CN109934719A - The detection method and detection device of medical insurance unlawful practice, medical insurance control charge system - Google Patents
The detection method and detection device of medical insurance unlawful practice, medical insurance control charge system Download PDFInfo
- Publication number
- CN109934719A CN109934719A CN201711367590.1A CN201711367590A CN109934719A CN 109934719 A CN109934719 A CN 109934719A CN 201711367590 A CN201711367590 A CN 201711367590A CN 109934719 A CN109934719 A CN 109934719A
- Authority
- CN
- China
- Prior art keywords
- data
- medical
- medical insurance
- detected
- module
- 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.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 67
- 239000013598 vector Substances 0.000 claims abstract description 29
- 230000006399 behavior Effects 0.000 claims abstract description 17
- 238000010801 machine learning Methods 0.000 claims abstract description 12
- 238000013135 deep learning Methods 0.000 claims abstract description 10
- 238000000605 extraction Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 27
- 239000003814 drug Substances 0.000 claims description 16
- 238000000034 method Methods 0.000 claims description 15
- 201000010099 disease Diseases 0.000 claims description 14
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 14
- 238000012795 verification Methods 0.000 claims description 13
- 238000003745 diagnosis Methods 0.000 claims description 11
- 229940079593 drug Drugs 0.000 claims description 9
- 238000013500 data storage Methods 0.000 claims description 8
- 238000007689 inspection Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 239000000126 substance Substances 0.000 claims description 5
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000000586 desensitisation Methods 0.000 description 4
- 230000003542 behavioural effect Effects 0.000 description 2
- 238000003759 clinical diagnosis Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000012946 outsourcing Methods 0.000 description 1
- 230000037361 pathway Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001225 therapeutic effect Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Landscapes
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The present invention provides a kind of detection method of medical insurance unlawful practice and detection devices, medical insurance control charge system, are related to medical insurance control expense technical field, can be improved accuracy, the timeliness, validity of medical insurance unlawful practice detection.Wherein the detection method includes: to obtain medical insurance data and its corresponding medical data, and medical data is described from multiple angles；Feature extraction is carried out to medical data and medical insurance data using machine learning algorithm, obtains feature vector；Feature vector and medical insurance data are trained using deep learning algorithm, generate multidimensional characteristic model；Search medical data corresponding to medical insurance data to be detected, medical insurance data to be detected and its corresponding medical data are inputted into multidimensional characteristic model, type belonging to the medical insurance behavior that medical insurance data to be detected are included is detected, the type includes normal, doubtful and violation.Above-mentioned detection method is applied to during medical insurance control expense, to examine the compliance of medical insurance reimbursement behavior.
Description
Technical field
The present invention relates to medical insurance control expense technical field more particularly to the detection methods and detection dress of a kind of medical insurance unlawful practice
It sets, medical insurance control charge system.
Background technique
With the rapid growth of medical expense, there is reality unable to make ends meet in medical insurance fund, thus medical insurance control expense at
For the most important thing instantly.Medical insurance control expense can rely on medical insurance control charge system to carry out, and traditional medical insurance control charge system majority uses medical insurance
Rule base judgment model carries out compliance analysis to medical insurance data according to medical insurance rule base, detects medical insurance unlawful practice.But
It is that the mathematical logic of medical insurance rule base is not related to the analysis to medical procedure compliance, this directly results in medical insurance rule base and verifies institute
Limited fund is saved, and can not in time, accurately and efficiently detect the medical insurance unlawful practice in medical procedure.
In view of the above-mentioned problems, major part is according to Expert Rules such as medicinal knowledge libraries to the doctor in medical procedure in the prior art
Unlawful practice is protected to be detected.And actually due to the complexity of the multi-source heterogeneous characteristic of data and medical act itself,
It is difficult just to detected all medical insurance unlawful practices with some specific rules；And with currently per capita charge, pre-payment,
The execution of the controls expense mode such as DRGs-based payment system, medical insurance unlawful practice can quickly change, these factors lead to existing doctor
Protect the failure and hysteresis quality of control charge system and the detection method to medical insurance unlawful practice.
Summary of the invention
For above-mentioned the problems of the prior art, the embodiment of the present invention provide a kind of medical insurance unlawful practice detection method and
Detection device, medical insurance control charge system, to improve accuracy, the timeliness, validity of medical insurance unlawful practice detection.
In order to achieve the above objectives, the embodiment of the present invention adopts the following technical scheme that
In a first aspect, the embodiment of the invention provides a kind of detection method of medical insurance unlawful practice, which includes
Following steps: S1: medical insurance data and its corresponding medical data are obtained, the medical data is retouched from multiple angles
It states；S2: feature extraction is carried out to the medical data and the medical insurance data using machine learning algorithm, obtains feature vector；
S3: being trained described eigenvector and the medical insurance data using deep learning algorithm, generates multidimensional characteristic model；S4:
Medical data corresponding to medical insurance data to be detected is searched, the medical insurance data to be detected and its corresponding medical data are defeated
Enter the multidimensional characteristic model, detects type belonging to the medical insurance behavior that the medical insurance data to be detected are included；Wherein, described
Type includes normal, doubtful and violation.
