CN109448846A - A kind of analysis method for calculating rare sick disease incidence based on medical insurance big data - Google Patents
A kind of analysis method for calculating rare sick disease incidence based on medical insurance big data Download PDFInfo
- Publication number
- CN109448846A CN109448846A CN201811045882.8A CN201811045882A CN109448846A CN 109448846 A CN109448846 A CN 109448846A CN 201811045882 A CN201811045882 A CN 201811045882A CN 109448846 A CN109448846 A CN 109448846A
- Authority
- CN
- China
- Prior art keywords
- disease
- month
- denominator
- insured
- diagnosis
- 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
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Technology Law (AREA)
- Economics (AREA)
- Pathology (AREA)
- Development Economics (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention discloses a kind of methods that rare sick disease incidence is calculated based on medical insurance big data, are related to medical insurance data process&analysis technology, and calculated disease is thoroughly to cure, once diagnosing, suffer from throughout one's life.By summarizing multiple key parameters of monthly medical insurance data, molecule and denominator information needed for disease incidence calculates are obtained, and then disease incidence is calculated；Required molecule is the kainogenesis case load of target disease in a certain range crowd in specific time；Required denominator is the population of exposure number in specific time, it is possible to the people of target disease occur, and need to exclude illness can not become the people of new cases again in specific time.The disease incidence data of rare disease can be obtained by the method for the invention, promoted the development of rare sick epidemiological study, provided data and technical support for rational clinical guidelines；The Transformation Application for further promoting medical insurance big data fills up the blank of the rare sick epidemiologic data in China.
Description
Technical field
The present invention relates to medical insurance data process&analysis technologies, and in particular to one kind calculates rare disease based on medical insurance big data
The analysis method of disease incidence (Incidence).
Background technique
Rare disease (Rare diseases) refers to the rare disease of the very low one kind of illness rate, disease incidence, and China lacks at present
The weary information, including disease incidence, illness rate of basic epidemiological features etc. to this kind of disease.Medicare data (Claims
It data is usually) in medical insurance chain of command, the data formed by payment information integration, including insured people are substantially special
The information such as sign, diagnosis, treatment, data volume is huge, not only comprehensive, timeliness is good, at low cost, and operability is high, even more comes from
In the longitudinal data of real world, is conducive to development epidemiological study rapidly and efficiently, is especially available with national medical insurance
Data provide new approaches to solve the problems, such as that the rare sick epidemic data in China lacks this.
Unlike other epidemiological studies, calculates disease incidence and need the new cases quantity in clear certain time
Disease population quantity is had suffered from in the regular period.In the analysis method for rare sick disease incidence external at present, Raghu G etc. is learned
Person utilizes U.S. Medicare medical insurance data, idiopathic pulmonary fibrosis disease incidence etc. during calculating 2001-2011, but these are sent out
The calculating of sick rate research be based on the individual initial data in medical insurance data, for the magnanimity medical insurance data in China, no matter data
Phase of storage not, format and the scale of construction or the span of data target, missing and individual remove-insurance etc. with external medical insurance data not
Together, therefore directly the above method can not be used in the epidemiological study of China's medical insurance database；And it makes a general survey of the country and utilizes medical insurance
Data the study found that it is current be mostly focused on by excavate medical insurance data find fraud, improve disease therapeuticing effect with
And modification formulation of accompanying policy etc., rare that medical insurance data is utilized to carry out epidemiology, especially rare sick disease incidence is suffered from
The research of sick rate, so that being difficult to make full use of medical insurance big data at present, the relevant epidemiologic feature for carrying out rare disease is divided
Analysis.
Summary of the invention
In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a kind of based on medical insurance big data measuring and calculating disease incidence
New method, based on optimization data intermediate storage format, by the multiple key parameters for summarizing monthly medical insurance data, comprising: every
Month insured individual is total, monthly increase insured number of individuals newly, monthly diagnosis records sum, monthly diagnosis records diagnosis missing is total,
Neopathy number of cases in regular period, in the regular period number of patients (regular period according to research it needs to be determined that, can be one
Moon, two months, 1 year etc.), the molecule and denominator information needed for disease incidence calculates are obtained, and then disease incidence is calculated.This hair
Bright method, which is related to summarizing, efficiently counts the statistical calculation that disease incidence corresponds to molecule, denominator under data format, be that one kind meets China
The method of the rare sick disease incidence of the calculating of medical insurance data characteristics, can be used for rare sick epidemiological analysis.
The diseases demands that the present invention can calculate cannot be cured thoroughly, once diagnosis, is suffered from throughout one's life.
The principle of the present invention is: the concept based on the man month, and monthly insured individual is total, monthly increases insured individual newly for counting
Number, monthly diagnosis records sum, monthly diagnosis records diagnosis missing sum, combining target disease definition extracting target patient, and
" the stealthy patient " under diagnosis missing at random situation is derived, calculates disease incidence in conjunction with the derivation of equation.What the method for the present invention can be calculated
Disease includes the institutes such as Huppert's disease, plasma cell leukemia, plasma cell dyscrasias, idiopathic pulmonary interstitial fibrosis, POEMS syndrome
Have and meets and cannot thoroughly cure, once diagnosis, suffers from the rare disease of feature throughout one's life.Rare disease can be obtained by the method for the invention
Disease incidence data, provide data and technical support for rational clinical guidelines, further promote medical insurance big data conversion
Using.
