electronic health record using the temporal expression to prestore:Before 3 months, 1 week
Before, today.Temporal information can be converted to using application scheme:1st day, the 83rd day, the 90th day.
It is understood that the temporal information extracted can also be normalized to by the application using other normalization modes
The relative time of time, obtains the timing node for symptom occur, this is no longer going to repeat them on the basis of the disease hair time.
It, can be to each disease after obtaining disease type information, symptom description information and the timing node of symptom occur
More parts of electronic health records of sick type obtain result and are integrated, and obtain symptom description information and the time of each disease type information
The correspondence model of node.Since the timing node in each electronic health record is the opposite of the time on the basis of the disease hair time
Time, and the disease hair time of same disease is largely identical, therefore can be by the corresponding symptom description information of each disease type
It is integrated with the timing node for symptom occur, obtains the symptom description information of each disease type information and pair of timing node
Relational model is answered, record has the symptom that disease occurs under different time node in the correspondence model, therefore can utilize
The correspondence model carries out diseases analysis etc..
In one example, for every part of electronic health record, according to the disease type information of acquisition, symptom description information and
The timing node for symptom occur, integrates out the symptom description information that disease occurs in different time node in every part of electronic health record;
According to the symptom description information that disease in every part of electronic health record occurs in different time node, each disease type information is integrated out
Symptom description information and timing node correspondence model.
In the present embodiment, since the sequencing that timing node occurs in electronic health record generally defers to what symptom actually occurred
Time sequencing, and electronic health record is cut into description in different time periods by timing node, the symptom extracted in every section represents
The detailed illness of current slot, therefore, for every part of electronic health record, the sequencing that can occur according to timing node, with
And the correspondence of timing node and symptom description information, the sequencing of symptom description information intermediate node on time is arranged
Sequence, to integrate out the symptom description information that disease occurs in different time node in every part of electronic health record, and due to the electronics
Record has disease type information in case history, it is hereby achieved that disease-state-corresponding time relationship of the electronic health record.
After disease-state-corresponding time relationship of every part of electronic health record determines, each disease type letter can be integrated out
The symptom description information of breath and the correspondence model of timing node.
For example, according to disease-state-corresponding time relationship, symptom-corresponding time relationship is divided by disease type
Symptom-corresponding time relationship of class, same disease type is divided into same class, and symptom-time of various disease type corresponds to
Relationship is divided into inhomogeneity.For same disease type, first judge in different symptoms-corresponding time relationship, the disease hair time pair
Whether the symptom answered is identical, if identical, then it represents that timing node is at the same time in different symptom-corresponding time relationships
On the basis of the time, different symptom-corresponding time relationships can be integrated;If it is not the same, then indicating different symptoms-
Timing node is not the relative time of the time on the basis of at the same time in corresponding time relationship, and hand is handled as one of which
Section can extract the identical symptom-corresponding time relationship of disease hair time corresponding symptom, and the symptom of extraction-time is corresponded to
Relationship is integrated, and the symptom-corresponding time relationship not extracted does not do integration processing.As another processing means, can incite somebody to action
Different symptom-corresponding time relationships are compared, thus it is speculated that go out symptom when really disease hair time and disease hair, and will be different
Symptom-corresponding time relationship in timing node be normalized to by speculate disease hair the time on the basis of the time relative time, then
Correspondence after normalization is integrated.
As seen from the above-described embodiment, the present embodiment is first integrated out disease in every part of electronic health record and is occurred in different time node
Symptom description information integrate further according to the symptom description information that disease in every part of electronic health record occurs in different time node
Go out the symptom description information of each disease type information and the correspondence model of timing node, integration efficiency can be improved, and
It is easy to implement.
After building correspondence model, diseases analysis can be carried out using the correspondence model of structure.For example, obtaining
After targeted exposure symptoms information and its timing node, in the correspondence model built in advance, search with the targeted exposure symptoms information and
The correspondence that its timing node matches, and diseases analysis is carried out according to lookup result.
Wherein, targeted exposure symptoms information can be the symptom information input by user for needing to inquire, and timing node is that mesh occur
Mark the timing node of symptom.Since the symptom that the correspondence model built in advance includes each disease type information describes letter
Therefore the correspondence of breath and timing node can be searched and the targeted exposure symptoms information and its time in correspondence model
The correspondence that node matches, and diseases analysis is carried out according to lookup result.
Since the symptom description information that correspondence model includes each disease type information is corresponding with timing node
Relationship, therefore a variety of diseases analysis can be carried out.In one example, diseases analysis may include medical diagnosis on disease, then can root
The corresponding disease type of the targeted exposure symptoms is determined according to lookup result.In another example, diseases analysis can also include disease
Disease forecasting determines the corresponding disease type of targeted exposure symptoms and disease type after the timing node according to lookup result
The symptom being likely to occur.As it can be seen that correspondence model can be utilized to obtain the information such as symptom, the tendency of disease, it can be used for medicine
The fields such as education, monitoring, clinical decision support.
It is understood that diseases analysis can also be other analyses, as long as the symptom dependent on disease type information
The analysis that the correspondence of description information and timing node carries out, will not enumerate herein.