In the detection method of above-mentioned medical insurance unlawful practice, medical number is described by introducing medical data, and from multi-angle
According to so that data source based on detection is more comprehensive, and use machine learning and deep learning algorithm instruct data
Practice modeling, to realize detection to medical insurance unlawful practice, thus improve the accuracy of medical insurance unlawful practice detection, timeliness,
Validity helps to improve the effect of medical insurance control expense.
Second aspect, the embodiment of the invention provides a kind of detection device of medical insurance unlawful practice, which includes:
Medical insurance data memory module, for storing medical insurance data.Medical data storage module, for storing corresponding to the medical insurance data
Medical data, and the medical data is the data being described from multiple angles.With the medical insurance data memory module and
The connected characteristic extracting module of the medical data storage module, for using machine learning algorithm to the medical data and institute
It states medical insurance data and carries out feature extraction, obtain feature vector.With the characteristic extracting module and the medical insurance data memory module
Connected training module is generated for being trained using deep learning algorithm to described eigenvector and the medical insurance data
Multidimensional characteristic model.The detection module being connected with the training module and the medical insurance data memory module, it is to be checked for searching
Medical data corresponding to medical insurance data is surveyed, the medical insurance data to be detected and its corresponding medical data input are described more
Dimensional feature model detects type belonging to the medical insurance behavior that the medical insurance data to be detected are included；Wherein, the type includes
Normally, doubtful and violation.
Medical insurance described in the beneficial effect and first aspect that the detection device of above-mentioned medical insurance unlawful practice can be generated is in violation of rules and regulations
The beneficial effect of the detection method of behavior is identical, and details are not described herein again.
The third aspect, the embodiment of the invention also provides a kind of medical insurance control charge system, which includes such as the
Detection device described in two aspects.
The detection of medical insurance unlawful practice described in the beneficial effect and second aspect that above-mentioned medical insurance control charge system can be generated
The beneficial effect of device is identical, and details are not described herein again.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the basic flow chart of the detection method of medical insurance unlawful practice provided by the embodiment of the present invention；
Fig. 2 is the basic block diagram of the detection device of medical insurance unlawful practice provided by the embodiment of the present invention.
Description of symbols:
1- medical insurance data memory module；2- medical data storage module；
3- characteristic extracting module；4- training module；
5- detection module；6- manual examination and verification module.
Specific embodiment
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, implement below in conjunction with the present invention
Attached drawing in example, technical scheme in the embodiment of the invention is clearly and completely described.Obviously, described embodiment
Only a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, the common skill in this field
Art personnel all other embodiment obtained without creative labor belongs to the model that the present invention protects
It encloses.
As described in background, in the prior art mainly according to Expert Rules such as medicinal knowledge libraries in medical procedure
Medical insurance unlawful practice is detected, but the accuracy of this detection mode, timeliness, validity are poor.For this problem, originally
The inventor of invention proposes: by introducing medical data, medical data is described from multi-angle, to improve data based on detection
Source it is comprehensive, feature extraction then is carried out to medical insurance data and medical data using machine learning algorithm, utilizes deep learning
Algorithm is trained modeling to feature vector and medical insurance data, realizes the detection to medical insurance unlawful practice, to improve medical insurance in violation of rules and regulations
Accuracy, timeliness, the validity of behavioral value.
The above are the designs of the basic fundamental of the embodiment of the present invention, are conceived based on the basic fundamental, implement below to the present invention
The technical solution of example describes in detail.
A kind of detection method of medical insurance unlawful practice is present embodiments provided, referring to Figure 1, which includes following
Step:
S1: obtaining medical insurance data and its corresponding medical data, carries out from multiple angles to acquired medical data
Description.
Three kinds of fraud, waste, abuse problems are primarily present in medical procedure, to the medical insurance unlawful practice in medical procedure
Detection essentially consists in how to carry out true, in due course, appropriate, suitable description to medical act, therefore from multiple angles in step S1
Medical data is described in degree, in terms of this mainly includes following four:
(1) from medical resource consumption angle, using medical diagnosis on disease associated packets (Diagnosis Related Groups,
Abbreviation DRGs) mode, medical data is described.
Wherein, so-called " medical diagnosis on disease associated packets " refer to according to the age of patient, gender, length of stay, clinical diagnosis,
It illness, operation, disease severity, complication and complication and the factors such as lapses to, patient is divided into 500~600 diagnosis phases
Then pass group determines how much compensate to hospital.