Present invention provide the technical scheme that
A method of rare sick disease incidence being calculated based on medical insurance big data, the diseases demands of measuring and calculating are thoroughly to control
More；Based on medical insurance database, multiple key parameters by summarizing monthly medical insurance data (include: monthly insured individual total, every
Increase insured number of individuals, monthly diagnosis records sum, monthly diagnosis records diagnosis missing sum, the neopathy in the regular period moon newly
Number of cases, number of patients in the regular period), the molecule and denominator information needed for disease incidence calculates are obtained, and then hair is calculated
Sick rate；Disease incidence calculate molecule refer in specific time, the kainogenesis case load of target disease, denominator in a certain range of crowd
Population of exposure number i.e. in specific time, it is possible to the people of target disease occur, need to exclude illness can not in specific time
The people of new cases can be become again；
Include the following steps:
A1. medical insurance database range (such as time span, Regional Distribution, outpatient service/be hospitalized) is determined；
A2. the definition of the basic cleaning and target disease of database；
The basic cleaning of database includes following basic step: (1) integrality and logicality of variable are verified in database；
(2) in database content of text code standardized and natural language processing；(3) International Classification of Diseases in database
The version of (International Classification of Diseases, ICD) is determining and uniformly.
In the present invention, the definition of target disease, which is subject in medical insurance database, there is corresponding to the title of disease or ICD coding,
It specifically needs to fully consider a variety of expression-forms of text and ICD coding, and comprehensively includes as far as possible by participle technique building
The dictionary library of target disease diagnosis name expression way.
The process of present invention building dictionary library are as follows:
The text envelope comprising targeted diagnostics disease name (such as Huppert's disease) is extracted first from medical insurance database
Breath, this partial text information may include the disease name expression way of mistake, it is also possible to and it include other diagnosis names, it can not
Directly utilize；
It needs that in the text information for obtaining extraction, it is (multiple to be related to targeted diagnostics disease name using participle technique
Myeloma) field extract；
Then by manually judging whether the expression of these disease names is correct, and correct text representation is determined as just one by one
The dictionary library of step；
All information comprising targeted diagnostics are extracted from medical insurance database again according to preliminary dictionary library, recycles and divides
Word technology is identified；
Repeatedly repeatedly, until the text representation accuracy rate of targeted diagnostics disease name is determined as most up to 95% or more
Whole dictionary library.Purpose is to obtain the expression way comprising all targeted diagnostics disease names as far as possible, so as to subsequent determining trouble
It will not be missed when person.
A3. denominator information summarizes；
It is specifically divided into four groups: insured but never submit an expense account individual, insured and thering is reimbursement to record but do not occur target disease to examine
Disconnected individual, the insured individual for thering is reimbursement to record and the individual of target disease diagnosis occur, have suffered from target disease in certain time.
According to the insured state of every individual in every month, it is included in insured person-time, rejects not insured person-time.
Specifically, according to insured state, person-time at this time is included in if having insured record, if deleted without insured record
It removes.
First group of denominator: the insured but individual never submitted an expense account corresponds to calculation formula such as formula 1 by man month summation:
Wherein, t indicates t-th of month；Insurancet,nFor n-th of group individual t-th of month insured state；N
Represent the man month summation of first group of denominator.
Second group: it is insured and have reimbursement record but do not occur the individual of targeted diagnostics, including three kinds of situations；
The first situation: not because the disease medical man month without reimbursement record is directly brought into denominator, for monthly, i.e., not because of disease
The medical of that month number m without reimbursement record1,1；
Second situation: it should be included in denominator because disease is medical and diagnoses the complete man month and calculate, for monthly, i.e., just because of disease
It examines and diagnoses complete of that month number m1,2；
The third situation: have medical but diagnosis missing man month be considered as it is subsequent fill up, extract because disease is medical but diagnosis lacks
The of that month number m of mistake1,3。
By taking every month as an example, the man month summation of second group of denominator corresponds to calculation formula such as formula 2:
Wherein, t indicates t-th of month；Insurancet,mFor m-th of group individual t-th of month insured state；M
Represent the man month summation of second group of denominator.
Third group: insured to there is reimbursement to record and the individual of targeted diagnostics, including three kinds of situations occur；
The first situation: not because the disease medical man month without reimbursement record is directly brought into denominator, for monthly, i.e., not because of disease
The medical of that month number k without reimbursement record1,1；
Second situation: it should be included in denominator because disease is medical and diagnoses the complete man month and calculate, for monthly, i.e., just because of disease
It examines and diagnoses complete of that month number k1,2；
The third situation: have medical but diagnosis missing man month be considered as it is subsequent fill up, for monthly, i.e., because disease is medical
But diagnose the of that month number k of missing1,3。
By taking every month as an example, the man month summation of third component mother corresponds to calculation formula such as formula 3:
Wherein, t indicates t-th of month；Insurancet,kFor k-th of group individual t-th of month insured state；K
Represent the man month summation for showing third component mother.
4th group: the individual of illness；
4th group of summation corresponds to calculation formula such as formula 4:
Wherein, t1Indicate the regular period；P represents the number of patients summation in the period.
A4. molecular information summarizes, including two groups；
For target disease, corresponding molecular information extracting is carried out, is specifically divided into two groups: is new to send out patient and measuring and calculating be filled up
New hair patient.During new hair patient refers to centainly, the kainogenesis case load of target disease in a certain range crowd；The latter (needs
Fill up the new hair patient of measuring and calculating) diagnosis missing of the measuring and calculating based on diagnosis information with whether suffer from certain rare sick being associated with and be not present
Statistical significance.