In an optional realization method, the sequelae data of disease are can be combined with, analyze disease time and symptom
The correlativity of sequelae, to speculate the probability for suffering from sequelae.The present embodiment builds sequelae model in advance.As shown in figure 3,
Fig. 3 is one embodiment flow chart that sequelae model is built in the application medical information processing method, includes the following steps 301
To step 303:
In step 301, from least a electronic health record, while obtaining disease type information, sequelae letter is obtained
The symptom description information occurred when breath and for the first time medical treatment.
When certain disease is there are when corresponding sequelae, often record has sequelae information in electronic health record, therefore can
To extract sequelae information from electronic health record.When not only having included disease type information in electronic health record, but also including sequelae information
When, disease type is Newly diagnosed as a result, sequelae is certain disease symptoms carried over after disease turns for the better by and large.It should
In embodiment, Waiting time can be first determined, then the corresponding symptom of Waiting time is determined as the symptom that the when of seeing a doctor occurs and is described
Information.It is illustrated below with a specific example:
For example, electronic health record:It is slightly unfavorable without occurring double lower limb activity under apparent inducement before patient's 1 year and a half, occur within more than 1 year
Double lower limb moving obstacle, inconvenient walking and with dizzy, distortion of commissure symptom hospitalize our hospital, diagnose " cerebral infarction ", give silver
Apricot is slightly better after improving the treatments such as cycle up to not needle, ozagrel needle.Double lower limb activity is slightly unfavorable before 1 month, today patient
Double lower limb activity is unfavorable, and appearance is weak, but without becoming thin, no dizziness headache, no pale complexion, no fever chilly, no speech is ambiguous,
It is to our hospital to ask further diagnosis and treatment, it is hospitalized and is admitted to hospital with " sequelae for cerebral infraction ".
In the electronic health record, disease type information is:Cerebral infarction;Waiting time is for the first time:More than 1 year, when medical treatment, occurred
Symptom description information be:Double lower limb moving obstacle, inconvenient walking, dizziness, distortion of commissure;The sequelae information of cerebral infarction is:
Double lower limb activity is unfavorable, weak.Further, the symptom not occurred can also be supplemented, for example, without becoming thin, it is without a head
Dizzy headache, no pale complexion, no fever chilly, no speech are ambiguous.
In step 302, based on disease type information, the sequelae information and for the first time obtained in more parts of electronic health records
The symptom description information occurred when doctor, symptom description information that when medical treatment for establishing each disease type information occurs and rear loses
The correspondence of disease information.
In disease type information, sequelae information and the symptom that occurs when seeing a doctor for the first time in obtaining every part of electronic health record
After description information, symptom description information that when medical treatment that can establish each disease type information occurs and sequelae information
Correspondence.
In step 303, according to the frequency of occurrences of the correspondence, the probability that patient suffers from sequelae is calculated, and according to
After result of calculation builds the correspondence of the symptom description information and sequelae probability occurred when the medical treatment of disease type information
Lose disease model.
Wherein, about the frequency of occurrences of correspondence, in one example, the frequency of occurrences can be based on disease type and believe
Occur when the correspondence of the symptom description information-sequelae information occurred when breath-medical treatment and disease type information-medical treatment
The ratio of the correspondence of symptom description information generates.
Using the frequency of occurrences of correspondence as the factor for suffering from sequelae probability, suffer from sequelae to calculate acquisition patient
Probability, and build the rear something lost of the correspondence of the symptom description information and sequelae probability occurred when the medical treatment of disease type information
Disease model.
After obtaining sequelae model, it can be predicted using the probability that sequelae model suffers from sequelae to patient.Tool
Body, after obtaining the symptom description information occurred when target disease type information and medical treatment, in the sequelae mould built in advance
In type, the correspondence to match with the symptom description information occurred when the target disease type information and medical treatment is searched, and
The probability that patient suffers from sequelae is determined according to lookup result.
As seen from the above-described embodiment, different Waiting times cause the probability for suffering from sequelae different after symptom due to occurring, this
Mode is after building sequelae model, in the sequelae model built in advance, search with the target disease type information and
The correspondence that the symptom description information occurred when medical treatment matches, and determine that patient suffers from the general of sequelae according to lookup result
Rate, to realize the prediction for the probability for suffering from sequelae to patient.
Various technical characteristics in embodiment of above can be arbitrarily combined, as long as the combination between feature is not present
Conflict or contradiction, but as space is limited, it is not described one by one, therefore the various technical characteristics in the above embodiment is arbitrary
It is combined the range for also belonging to this disclosure.
One of which combination is exemplified below to be illustrated.As shown in figure 4, Fig. 4 is the application medical information processing side
Another embodiment flow chart of method.The building process of correspondence model is mainly introduced in the flow chart.
41, the time extracts.Using the temporal expression to prestore, from the upper of at least symptom description information of portion electronic health record
Hereinafter extracting time information, the temporal expression are the time description informations not comprising the specific time.In electronic health record,
The sequencing that temporal expression occurs generally defers to the time sequencing that symptom actually occurs.Temporal expression cuts case history text
It is divided into description in different time periods, the symptom extracted in every section represents the detailed symptom of current slot.
42, symptom is extracted.