(2) from medical behavioral agent angle, by hospital, the essential information and relationship of doctor, patient three, to medical number
According to being described.
(3) rule angle is closed from medical act, it is right from directions such as clinical path (Clinical pathway), Expert Rules
Medical data is described.
Wherein, so-called " clinical path ", which refers to, establishes a set of standardized therapeutic mode and treatment procedure for a certain disease,
It is the aggregative model in relation to clinical treatment, promotes to treat tissue and disease pipe so that evidence-based medical and guide are guidance
The method of reason.Expert Rules include the contents such as medicinal knowledge library.
(4) from the data direction of multiclass different characteristics, such as discrete data, continuous data, time series data direction, to medical treatment
Data are described.
It should be noted that the medical data being described refers to that medical insurance reimbursement application (medical insurance data) is right in this step
Medical data after the desensitization answered, to guarantee the authenticity and comprehensive description of medical data.So-called " desensitization " refers to certain quick
Feel information by desensitization rule carry out data deformation, realize privacy-sensitive data reliably protecting, thus can exploitation,
Safely using the real data set after desensitization in test and other nonproductive environment and Outsourcing Environment.
S2: the medical data using machine learning algorithm to acquired medical insurance data and by multi-angle description carries out special
Sign is extracted, and feature vector is obtained.
In above-mentioned steps S2, so-called " machine learning algorithm ", which refers to, specializes in the mankind are simulated or realized to computer how
Learning behavior reorganize the existing structure of knowledge to obtain new knowledge or skills and be allowed to constantly improve the performance of itself.
Above-mentioned steps S2 specifically may include following three aspects:
(1) according to medical diagnosis on disease associated packets, according to the age of patient, gender, length of stay, clinical diagnosis, illness, hand
It art, disease severity, complication and complication and the factors such as lapses to, patient is divided into diagnosis relevant group, for each diagnosis correlation
Group provides the medical insurance section under the conditions of same medical resource.
(2) by hospital, the essential information and relationship of doctor, patient three, medical data is analyzed, and uses a variety of machines
Learning algorithm carries out data normalization to medical data.
Wherein, a variety of machine learning algorithms concretely a variety of unsupervised machine learning algorithms.
It is specific that data normalization is carried out to medical data can include:
(i) statistical distribution mode is used, by expense, length of stay data normalization to [0,1] section.Detailed process are as follows:
Using 5 quartiles as initial minimum, using 95 quartiles as original maximum, less than 5 quartiles and greatly in expense, length of stay data
1 is set in the data of 95 quartiles, in expense, length of stay data >=data of 5 quartiles of quartile Qie≤95 set 0.
(ii) by clustering algorithm, to Western medicine medication record, check that chemical examination recording gauge calculates data record Euclidean distance, then
It is normalized.Detailed process are as follows: with International Classification of Diseases (International Classification of
Diseases, abbreviation ICD) it is group basis, Western medicine medication record, inspection chemical examination record are clustered in same disease, obtained
To Western medicine types of medicines, check test item；Western medicine types of medicines in drug white list is labeled as 1 (as outlier), no
Then it is labeled as 0 (as positive constant)；The inspection test item in chemical examination white list is being checked labeled as 1, otherwise labeled as 0.
(3) it is inputted Expert Rules as feature, obtain Expert Rules returns the result data, this is returned the result data
It is divided into the data of multiclass different characteristics, such as: then Various types of data is normalized to by two-value type data, continuous data respectively
[0,1] section.Wherein, the data that doubtful degree is bigger in data are returned the result, score value is higher.
S3: being trained feature vector obtained and acquired medical insurance data using deep learning algorithm, generates
Multidimensional characteristic model.
Above-mentioned steps S3 specifically includes following procedure:
Acquired medical insurance data, the medical data by multi-angle description and feature vector obtained are spelled
It connects, forms the training data of N-dimensional.Wherein, aforementioned medical data is N1 dimension data, and aforementioned medical insurance data are N2 dimension data, aforementioned
Feature vector is N3 dimension data, N-dimensional=N1 dimension+N2 dimension+N3 dimension.
Value is carried out on the training data for be formed by N-dimensional with the vector that M is tieed up, default step-length is set as 1, generates 2M+1
The vector of a M dimension, the vector of 2M+1 M dimension is handled by the forest of preset kind, generate respectively 2M+1 a 3 tie up to
The 2M+1 3 dimensional vectors are connected to one, generate a G dimensional vector by amount.Wherein, M=N/3, G=3 × (2M+1) × forest
Number.
For generated G dimensional vector, the characteristic information for all receiving feature vector using every layer of cascade system and being included, warp
It crosses the combination of multilayer polymorphic type forest and carries out feature learning, each layer of output result is spliced with generated G dimensional vector, then
It is secondary to carry out feature learning by every layer of forest combination, each layer parameter is saved as multidimensional spy according to frequency of training and convergence index
Levy model.