First group of molecule: newly sending out patient
The all new hair patients of (for example, can monthly or can be per year) are included in the regular period, are denoted as
Wherein, t1Indicate the regular period；Case_new indicates the number that target disease is newly diagnosed as in the regular period.New hair patient's sentences
Disconnected method is: the patient of targeted diagnostics does not occur before the specific time that research calculates disease incidence, according to study of disease
Difference selects the different elution phases.Such as calculate the disease incidence in certain rare disease a certain year, then within the scope of database this year it
Do not occur targeted diagnostics in the preceding time and is judged as new hair patient.
Second group of molecule: the new hair patient of measuring and calculating need to be filled up
A5. the essential characteristic that molecule newly sends out patient verify with uniformly, such as age, gender, nationality, household register
Medical insurance data are divided into " insurant information table ", " common door (urgency) consultation fee is used and settlement information table " and " outpatient service is big
Disease, outpatient service are planned as a whole, are hospitalized, hospital bed set up at a patient's home expense and settlement information table " 3 lists, by associated variable between each list, to year
Age, gender, nationality, household register etc. need to carry out multi-section verification and uniformly, and corresponding to unique identity to reach each associated variable knows
Other ID (such as identity card), while the information insides one such as each unique identification ID corresponding age, gender, nationality and household register
It causes.
The above-mentioned molecular information summarized and denominator information are asked quotient, calculate disease incidence by the A6. calculating of disease incidence.
The calculation formula of disease incidence Incidence (for calculating 1 year disease incidence)
Wherein, New Case indicates the new hair patient populations in observation year, and the new hair including observing in database is suffered from
Person and the sum of new hair patient of measuring and calculating is filled up, is indicated with ∑ NewCase；PersonYear indicates population of exposure number of the same period,
Refer to the crowd that the disease may occur in the observation year in observation area, is indicated with ∑ Personmonth；∑tCase is every
The sum of the new hair patient of measuring and calculating need to be filled up within a month,T indicates t-th of month,
Caseimpute_mIndicate every month according to insured and there is reimbursement to record but do not occur in the individuals of targeted diagnostics, according to have it is medical but
The target patient number that the number of diagnosis missing is estimated.It indicates newly to be diagnosed as target disease in the regular period
Number, t1Indicate the period.
The denominator sum ∑ of every monthtPersonMonth is calculate by the following formula to obtain:
∑tPersonMonth=∑tPersonMonth1+∑tPersonMonth2+∑tPersonMonth3
Wherein, t indicates t-th of month.∑tPersonMonth1 corresponds to the people that individual that is insured but never submitting an expense account is contributed
Month, ∑tPersonMonth2 is represented insured and is had reimbursement to record but do not occur the man month that the individual of targeted diagnostics is contributed, ∑tPersonMonth3 represent it is insured there is reimbursement to record and the man month that the individual of targeted diagnostics is contributed occur,Indicate the individual of illness in the regular period, t1Indicate the period.
The beneficial effects of the present invention are:
Rare sick disease incidence measuring method provided by through the invention based on medical insurance big data, is related to total amount
On the one hand the statistical calculation for corresponding to molecule, denominator according to disease incidence is efficiently counted under format can obtain the morbidity of the rare disease in China
Rate and Disease Spectrum data (including disease (disease), disability (disability) and premature death (premature
Death) to the pressure of entire society's economy and health), data and technical support are provided for rational clinical guidelines；Another party
Face can promote turning for medical insurance big data the present invention provides the new method that disease incidence calculates is solved using medical insurance data
Change application, fills up the disease incidence data blank of the rare disease in China conscientiously.The method of the present invention is that one kind meets China's medical insurance data spy
The method of the rare sick disease incidence of the calculating of sign, can be used for rare sick epidemiological analysis.
Detailed description of the invention
Fig. 1 is the flow diagram of the method provided by the invention for calculating disease incidence.
Specific embodiment
With reference to the accompanying drawing, the present invention, the model of but do not limit the invention in any way are further described by embodiment
It encloses.
The present invention provides a kind of new method based on medical insurance big data measuring and calculating disease incidence, is deposited among data based on optimizing
Format is stored up, by summarizing multiple key parameters of monthly medical insurance data, molecule and denominator information needed for obtaining disease incidence calculating,
And then disease incidence is calculated；The molecule that disease incidence calculates refers in specific time that target disease is new in a certain range of crowd
Case load, the population of exposure number in denominator, that is, specific time, it is possible to the people of target disease occur, illness need to be excluded and existed occurs
The people of new cases can not be become in specific time again.
It is the process provided by the invention for calculating disease incidence method shown in Fig. 1, a specific embodiment of the invention is as follows:
A1. database range (such as time span, Regional Distribution, outpatient service/be hospitalized) is determined；
A2. the definition of the basic cleaning and target disease of database；
The basic cleaning of database includes following basic step: (1) integrality and logicality of variable are verified in database；
(2) in database content of text code standardized and natural language processing；(3) International Classification of Diseases in database
The version of (International Classification of Diseases, ICD) is determining and uniformly.
In the present invention, the definition of target disease, which is subject in medical insurance database, there is corresponding to the title of disease or ICD coding,
The a variety of expression-forms for specifically needing to fully consider text and ICD coding, construct dictionary library comprehensive as far as possible.