421, seed is extracted.Record has known symptom description information (seed) in seed database, utilizes matching algorithm
Seed is extracted from least a electronic health record.
422, pattern learning.Based on extracted seed, extracted from the context of seed in electronic health record character and
Identify the position relationship of seed and character;According to the frequency of occurrences of the character extracted and the position relationship identified, really
Determine symptom and describes pattern.
423, pattern match.Symptom is described into the character in pattern, is matched at least a electronic health record, it is described
It includes the character and the position of symptom description information and character pass that symptom description information context will appear that symptom, which describes pattern,
System;According to the position relationship of the symptom description information and character, symptom description information is obtained from the context of match information.
Wherein, for the correspondence for determining symptom description information Yu the timing node of symptom occur, first with timetable
Electronic health record is cut into description in different time periods up to formula, when pattern match, pattern can be described using symptom from cutting
Description in carry out character match, to which the symptom description information of acquisition and the period are established correspondence, i.e. time
Section is the symptom time of occurrence of the symptom description information.
424, seed is obtained.Using the symptom description information of acquisition as seed, repeat pattern learning, pattern match and
Seed step is obtained, until no longer finding new symptom description information.
43, symptom & time normalizations.The symptom original description information of same symptom is normalized to identical symptom description
Information.The temporal information extracted is normalized to the relative time of the time on the basis of the disease hair time, symptom occurs in acquisition
Timing node.
44, correspondence model.Result is obtained to more parts of electronic health records of each disease type to integrate, and obtains each
The symptom description information of disease type information and the correspondence model of timing node.
45, sequelae model.The construction step of the sequelae model includes:From at least a electronic health record, obtain
While disease type information, the symptom description information occurred when sequelae information and for the first time medical treatment is obtained;Based on more parts of electricity
The symptom description information occurred when the disease type information, sequelae information and the medical treatment for the first time that are obtained in sub- case history, is established every
The correspondence of the symptom description information and sequelae information that occur when the medical treatment of kind disease type information;According to the corresponding pass
The frequency of occurrences of system calculates patient and suffers from the probability of sequelae, and while building the medical treatment of disease type information according to result of calculation goes out
The sequelae model of existing symptom description information and the correspondence of sequelae probability.
46, diseases analysis.After obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance
In, the correspondence to match with the targeted exposure symptoms information and its timing node is searched, and disease point is carried out according to lookup result
Analysis.After obtaining the symptom description information occurred when target disease type information and medical treatment, in the sequelae model built in advance
In, search the correspondence to match with the symptom description information occurred when the target disease type information and medical treatment, and root
The probability that patient suffers from sequelae is determined according to lookup result.
In the embodiment of the present application, the electronic health record of all section office of Different hospital can be uploaded in cloud server terminal,
Such as it is uploaded on Ali's cloud.Correspondence model and sequelae are established using the data in application scheme and cloud server terminal
Model.If receive client transmission diseases analysis request, according to diseases analysis ask obtain targeted exposure symptoms information and its
Timing node, in the correspondence model built in advance, lookup matches with the targeted exposure symptoms information and its timing node
Correspondence, and diseases analysis is carried out according to lookup result, and analysis result is back to the client.If receiving visitor
The sequelae probabilistic query request that family end is sent, according to sequelae probabilistic query request acquisition target disease type information and just
The symptom description information occurred when doctor, in the sequelae model built in advance, search with the target disease type information and
The correspondence that the symptom description information occurred when medical treatment matches, and determine that patient suffers from the general of sequelae according to lookup result
Rate;And the probability that patient suffers from sequelae is sent to the client.
All tidal data recoverings are obtained fairly perfect electronic health record library by the application.Data volume is bigger in electronic health record library,
After building correspondence model and sequelae model according to the electronic health record in electronic health record library, using correspondence model with after
It is more accurate to lose disease model progress diseases analysis.
Corresponding with the embodiment of the application medical information processing method, present invention also provides medical information processing to fill
It sets, the embodiment of readable medium and electronic equipment.
The application provides one or more machine readable medias, instruction is stored thereon with, when by one or more processors
When execution so that terminal device executes medical information processing method as described above.
The embodiment of the application medical information processing device can be applied on various electronic equipments, for example, the electronics is set
Standby may include mobile phone, tablet computer, PC etc..Wherein, device embodiment can by software realization, can also by hardware or
The mode of person's software and hardware combining is realized.It is by electricity where it as the device on a logical meaning for implemented in software
Corresponding computer program instructions in nonvolatile memory are read what operation in memory was formed by the processor of sub- equipment.From
For hardware view, as shown in figure 5, for a kind of hardware configuration of 531 place electronic equipment of the application medical information processing device
Figure is implemented other than processor 510 shown in fig. 5, memory 530, network interface 540 and nonvolatile memory 520
Electronic equipment in example where device can also include other hardware generally according to the actual functional capability of the equipment, not another in Fig. 5
One shows.
It is one embodiment block diagram of the application medical information processing device referring to Fig. 6:
The device includes:Model construction module 610 and information analysis module 620.
Model construction module 610, for from least a electronic health record, obtaining disease type information, symptom is retouched in advance
It states information and the timing node of symptom occurs, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease;It is right
More parts of electronic health records of each disease type obtain result and are integrated, and obtain the symptom description information of each disease type information
With the correspondence model of timing node.