A3. disease incidence corresponds to summarizing for denominator information；
The denominator of disease incidence is specifically divided into four groups
First group: the insured but individual never submitted an expense account
The some patientss are never medical because of disease, only insured record, no reimbursement record, are used only as in disease incidence calculating point
It is female.It specifically needs in the statistical observation time, the insured state (1=is insured, and 0=is not insured) of each observation object in every month,
Then the not insured man month is rejected, the insured man month is added up to and is put into denominator.By taking every month as an example, first group of denominator presses people
Month summation corresponds to calculation formula such as formula 1:
Wherein, t indicates t-th of month；Insurancet,nFor n-th of group individual t-th of month insured state；N
Represent first group of man month summation of denominator.
Second group: it is insured and have reimbursement record but do not occur the individual of targeted diagnostics
The some patientss were once medical because of disease, but did not occurred targeted diagnostics, while possessing insured record, reimbursement record, equally
Denominator is used only as when disease incidence calculates.It specifically needs in the statistical observation time, the insured shape of each observation object in every month
State (1=is insured, and 0=is not insured) then equally rejects not insured man month, but the insured man month cannot be directly placed into point
Mother, but three kinds of situations are divided into according to diagnostic state:
The first situation: not because the disease medical man month without reimbursement record is directly brought into denominator (such as attached drawing 1), to monthly
Speech, i.e., not because of the medical of that month number m without reimbursement record of disease1,1；
Second situation: should be included in denominator and calculate (such as attached drawing 1) because disease is medical and diagnoses the complete man month, to monthly
Speech, i.e., because disease is medical and the complete of that month number m of diagnosis1,2；
The third situation: have medical but diagnosis missing man month be considered as it is subsequent fill up (such as attached drawing 1), extract because disease is medical
But diagnose the of that month number m of missing1,3。
By taking every month as an example, the man month summation of second group of denominator corresponds to calculation formula such as formula 2:
Wherein, t indicates t-th of month；Insurancet,mFor m-th of group individual t-th of month insured state；M
Represent the man month summation of second group of denominator.
Third group: insured to there is reimbursement to record and the individual of targeted diagnostics occur
The some patientss have because disease is medical, and targeted diagnostics occur, while possessing insured record, reimbursement record, are suffering from
Sick rate, disease incidence are used as molecule and denominator when calculating.It for denominator, specifically needs in the statistical observation time, each observation pair
As insured state (1=is insured, and 0=is not insured) in every month, the not insured man month is equally then rejected (such as attached drawing 1),
But the insured man month cannot still be directly placed into denominator, but be divided into three kinds of situations according to diagnostic state:
The first situation: not because the disease medical man month without reimbursement record is directly brought into denominator (such as attached drawing 1), to monthly
Speech, i.e., not because of the medical of that month number k without reimbursement record of disease1,1；
Second situation: should be included in denominator and calculate (such as attached drawing 1) because disease is medical and diagnoses the complete man month, to monthly
Speech, i.e., because disease is medical and the complete of that month number k of diagnosis1,2；；
The third situation: have medical but diagnosis missing man month be considered as it is subsequent fill up (such as attached drawing 1), for monthly,
I.e. because disease is medical but the of that month number k of diagnosis missing1,3。
By taking every month as an example, the man month summation of third component mother corresponds to calculation formula such as formula 3:
Wherein, t indicates t-th of month；Insurancet,kFor k-th of group individual t-th of month insured state；K
Represent the man month summation for showing third component mother.
4th group: the individual of illness
It since the disease that can calculate of the present invention is once diagnose, suffers from all the life, therefore denominator should be that target disease may occur
Crowd, i.e. population of exposure number, therefore need to remove the individual of illness in the regular period when calculating denominator, the 4th group of summation pair
Answer calculation formula such as formula 4:
Wherein, t1Indicate the regular period；P represents the number of patients summation in the period.
A4. molecular information summarizes；
After the definition of target disease, corresponding molecular information extracting is carried out, is specifically divided into two groups:
First group of molecule: newly sending out patient
The all new hair patient being included in the regular period, is denoted as ∑tCase_new, wherein t was indicated in the specific period；
Case_new indicates the number that target disease is newly diagnosed as in the regular period.The judgment method of new hair patient is: calculating in research
There is not the patient of targeted diagnostics before the specific time of disease incidence, different elutions is selected according to the difference of study of disease
Phase.Such as calculate the disease incidence in certain rare disease a certain year, then do not occur in the time within the scope of database before this year
Targeted diagnostics are judged as new hair patient.
Second group of molecule: the new hair patient of measuring and calculating need to be filled up
There is diagnosis missing in part diagnosis records, including insured and medical but diagnosis missing non-targeted Disease and ginseng
Protect and go to a doctor but diagnose the target disease patient of missing, i.e. m1,3And k1,3.Due to calculate disease incidence require patient be newly send out, namely
There are targeted diagnostics for the first time, so for k1,3Although the of that month diagnosis missing of patient, because being diagnosed as target before
Patient is not considered new hair patient, therefore is not included in molecule.Therefore, the part for needing exist for filling up measuring and calculating is m1,3, part record
It should not directly reject, as shown in table 1,
Molecule fills up schematic diagram when 1 disease incidence of table calculates
The calculation formula of disease incidence Incidence ideally is formula 5:
Wherein, a represents the target disease that can be grabbed in individual that is insured and having the diagnosis of medical and diagnosis records not lack
Patient's number, b represent the non-targeted disease that can be grabbed in individual that is insured and having the diagnosis of medical and diagnosis records not lack
Patient's number, c represent the target disease that can theoretically grab in individual that is insured and having medical and diagnosis records diagnosis missing
Patient's number, d represent the non-targeted disease that can be grabbed in individual that is insured and having medical and diagnosis records diagnosis missing
Patient numbers, e represent insured but never medical individual number.
And the disease incidence calculation formula after directly rejecting is expressed as formula 6:
Disease incidence ideally is obviously differed with both disease incidence after directly rejecting.Therefore, it is necessary to lack to diagnosis
Part is lost, based on certain it is assumed that suitably being estimated, to obtain the numerical value of c and d.The original hypothesis that the present invention uses is medical
Information diagnosis missing with whether suffer from certain it is rare disease be associated be not present statistical significance, i.e.,If meeting this
It is assumed that then c is expressed as formula 7:
Wherein, c+d is the total number of records of diagnosis missing, can directly be counted to get.
The number for needing to fill up in molecule is calculated according to formula 7.Note that the new hair patient diagnosis for needing to fill up measuring and calculating lacks
Lose part are as follows: insured and there is reimbursement to record but do not occur of that month in the individuals of targeted diagnostics having medical but diagnosis missing number
m1,3, filled up according to above-mentioned hypothesis, Caseimpute_m=c, the target disease after filling up newly send out patient numbers' sum and areWherein, t indicates t-th of month, Caseimpute_mIndicate every month in, according to it is insured and have reimbursement record
But does not occur the of that month target patient for thering is medical but diagnosis missing number to be estimated in the individual of targeted diagnostics and newly send out number.
A5. molecule patient essential characteristic verify with uniformly
Medical insurance data are divided into " insurant information table ", " common door (urgency) consultation fee is used and settlement information table " and " outpatient service is big
Disease, outpatient service are planned as a whole, are hospitalized, hospital bed set up at a patient's home expense and settlement information table " 3 lists, by associated variable between each list, to year
Age, gender, nationality, household register etc. need to carry out multi-section verification and uniformly, and corresponding to unique identity to reach each associated variable knows
Other ID (such as identity card), while the information insides one such as each unique identification ID corresponding age, gender, nationality and household register
It causes.
A6. the calculating of disease incidence
Disease incidence calculation formula (as unit of man-year)
Wherein, ∑ NewCase indicates the new hair patient populations in observation year, and the new hair including observing in database is suffered from
Person and fill up the sum of new hair patient of measuring and calculating；PersonYear indicates population of exposure number of the same period, that is, referring to should in the observation year
The crowd of the disease may occur in observation area；∑tCase is that need to fill up the sum of new hair patient of measuring and calculating every month, T indicates t-th of month, Caseimpute_mIndicate every month according to it is insured and have reimbursement remember
In the individual for recording but not occurring targeted diagnostics, according to the target patient number for thering is medical but diagnosis missing number to be estimated.Indicate the number that target disease is newly diagnosed as in the regular period, t1Indicate the period.
The denominator sum ∑ of every monthtPersonMonth is calculate by the following formula to obtain:
∑tPersonMoth=∑tPersonMonth1+∑tPersonMonth2+∑tPersonMonth3
Wherein, t indicates t-th of month.∑tPersonMonth1 corresponds to the people that individual that is insured but never submitting an expense account is contributed
Month, ∑tPersonMonth2 is represented insured and is had reimbursement to record but do not occur the man month that the individual of targeted diagnostics is contributed, ∑tPersonMonth3 represent it is insured there is reimbursement to record and the man month that the individual of targeted diagnostics is contributed occur,Indicate the individual of illness in the regular period, t1Indicate the period.
Below by example, the present invention will be further described.
This example chooses the insured library of certain province urban employees and the insured library of town dweller of 2012-2016, is within selection 4 years
The elution phase, then the illness rate that Huppert's disease is calculated based on the medical insurance database for saving 2016 is calculated, database includes
Urban employees' insurant (217,342,112 people) in 2016, town dweller's insurant (145,714,765 people).
(date, medical date variable missing, exception are such as submitted an expense account) after completing basic data cleansing, Huppert's disease
Clinical diagnostic modalities combination text, ICD coding statement is such as table 2:
The diagnosis description of 2 Huppert's disease of table and ICD coding are enumerated
Field name in database comprising diagnostic message shares 6, is Main Diagnosis title, Main Diagnosis volume respectively
Code wants diagnosis name for the first time, wants diagnosis coding for the first time, wants diagnosis name for the second time and want diagnosis coding for the second time, then real
Border definition is defined as follows according to field structure in database:
Above-mentioned each field must include field value (being "or" relationship between field): myeloma, card are strangled, and bone marrow cancer/
Marrow ca, myelopathy, C90, M9732,203.0；
Above-mentioned each field must exclude field value (being "or" relationship between field): thick liquid cell, isolatism, C90.1,
C90.2
When specific molecule crawl, for Main Diagnosis, for the first time to diagnose, to diagnose for second, Main Diagnosis coding, the
One secondary diagnosis coding and want diagnosis coding totally six fields expansion for the second time, it is desirable that all in six fields, it is necessary to have to
It is one " must include field " few, but " field must be excluded " must not contained.
Then it carries out denominator information to summarize, wherein monthly insured individual sum, monthly diagnosis records sum and monthly
It is as shown in table 3 to examine the distribution of record diagnosis missing sum in every month:
3 molecule denominator of table calculates the medical insurance data parameters summary sheet for needing to use
It should be noted that monthly it is insured individual sum=it is insured but never submit an expense account individual+it is insured and have reimbursement record
The individual of targeted diagnostics+insured, which occurs, in unit has reimbursement to record and the individual of targeted diagnostics occur, therefore only lists medical insurance here
Monthly insured individual is total for key parameter in data.
One, the calculating of denominator
The calculating of denominator is divided into two parts: the number of individuals of average annual insured numbers in 2016 and illness before in 2016.