Information analysis module 620, after obtaining targeted exposure symptoms information and its timing node, in the corresponding pass built in advance
It is to search the correspondence to match with the targeted exposure symptoms information and its timing node, and carry out according to lookup result in model
Diseases analysis.
In an optional realization method, the model construction module 610 includes (Fig. 6 is not shown):
Information matches module, for the symptom to prestore to be described the character in pattern, at least a electronic health record into
Row matching, the symptom describe pattern include character that symptom description information context will appear and symptom description information with
The position relationship of character.
Symptom obtains module, for the position relationship according to the symptom description information and character, from the upper of match information
Hereinafter obtain symptom description information.
In an optional realization method, the model construction module 610 further includes that (Fig. 6 does not show mode decision module
Go out), it is used for:
Using known symptom description information as seed, and kind is extracted from least a electronic health record using matching algorithm
Son;
Based on extracted seed, character is extracted from the context of seed in electronic health record and identify seed with
The position relationship of character;
According to the frequency of occurrences of the character extracted and the position relationship identified, determine that symptom describes pattern.
In an optional realization method, the symptom obtains module and is specifically used for:
According to the position relationship of the symptom description information and character, it is original that symptom is extracted from the context of match information
Description information;
The symptom original description information of same symptom is normalized to identical symptom description information.
In an optional realization method, the symptom obtains module and is specifically used for:
For partial symptoms original description information in extraction information, each symptom that same time in same disease is occurred is former
Beginning description information is divided into different clustering clusters;
The similarity of current symptomatic original description information and symptom original description information in clustering cluster is calculated, and according to similar
Degree determines whether the clustering cluster or newly-built clustering cluster to be added in current symptomatic original description information and by current symptomatic original description
Newly-built clustering cluster is added in information, and the current symptomatic original description information is to extract the symptom for not having that clustering cluster is added in information
Original description information;
After corresponding clustering cluster is added in all symptom original description information, retouched according to symptom is original between different clustering clusters
The highest similarity value for stating information, judges whether to merge clustering cluster, and executes corresponding processing;
After all clustering clusters merge judgement and processing, it is by the symptom original description information unification of same clustering cluster
Same symptoms description information.
In an optional realization method, the model construction module 610 includes (Fig. 6 is not shown):
Information extraction modules, for using the temporal expression to prestore, letter to be described from the symptom of at least a electronic health record
Extracting time information in the context of breath, the temporal expression are the time description informations not comprising the specific time.
Time normalizing module, for by the temporal information extracted be normalized to by disease send out the time on the basis of the time it is opposite
Time obtains the timing node for symptom occur.
In an optional realization method, the time normalizing module is specifically used for:
The number in the temporal information is extracted, regard the number as time absolute value;
The chronomere in the temporal information is extracted, when time absolute value being scaled unified according to the chronomere
Between unit time value;
Extract the information for describing time relativeness in the temporal information;
Determine the disease hair time of disease in the electronic medical records;
According to the time value, the information and disease hair time for describing time relativeness, by the time
Information is normalized to the relative time of the time on the basis of the disease hair time, obtains the timing node for symptom occur.
In an optional realization method, the model construction module 610 includes that (Fig. 6 does not show information integration module
Go out), it is used for:
For every part of electronic health record, according to the disease type information of acquisition, symptom description information and occur symptom when
Intermediate node integrates out the symptom description information that disease occurs in different time node in every part of electronic health record;
According to the symptom description information that disease in every part of electronic health record occurs in different time node, each disease is integrated out
The symptom description information of type information and the correspondence model of timing node.
In an optional realization method, described device further includes probability analysis module, is used for:
After obtaining the symptom description information occurred when target disease type information and medical treatment, in the sequelae built in advance
In model, the correspondence to match with the symptom description information occurred when the target disease type information and medical treatment is searched,
And the probability that patient suffers from sequelae is determined according to lookup result;
The model construction module is additionally operable to:
From at least a electronic health record, while obtaining disease type information, sequelae information and for the first time is obtained
The symptom description information occurred when doctor;
The disease occurred when based on the disease type information, sequelae information and medical treatment for the first time obtained in more parts of electronic health records
Shape description information, symptom description information that when medical treatment for establishing each disease type information occurs is corresponding with sequelae information to close
System;
According to the frequency of occurrences of the correspondence, calculates patient and suffer from the probability of sequelae, and built according to result of calculation
The sequelae model of the symptom description information occurred when the medical treatment of disease type information and the correspondence of sequelae probability.
Based on this, the application also provides a kind of electronic equipment, including:
Processor;Memory for storing the processor-executable instruction;
Wherein, the processor is configured as:
After obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance, search and the mesh
The correspondence that mark symptom information and its timing node match, and diseases analysis is carried out according to lookup result;
The building process of the correspondence model includes:
From at least a electronic health record, obtains disease type information, symptom description information and the time of symptom occur
Node, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease;
Result is obtained to more parts of electronic health records of each disease type to integrate, and obtains the disease of each disease type information
The correspondence model of shape description information and timing node.