First part's average annual insured number in 2016 uses the method using 12 monthly average, obtains: (30152092+
28539556+30571984+30615202+28779370+30912654+31344596+28196530+30464440+
30734528+32040068
+ 30705856)/12=30254740
The number of individuals of illness before second part 2016 obtains by the following method: by the number of province 2012-2016
It is divided into 2012-2015 and 2016 year two parts according to library, each section all guarantees that each individual only has an idagnostic logout, will
Two databases carry out horizontal meaders with software, matched part be had in 2016 diagnosis records and before 2016 by
It is diagnosed as patient's number of Huppert's disease, obtaining the number of individuals for saving 2016 Nian Qianyi illness by this method is 3821 people.
Two, the calculating of molecule
The calculating of molecule is divided into two parts: new hair patient's number and the patient that need to fill up measuring and calculating
Patient's number is newly sent out by first part, is obtained using method same as illness number of individuals, it may be assumed that by province 2012-
Database in 2016 is divided into 2012-2015 and 2016 year two parts, and each section all guarantees each individual only one
Two databases are carried out horizontal meaders with software, only occurred in database in 2016 in matching result by idagnostic logout
Part is the new hair patient number in 2016 Huppert's diseases, this obtained by this method save 2016 Nian Xinfa individual patients numbers be
417 people.
Patient's number that second part need to be filled up calculates following (this example is calculated as unit of year need to fill up patient's number), first calculates
Average annual diagnosis records number in 2016 out are as follows:
(5266439+4685125+5446965+4767414+5356961+4962207+5627907+4545582+
4700331+5365805+5684190+5739000)/12=5178994
Average annual diagnosis records diagnose missing number within 2016 are as follows:
(331479+314970+333140+315726+300708+271454+274475+282899+273228+
284421+277799+280500)/12=295067
Then according to new 417 people of hair patient in 2016, the patient's number that need to be filled up is calculated are as follows: (417*295067)/
(5178994-295067)=25
Then disease incidence calculates as follows:
It should be noted that the purpose for publicizing and implementing example is to help to further understand the present invention, but the skill of this field
Art personnel, which are understood that, not to be departed from the present invention and spirit and scope of the appended claims, and various substitutions and modifications are all
It is possible.Therefore, the present invention should not be limited to embodiment disclosure of that, and the scope of protection of present invention is with claim
Subject to the range that book defines.
Claims (6)
1. a kind of analysis method for calculating rare sick disease incidence based on medical insurance big data, characterized in that calculated disease is cannot
Thoroughly cure, once diagnosis, is suffered from throughout one's life；Based on medical insurance database, by summarizing multiple key parameters of monthly medical insurance data,
The molecule and denominator information needed for disease incidence calculates are obtained, and then disease incidence is calculated；
Multiple key parameters include: that monthly insured individual is total, monthly increase insured number of individuals newly, monthly diagnosis records sum, every
Month diagnosis records diagnosis missing sum, the neopathy number of cases in the regular period, number of patients in the regular period；
Disease incidence calculates the kainogenesis case load that required molecule is target disease in a certain range of crowd in specific time；
Denominator needed for disease incidence calculates is the population of exposure number in specific time, it is possible to the people of target disease occur, and need
The people of new cases can not be become again in specific time by excluding illness；
Include the following steps:
A1. medical insurance database range is determined, comprising: time span, Regional Distribution, outpatient service/be hospitalized；
A2. the basic cleaning of database defines target disease and constructs target disease dictionary library；
Database it is basic cleaning include:
A21) in verification database variable integrality and logicality；
A22) in database content of text code standardized and natural language processing；
A23) the version of the International Classification of Diseases ICD coding in determining and unified database；
The definition of target disease, which is subject in medical insurance database, there is corresponding to the title of disease or ICD coding, including text data
With the diversified forms of ICD coding；And pass through dictionary library of the participle technique building comprising the expression of target disease diagnosis name；
A3. summarize denominator information；It is specifically divided into four groups:
First group of denominator information: the insured but individual never submitted an expense account；
Second group of denominator information: it is insured and have reimbursement record but do not occur target disease diagnosis individual；
Denominator information third group: the insured individual for thering is reimbursement to record and target disease diagnosis occur；
The 4th group of denominator information: the individual of target disease is had suffered from certain time；
According to the insured state of every individual in every month, it is included in insured person-time if having insured record, is otherwise not insured people
It is secondary, it is rejected；
First group of denominator information is calculated especially by formula 1 by man month summation correspondence:
Wherein, t indicates t-th of month；Insurancet,nFor n-th of group individual t-th of month insured state；N is represented
First group of man month summation of denominator；
Second group of denominator information includes three kinds of situations；
Second group of the first situation of denominator information: not because the disease medical man month without reimbursement record is directly brought into denominator, to monthly
Speech,
I.e. not because of the medical of that month number m without reimbursement record of disease1,1；
Second group of second situation of denominator information: should be included in denominator and calculate because disease is medical and diagnoses the complete man month, to monthly
Speech,
I.e. because disease is medical and the complete of that month number m of diagnosis1,2；
Second group of the third situation of denominator information: have medical but diagnosis missing man month be considered as it is subsequent fill up, extract because of disease just
Examine but diagnose the of that month number m of missing1,3。
Second group of denominator information is calculated especially by formula 2 by man month summation:
Wherein, t indicates t-th of month；Insurancet,mFor m-th of group individual t-th of month insured state；M is represented