The function of modules and the realization process of effect specifically refer to and correspond to step in the above method in above-mentioned apparatus
Realization process, details are not described herein.
For device embodiments, since it corresponds essentially to embodiment of the method, so related place is referring to method reality
Apply the part explanation of example.The apparatus embodiments described above are merely exemplary, wherein described be used as separating component
The module of explanation may or may not be physically separated, and the component shown as module can be or can also
It is not physical module, you can be located at a place, or may be distributed on multiple network modules.It can be according to actual
It needs that some or all of module therein is selected to realize the purpose of application scheme.Those of ordinary skill in the art are not paying
In the case of going out creative work, you can to understand and implement.
Those skilled in the art will readily occur to its of the application after considering specification and putting into practice the invention applied here
Its embodiment.This application is intended to cover any variations, uses, or adaptations of the application, these modifications, purposes or
Person's adaptive change follows the general principle of the application and includes the common knowledge in the art that the application does not apply
Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the application are by following
Claim is pointed out.
It should be understood that the application is not limited to the precision architecture for being described above and being shown in the accompanying drawings, and
And various modifications and changes may be made without departing from the scope thereof.Scope of the present application is only limited by the accompanying claims.
Claims (19)
1. a kind of medical information processing method, which is characterized in that the method includes:
In advance from least a electronic health record, obtains disease type information, symptom description information and the time of symptom occur
Node, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease;
Result is obtained to more parts of electronic health records of each disease type to integrate, obtain the disease of each disease type information in advance
The correspondence model of shape description information and timing node;
After obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance, search and the target disease
The correspondence that shape information and its timing node match, and diseases analysis is carried out according to lookup result.
2. according to the method described in claim 1, it is characterized in that, obtaining symptom from least a electronic health record in advance and retouching
Information is stated, including:
The symptom to prestore is described into the character in pattern, is matched at least a electronic health record, the symptom describes mould
Formula includes the character that symptom description information context will appear and the position relationship of symptom description information and character;
According to the position relationship of the symptom description information and character, symptom description letter is obtained from the context of match information
Breath.
3. according to the method described in claim 2, it is characterized in that, the determination step that the symptom describes pattern includes:
Using known symptom description information as seed, and seed is extracted from least a electronic health record using matching algorithm;
Based on extracted seed, character is extracted from the context of seed in electronic health record and identifies seed and character
Position relationship;
According to the frequency of occurrences of the character extracted and the position relationship identified, determine that symptom describes pattern.
4. according to the method described in claim 2, it is characterized in that, the position according to the symptom description information and character
Relationship obtains symptom description information from the context of match information, including:
According to the position relationship of the symptom description information and character, symptom original description is extracted from the context of match information
Information;
The symptom original description information of same symptom is normalized to identical symptom description information.
5. according to the method described in claim 4, it is characterized in that, the symptom original description information normalizing by same symptom
Identical symptom description information is turned to, including:
For partial symptoms original description information in extraction information, each symptom that same time in same disease occurs original is retouched
It states information and is divided into different clustering clusters;
The similarity of current symptomatic original description information and symptom original description information in clustering cluster is calculated, and true according to similarity
It is fixed whether the clustering cluster or newly-built clustering cluster to be added in current symptomatic original description information and by current symptomatic original description information
Newly-built clustering cluster is added, the current symptomatic original description information is to extract not having the symptom of addition clustering cluster original in information
Description information;
After corresponding clustering cluster is added in all symptom original description information, believed according to symptom original description between different clustering clusters
The highest similarity value of breath, judges whether to merge clustering cluster, and executes corresponding processing;
It is identical by the symptom original description information unification of same clustering cluster after all clustering clusters merge judgement and processing
Symptom description information.
6. according to the method described in claim 1, it is characterized in that, in advance from least a electronic health record, there is disease in acquisition
The timing node of shape, including:
Using the temporal expression to prestore, the extraction time letter from the context of the symptom description information of at least a electronic health record
Breath, the temporal expression are the time description informations not comprising the specific time;
By the temporal information extracted be normalized to by disease send out the time on the basis of the time relative time, obtain occur symptom when
Intermediate node.
7. according to the method described in claim 6, it is characterized in that, described be normalized to send out with disease by the temporal information extracted
The relative time of time on the basis of time obtains the timing node for symptom occur, including:
The number in the temporal information is extracted, regard the number as time absolute value;
The chronomere in the temporal information is extracted, time absolute value is scaled by unified time list according to the chronomere
The time value of position;
Extract the information for describing time relativeness in the temporal information;
Determine the disease hair time of disease in the electronic medical records;
According to the time value, the information and disease hair time for describing time relativeness, by the temporal information
It is normalized to the relative time of the time on the basis of the disease hair time, obtains the timing node for symptom occur.
8. method according to any one of claims 1 to 7, which is characterized in that described in advance to the more of each disease type
Part electronic health record obtains result and is integrated, and the symptom description information for obtaining each disease type information is corresponding with timing node
Relational model, including:
For every part of electronic health record, according to the disease type information of acquisition, symptom description information and there is the when segmentum intercalaris of symptom
Point integrates out the symptom description information that disease occurs in different time node in every part of electronic health record;
According to the symptom description information that disease in every part of electronic health record occurs in different time node, each disease type is integrated out
The symptom description information of information and the correspondence model of timing node.