The man month summation of second group of denominator；
Denominator information third group includes three kinds of situations；
The first situation of denominator information third group: not because the disease medical man month without reimbursement record is directly brought into denominator, to monthly
Speech,
I.e. not because of the medical of that month number k without reimbursement record of disease1,1；
Denominator information third group second situation: should be included in denominator and calculate because disease is medical and diagnoses the complete man month, to monthly
Speech,
I.e. because disease is medical and the complete of that month number k of diagnosis1,2；
The third situation of denominator information third group: have medical but diagnosis missing man month be considered as it is subsequent fill up, for monthly,
I.e. because disease is medical but the of that month number k of diagnosis missing1,3；
Denominator information third group is calculated especially by formula 3 by man month summation:
Wherein, t indicates t-th of month；Insurancet,kFor k-th of group individual t-th of month insured state；K is represented
Show the man month summation of third component mother；
The 4th group of denominator information passes through the summation that the individual of illness is calculated in formula 4:
Wherein, t1It indicates in the regular period；P represents the number of patients summation in the period；
A4. summarize molecular information, including two groups: new hair patient and the new hair patient that measuring and calculating need to be filled up；
New hair patient be it is certain during in a certain range crowd target disease kainogenesis case load；The calculating side of new hair patient
Method is: selecting the different elution phases according to the difference of disease, target does not occur before the specific time for calculating disease incidence and examine
Disconnected patient, by regular period t1Interior all new hair patient is denoted asThe new hair patient that measuring and calculating need to be filled up surveys
Calculate based on diagnosis information diagnosis missing with whether suffer from certain it is rare disease be associated be not present statistical significance, be denoted as ∑tCase；
A5. to the essential characteristic of the new hair patient in molecular information verify with uniformly so that the data information of separate sources
Unanimously；
A6. it calculates disease incidence: the molecular information summarized and denominator information being asked into quotient, disease incidence is calculated.
Through the above steps, it realizes and disease incidence is calculated based on medical insurance big data.
2. the method as described in claim 1 based on medical insurance big data measuring and calculating disease incidence, characterized in that especially by formula 8
1 year disease incidence Incidence is calculated:
Wherein, New Case indicates the new hair patient populations in observation year, including the new hair patient that is observed in database and
The sum of the new hair patient for filling up measuring and calculating, is indicated with ∑ NewCase；PersonYear indicates population of exposure number of the same period, that is, refers to
The crowd that the disease may occur in the observation year in observation area, is indicated with ∑ Personmonth；∑tCase is every month
The sum of the new hair patient of measuring and calculating, ∑ need to be filled uptCase=∑tCaseimpute_m, t t-th of month of expression, Caseimpute_mTable
Show every month according to insured and there is reimbursement to record but do not occur in the individuals of targeted diagnostics, according to there is medical but diagnosis missing people
Several estimated target patient numbers；Indicate regular period t1The interior number for being newly diagnosed as target disease；
The denominator sum ∑ of every monthtPersonMonth is calculate by the following formula to obtain:
∑tPersonMonth=∑tPersonMonth1+∑tPersonMonth2+∑tPersonMonth3
Wherein, t indicates t-th of month；∑tPersonMonth1 corresponds to the man month that individual that is insured but never submitting an expense account is contributed,
∑tPersonMonth2 is represented insured and is had reimbursement to record but do not occur the man month that the individual of targeted diagnostics is contributed, ∑tPersonMonth3 represent it is insured have reimbursement record and there is the man month that the individual number of targeted diagnostics is contributed；Indicate regular period t1The individual number of inside illness.
3. the method as described in claim 1 based on medical insurance big data measuring and calculating disease incidence, characterized in that step A23) building
The process of target disease dictionary library specifically includes:
A231 the text information comprising targeted diagnostics disease name) is extracted first from medical insurance database；
A232 participle technique) is used, from extracting in obtained text information, the field for being involved in targeted diagnostics disease name is mentioned
It takes out；
A233) by manual type, judge whether disease name expression is correct one by one；Correct text representation is formed as preliminary
Dictionary library；
A234 all letters comprising targeted diagnostics disease name) are extracted from medical insurance database according to preliminary dictionary library again
Breath, then operation A232 is executed repeatedly)~A233) identified, until the text representation accuracy rate of targeted diagnostics disease name reaches
To 95% or more, i.e. building obtains target disease dictionary library.
4. the method for rare sick disease incidence is calculated based on medical insurance big data as claimed in claim 3, characterized in that targeted diagnostics disease
Name of disease is known as Huppert's disease.
5. the method as described in claim 1 based on medical insurance big data measuring and calculating disease incidence, characterized in that summarize molecular information
In, the number for needing to fill up in molecule is calculated especially by formula 7 for the measuring and calculating to the new hair patient that need to fill up measuring and calculating:
Wherein, a represents the target disease that can be grabbed in individual that is insured and having the diagnosis of medical and diagnosis records not lack and suffers from
Person's number；B represents the non-targeted disease that can be grabbed in individual that is insured and having the diagnosis of medical and diagnosis records not lack and suffers from
Person's number；C represents the target disease that can theoretically grab in individual that is insured and having medical and diagnosis records diagnosis missing and suffers from
Person's number；D represents the non-targeted Disease that can be grabbed in individual that is insured and having medical and diagnosis records diagnosis missing
Number；C+d is the total number of records of diagnosis missing.