9. method according to any one of claims 1 to 7, which is characterized in that the method further includes:
After obtaining the symptom description information occurred when target disease type information and medical treatment, in the sequelae model built in advance
In, search the correspondence to match with the symptom description information occurred when the target disease type information and medical treatment, and root
The probability that patient suffers from sequelae is determined according to lookup result;
The construction step of the sequelae model includes:
From at least a electronic health record, while obtaining disease type information, when obtaining sequelae information and seeing a doctor for the first time
The symptom description information of appearance;
The symptom occurred when based on the disease type information, sequelae information and medical treatment for the first time obtained in more parts of electronic health records is retouched
Information is stated, the correspondence for the symptom description information and sequelae information that when medical treatment for establishing each disease type information occurs;
According to the frequency of occurrences of the correspondence, the probability that patient suffers from sequelae is calculated, and disease is built according to result of calculation
The sequelae model of the symptom description information occurred when the medical treatment of type information and the correspondence of sequelae probability.
10. a kind of medical information processing device, which is characterized in that described device includes:
Model construction module, for from least a electronic health record, obtain in advance disease type information, symptom description information with
And there is the timing node of symptom, the duration that the timing node is used to describe afterwards to be passed through since being sent out disease;To each disease
More parts of electronic health records of type obtain result and are integrated, obtain the symptom description information of each disease type information with when segmentum intercalaris
The correspondence model of point;
Information analysis module, after obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance
In, the correspondence to match with the targeted exposure symptoms information and its timing node is searched, and disease point is carried out according to lookup result
Analysis.
11. device according to claim 10, which is characterized in that the model construction module includes:
Information matches module, for the symptom to prestore to be described the character in pattern, the progress at least a electronic health record
Match, it includes the character and symptom description information and character that symptom description information context will appear that the symptom, which describes pattern,
Position relationship;
Symptom obtains module, for the position relationship according to the symptom description information and character, from the context of match information
Middle acquisition symptom description information.
12. according to the devices described in claim 11, which is characterized in that the model construction module further includes that pattern determines mould
Block is used for:
Using known symptom description information as seed, and seed is extracted from least a electronic health record using matching algorithm;
Based on extracted seed, character is extracted from the context of seed in electronic health record and identifies seed and character
Position relationship;
According to the frequency of occurrences of the character extracted and the position relationship identified, determine that symptom describes pattern.
13. according to the devices described in claim 11, which is characterized in that the symptom obtains module and is specifically used for:
According to the position relationship of the symptom description information and character, symptom original description is extracted from the context of match information
Information;
The symptom original description information of same symptom is normalized to identical symptom description information.
14. device according to claim 13, which is characterized in that the symptom obtains module and is specifically used for:
For partial symptoms original description information in extraction information, each symptom that same time in same disease occurs original is retouched
It states information and is divided into different clustering clusters;
The similarity of current symptomatic original description information and symptom original description information in clustering cluster is calculated, and true according to similarity
It is fixed whether the clustering cluster or newly-built clustering cluster to be added in current symptomatic original description information and by current symptomatic original description information
Newly-built clustering cluster is added, the current symptomatic original description information is to extract not having the symptom of addition clustering cluster original in information
Description information;
After corresponding clustering cluster is added in all symptom original description information, believed according to symptom original description between different clustering clusters
The highest similarity value of breath, judges whether to merge clustering cluster, and executes corresponding processing;
It is identical by the symptom original description information unification of same clustering cluster after all clustering clusters merge judgement and processing
Symptom description information.
15. device according to claim 10, which is characterized in that the model construction module includes:
Information extraction modules, for utilizing the temporal expression to prestore, from at least symptom description information of portion electronic health record
Extracting time information in context, the temporal expression are the time description informations not comprising the specific time;
Time normalizing module, for by the temporal information extracted be normalized to by disease send out the time on the basis of the time it is opposite when
Between, obtain the timing node for symptom occur.
16. device according to claim 15, which is characterized in that the time normalizing module is specifically used for:
The number in the temporal information is extracted, regard the number as time absolute value;
The chronomere in the temporal information is extracted, time absolute value is scaled by unified time list according to the chronomere
The time value of position;
Extract the information for describing time relativeness in the temporal information;
Determine the disease hair time of disease in the electronic medical records;
According to the time value, the information and disease hair time for describing time relativeness, by the temporal information
It is normalized to the relative time of the time on the basis of the disease hair time, obtains the timing node for symptom occur.
17. according to claim 10 to 16 any one of them device, which is characterized in that the model construction module includes information
Module is integrated, is used for:
For every part of electronic health record, according to the disease type information of acquisition, symptom description information and there is the when segmentum intercalaris of symptom
Point integrates out the symptom description information that disease occurs in different time node in every part of electronic health record;
According to the symptom description information that disease in every part of electronic health record occurs in different time node, each disease type is integrated out
The symptom description information of information and the correspondence model of timing node.