6. the method as described in claim 1 based on medical insurance big data measuring and calculating disease incidence, characterized in that calculated disease packet
Include Huppert's disease, plasma cell leukemia, plasma cell dyscrasias, idiopathic pulmonary interstitial fibrosis, POEMS syndrome.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811045882.8A CN109448846B (en) | 2018-09-07 | 2018-09-07 | Analysis method for measuring and calculating rare disease incidence based on medical insurance big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811045882.8A CN109448846B (en) | 2018-09-07 | 2018-09-07 | Analysis method for measuring and calculating rare disease incidence based on medical insurance big data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109448846A true CN109448846A (en) | 2019-03-08 |
| CN109448846B CN109448846B (en) | 2021-04-30 |
Family
ID=65530399
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811045882.8A Active CN109448846B (en) | 2018-09-07 | 2018-09-07 | Analysis method for measuring and calculating rare disease incidence based on medical insurance big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109448846B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112967817A (en) * | 2021-02-02 | 2021-06-15 | 武汉大学 | Epidemiological research population screening method based on medical big data and storage medium |
| CN113488123A (en) * | 2021-04-21 | 2021-10-08 | 广州医科大学附属第一医院 | Method for establishing COVID-19 triage system based on diagnosis time efficiency, system and triage method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678104A (en) * | 2016-04-06 | 2016-06-15 | 电子科技大学成都研究院 | Method for analyzing health data of old people on basis of Cox regression model |
| WO2017152639A1 (en) * | 2016-03-10 | 2017-09-14 | 深圳市前海安测信息技术有限公司 | Medical insurance actuarial system and method based on medical big data |
| CN108198595A (en) * | 2018-01-18 | 2018-06-22 | 北京化工大学 | A kind of multi-source heterogeneous unstructured medical record data fusion method |
-
2018
- 2018-09-07 CN CN201811045882.8A patent/CN109448846B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017152639A1 (en) * | 2016-03-10 | 2017-09-14 | 深圳市前海安测信息技术有限公司 | Medical insurance actuarial system and method based on medical big data |
| CN105678104A (en) * | 2016-04-06 | 2016-06-15 | 电子科技大学成都研究院 | Method for analyzing health data of old people on basis of Cox regression model |
| CN108198595A (en) * | 2018-01-18 | 2018-06-22 | 北京化工大学 | A kind of multi-source heterogeneous unstructured medical record data fusion method |
Non-Patent Citations (2)
| Title |
|---|
| MORAN KI等: "Incidence, prevalence and complications of Budd–Chiari syndrome in South Korea: a nationwide, population‐based study", 《LIVER INTERNATIONAL》 * |
| 董选军: "镇海区百年一遇伤寒发病率测算", 《中国公共卫生管理》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112967817A (en) * | 2021-02-02 | 2021-06-15 | 武汉大学 | Epidemiological research population screening method based on medical big data and storage medium |
| CN112967817B (en) * | 2021-02-02 | 2022-06-10 | 武汉大学 | Epidemiological research population screening method based on medical big data and storage medium |
| CN113488123A (en) * | 2021-04-21 | 2021-10-08 | 广州医科大学附属第一医院 | Method for establishing COVID-19 triage system based on diagnosis time efficiency, system and triage method |
| CN113488123B (en) * | 2021-04-21 | 2023-07-18 | 广州医科大学附属第一医院 | Method for establishing diagnosis time-effect-based COVID-19 triage system, system and triage method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109448846B (en) | 2021-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Waghade et al. | A comprehensive study of healthcare fraud detection based on machine learning | |
| AU2011319965B2 (en) | System and method for machine based medical diagnostic code identification, accumulation, analysis and automatic claim process adjudication | |
| Tangcharoensathien et al. | Knowledge-based changes to health systems: the Thai experience in policy development | |
| US10319466B2 (en) | Intelligent filtering of health-related information | |
| Panda et al. | Big data in health care: A mobile based solution | |
| Konrad et al. | Key considerations when using health insurance claims data in advanced data analyses: an experience report | |
| CN110766332A (en) | Comprehensive quality index monitoring system for regional sanitation | |
| CN109448846A (en) | A kind of analysis method for calculating rare sick disease incidence based on medical insurance big data | |
| Lewer et al. | Data resource: the Kent integrated dataset (KID) | |
| Hristovski et al. | Using data warehousing and OLAP in public health care. | |
| Abdelfattah | Neutrosophic data envelopment analysis: An application to regional hospitals in Tunisia | |
| CN108630320A (en) | A method of disease illness rate is calculated based on medical insurance big data | |
| CN110827990B (en) | Typhoid fever syndrome differentiation reasoning system based on knowledge graph | |
| CN112992366A (en) | ICD (identity control document) code artificial intelligence audit quality control mode and system based on medical insurance disease payment system | |
| CN108320804A (en) | The method for realizing disease complexity quantitatively evaluating based on disease fault network complexity evaluation system | |
| Alterman et al. | State-level treatment outcome studies using administrative databases | |
| CN108198622A (en) | The system that disease complexity quantitatively evaluating is realized based on computer software | |
| CN110364265B (en) | Data value generation and implementation method based on health data bank | |
| Fowler et al. | Change in healthcare utilisation after surgical treatment: observational study of routinely collected patient data from primary and secondary care | |
| KR20210126408A (en) | Device and method for ai calculating damage using disease big data | |
| Parker et al. | Data quality: How the flow of data influences data quality in a small to medium medical practice | |
| Alghamdi | Health data warehouses: reviewing advanced solutions for medical knowledge discovery | |
| Liu et al. | Empirical study of data completeness in electronic health records in China | |
| CN110766004B (en) | Medical identification data processing method and device, electronic equipment and readable medium | |
| Barrett et al. | A description of an intermediate care service using routinely collected data |
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 |