18. according to claim 10 to 16 any one of them device, which is characterized in that described device further includes probability analysis mould
Block is used for:
After obtaining the symptom description information occurred when target disease type information and medical treatment, in the sequelae model built in advance
In, search the correspondence to match with the symptom description information occurred when the target disease type information and medical treatment, and root
The probability that patient suffers from sequelae is determined according to lookup result;
The model construction module is additionally operable to:
From at least a electronic health record, while obtaining disease type information, when obtaining sequelae information and seeing a doctor for the first time
The symptom description information of appearance;
The symptom occurred when based on the disease type information, sequelae information and medical treatment for the first time obtained in more parts of electronic health records is retouched
Information is stated, the correspondence for the symptom description information and sequelae information that when medical treatment for establishing each disease type information occurs;
According to the frequency of occurrences of the correspondence, the probability that patient suffers from sequelae is calculated, and disease is built according to result of calculation
The sequelae model of the symptom description information occurred when the medical treatment of type information and the correspondence of sequelae probability.
19. a kind of electronic equipment, which is characterized in that including:
Processor;Memory for storing the processor-executable instruction;
Wherein, the processor is configured as:
After obtaining targeted exposure symptoms information and its timing node, in the correspondence model built in advance, search and the target disease
The correspondence that shape information and its timing node match, and diseases analysis is carried out according to lookup result;
The building process of the correspondence model includes:
From at least a electronic health record, obtains disease type information, symptom description information and the timing node of symptom occurs,
The duration that the timing node is used to describe afterwards to be passed through since being sent out disease;
Result is obtained to more parts of electronic health records of each disease type to integrate, the symptom for obtaining each disease type information is retouched
State the correspondence model of information and timing node.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710093245.7A CN108461110B (en) | 2017-02-21 | 2017-02-21 | Medical information processing method, device and equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710093245.7A CN108461110B (en) | 2017-02-21 | 2017-02-21 | Medical information processing method, device and equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108461110A true CN108461110A (en) | 2018-08-28 |
CN108461110B CN108461110B (en) | 2021-07-23 |
Family
ID=63221710
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710093245.7A Active CN108461110B (en) | 2017-02-21 | 2017-02-21 | Medical information processing method, device and equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108461110B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109360658A (en) * | 2018-11-01 | 2019-02-19 | 北京航空航天大学 | A kind of the disease pattern method for digging and device of word-based vector model |
CN109978022A (en) * | 2019-03-08 | 2019-07-05 | 腾讯科技(深圳)有限公司 | A kind of medical treatment text message processing method and device, storage medium |
CN110391026A (en) * | 2019-07-25 | 2019-10-29 | 北京百度网讯科技有限公司 | Information classification approach, device and equipment based on medical probability graph |
CN111539478A (en) * | 2020-04-26 | 2020-08-14 | 成都见田科技有限公司 | Intelligent diagnosis method, system and diagnosis equipment for elevator faults |
CN111599483A (en) * | 2020-05-29 | 2020-08-28 | 安徽科大讯飞医疗信息技术有限公司 | Medical record set optimization method, device, equipment and storage medium |
CN111785386A (en) * | 2020-06-30 | 2020-10-16 | 安徽科大讯飞医疗信息技术有限公司 | Time interval dividing method, related device and readable storage medium |
CN112148882A (en) * | 2020-11-27 | 2020-12-29 | 北京惠及智医科技有限公司 | Medical record text processing method and device, electronic equipment and storage medium |
CN113094477A (en) * | 2021-06-09 | 2021-07-09 | 腾讯科技(深圳)有限公司 | Data structuring method and device, computer equipment and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101034410A (en) * | 2007-04-16 | 2007-09-12 | 浙江大学 | Medical treatment case index cast and electronic medical record system based on same |
CN101236579A (en) * | 2008-02-20 | 2008-08-06 | 杭州创业软件股份有限公司 | Dynamic structured electronic patient history |
CN100416583C (en) * | 2000-02-14 | 2008-09-03 | 第一咨询公司 | Automated diagnostic system and method |
CN103761437A (en) * | 2014-01-20 | 2014-04-30 | 中国中医科学院 | Scientific research data automatic generation system based on clinical data |
CN105389465A (en) * | 2015-10-29 | 2016-03-09 | 徐翼 | Medical record data processing method, apparatus and system |
CN106228188A (en) * | 2016-07-22 | 2016-12-14 | 北京市商汤科技开发有限公司 | Clustering method, device and electronic equipment |
-
2017
- 2017-02-21 CN CN201710093245.7A patent/CN108461110B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100416583C (en) * | 2000-02-14 | 2008-09-03 | 第一咨询公司 | Automated diagnostic system and method |
CN101034410A (en) * | 2007-04-16 | 2007-09-12 | 浙江大学 | Medical treatment case index cast and electronic medical record system based on same |
CN101236579A (en) * | 2008-02-20 | 2008-08-06 | 杭州创业软件股份有限公司 | Dynamic structured electronic patient history |
CN103761437A (en) * | 2014-01-20 | 2014-04-30 | 中国中医科学院 | Scientific research data automatic generation system based on clinical data |
CN105389465A (en) * | 2015-10-29 | 2016-03-09 | 徐翼 | Medical record data processing method, apparatus and system |
CN106228188A (en) * | 2016-07-22 | 2016-12-14 | 北京市商汤科技开发有限公司 | Clustering method, device and electronic equipment |
Non-Patent Citations (3)
Title |
---|
丁宝芬: "《医学信息学》", 30 September 2009, 东南大学出版社 * |
利用条件随机场实现中文病历文本中时间关系的自动提取: "利用条件随机场实现中文病历文本中时间关系的自动提取", 《中国生物医学工程学报》 * |
王建洪: "中文电子病历信息提取方法研究", 《万方数据知识服务平台》 * |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109360658A (en) * | 2018-11-01 | 2019-02-19 | 北京航空航天大学 | A kind of the disease pattern method for digging and device of word-based vector model |
CN109360658B (en) * | 2018-11-01 | 2021-06-08 | 北京航空航天大学 | Disease pattern mining method and device based on word vector model |
CN109978022A (en) * | 2019-03-08 | 2019-07-05 | 腾讯科技(深圳)有限公司 | A kind of medical treatment text message processing method and device, storage medium |
CN110490251A (en) * | 2019-03-08 | 2019-11-22 | 腾讯科技(深圳)有限公司 | Prediction disaggregated model acquisition methods and device, storage medium based on artificial intelligence |
CN110391026B (en) * | 2019-07-25 | 2022-04-26 | 北京百度网讯科技有限公司 | Information classification method, device and equipment based on medical probability map |
CN110391026A (en) * | 2019-07-25 | 2019-10-29 | 北京百度网讯科技有限公司 | Information classification approach, device and equipment based on medical probability graph |
CN111539478A (en) * | 2020-04-26 | 2020-08-14 | 成都见田科技有限公司 | Intelligent diagnosis method, system and diagnosis equipment for elevator faults |
CN111599483A (en) * | 2020-05-29 | 2020-08-28 | 安徽科大讯飞医疗信息技术有限公司 | Medical record set optimization method, device, equipment and storage medium |
CN111599483B (en) * | 2020-05-29 | 2023-12-22 | 讯飞医疗科技股份有限公司 | Medical record set optimization method, device, equipment and storage medium |
CN111785386A (en) * | 2020-06-30 | 2020-10-16 | 安徽科大讯飞医疗信息技术有限公司 | Time interval dividing method, related device and readable storage medium |
CN111785386B (en) * | 2020-06-30 | 2024-04-05 | 讯飞医疗科技股份有限公司 | Time interval division method, related device and readable storage medium |
CN112148882B (en) * | 2020-11-27 | 2021-09-10 | 北京惠及智医科技有限公司 | Medical record text processing method and device, electronic equipment and storage medium |
CN112148882A (en) * | 2020-11-27 | 2020-12-29 | 北京惠及智医科技有限公司 | Medical record text processing method and device, electronic equipment and storage medium |
CN113094477B (en) * | 2021-06-09 | 2021-08-31 | 腾讯科技(深圳)有限公司 | Data structuring method and device, computer equipment and storage medium |
CN113094477A (en) * | 2021-06-09 | 2021-07-09 | 腾讯科技(深圳)有限公司 | Data structuring method and device, computer equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN108461110B (en) | 2021-07-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108461110A (en) | Medical information processing method, device and equipment | |
Garcia-Ceja et al. | Mental health monitoring with multimodal sensing and machine learning: A survey | |
Ambekar et al. | Disease risk prediction by using convolutional neural network | |
CN108899064A (en) | Electronic health record generation method, device, computer equipment and storage medium | |
CN106951703B (en) | System and method for generating electronic medical record | |
CN110675944A (en) | Triage method and device, computer equipment and medium | |
CN109493979A (en) | A kind of disease forecasting method and apparatus based on intelligent decision | |
US11529105B2 (en) | Digital twin updating | |
CN111341456A (en) | Method and device for generating diabetic foot knowledge map and readable storage medium | |
CN109065183B (en) | Method and device for generating inquiry dialogue template and acquiring inquiry data | |
US20220165417A1 (en) | Population-level gaussian processes for clinical time series forecasting | |
US20180151254A1 (en) | High-speed similar case search method and device through reduction of large scale multi-dimensional time series health data to multiple dimensions | |
CN110413981A (en) | The based reminding method and device of the quality detecting method of electronic health record, similar case history | |
CN112447270A (en) | Medication recommendation method, device, equipment and storage medium | |
CN113679348A (en) | Blood glucose prediction method, blood glucose prediction device, blood glucose prediction apparatus, and storage medium | |
CN112489747A (en) | Chronic patient supervision method, device, equipment and medium based on analysis model | |
CN110752027A (en) | Electronic medical record data pushing method and device, computer equipment and storage medium | |
Eldib et al. | Discovering activity patterns in office environment using a network of low-resolution visual sensors | |
CN112233747A (en) | Twin network data analysis method and system based on personal digital | |
CN114373541A (en) | Intelligent traditional Chinese medicine diagnosis method, system and storage medium based on distribution | |
CN112507722B (en) | Method and device for controlling content quality of electronic medical record | |
CN115662562A (en) | Medical record diagnosis and treatment data management method, device, equipment and storage medium | |
CN109671502A (en) | Object portrait update method and device | |
Economou et al. | Parametric proportional odds frailty models | |
CN106326626B (en) | A kind of health sufferer information input system |
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 |