S4: searching medical data corresponding to medical insurance data to be detected, by medical insurance data to be detected and its corresponding doctor
It treats data and inputs multidimensional characteristic model generated, detect type belonging to the medical insurance behavior that medical insurance data to be detected are included.
Wherein, type belonging to medical insurance behavior includes normal, doubtful and violation.
S1~S4 through the above steps, multidimensional characteristic model is i.e. as the detection model to medical insurance unlawful practice, Neng Goushi
Medical insurance unlawful practice now accurate, timely, being effectively detected out in medical procedure, is saved to help to improve medical insurance control expense
Fund, control the reasonable growth of medical insurance fund.
In order to further increase accuracy, the timeliness, validity of medical insurance unlawful practice detection, as a kind of possible reality
Existing mode can increase following step after step s4:
S5: manual examination and verification are carried out to doubtful and violation medical insurance behavior is detected as, the result of manual examination and verification is as labeled
Training sample return step S3.
Medical insurance unlawful practice is capable of forming by step S5 and detects self study closed loop, to largely improve inspection
The sensibility for surveying model (i.e. multidimensional characteristic model), can detected emerging medical insurance unlawful practice more quickly.
It should be pointed out that the detection method of medical insurance unlawful practice provided by the present embodiment, in addition to helping to improve doctor
It protects control to take outside saved fund, additionally it is possible to carry out medical procedure the monitoring for rationally, closing rule, carry out for different unlawful practices
Prevent early warning, alarm prompt in thing in advance, ex-post analysis control guarantees the true, in due course, appropriate, appropriate of medical act.
Based on the detection method of above-mentioned medical insurance unlawful practice, the present embodiment additionally provides a kind of detection of medical insurance unlawful practice
Device, as shown in Fig. 2, the detection device includes: medical insurance data memory module 1, medical data storage module 2, feature extraction mould
Block 3, training module 4 and detection module 5.
Wherein, medical insurance data memory module 1 is for storing medical insurance data.
Medical data storage module 2 is deposited in the medical data and medical insurance data memory module 1 for storing medical data
The medical insurance data of storage are corresponding, and the medical data is to close rule, number from medical resource consumption, medical act main body, medical act
According to the data that multiple angles such as characteristic are described, to improve the authenticity of data and comprehensive.
Characteristic extracting module 3 is connected with medical insurance data memory module 1 and medical data storage module 2, is used for from medical insurance number
According to obtaining its medical insurance data stored in memory module 1, and its medical treatment stored is obtained from medical data memory module 2
Then data carry out feature extraction to acquired medical data and medical insurance data using machine learning algorithm, obtain feature to
Amount.
Training module 4 is connected with characteristic extracting module 3 and medical insurance data memory module 1, for storing mould from medical insurance data
Medical insurance data are obtained in block 1, and obtain feature vector from characteristic extracting module 3, then using deep learning algorithm to being obtained
The feature vector and medical insurance data taken is trained, and multidimensional characteristic model is generated, as detection model.
Detection module 5 is connected with training module 4 and medical insurance data memory module 1, for searching medical insurance data institute to be detected
Corresponding medical data, and multidimensional characteristic model is obtained from training module 4, then by medical insurance data to be detected and its corresponding
Medical data input the multidimensional characteristic model, the medical insurance for being included using multidimensional characteristic model inspection medical insurance data to be detected
Type belonging to behavior is which seed type in normal, doubtful and violation.
It is entirely square from multi-angle by introducing medical data using the detection device of medical insurance unlawful practice presented above
The description medical data of position utilizes machine learning algorithm combination deep learning then according to these medical datas and medical insurance data
Algorithm is trained modeling, thus realize it is accurate, timely, effectively medical insurance unlawful practice is detected.
As a kind of possible design, continuing with referring to fig. 2, detection device provided by the present embodiment can also include:
Manual examination and verification module 6, which is connected with detection module 5 and training module 4, for obtaining from detection module 5
To the testing result of medical insurance behavior, manual examination and verification then are carried out to being detected as doubtful and violation medical insurance behavior, and will manually examine
The result of core feeds back to training module 4 as labeled training sample.
By above structure, so that training module 4, detection module 5 and manual examination and verification module 6 form medical insurance unlawful practice inspection
Self study closed loop is surveyed, to largely improve the sensibility of detection model (i.e. multidimensional characteristic model), is occurred to new
Medical insurance unlawful practice can detected more quickly.
In said structure, a variety of access modes can be provided to use for manual examination and verification module 6, in order to auditor couple
Doubtful and violation medical insurance behavior is audited.
In addition, feeding back to trained mould for the result of manual examination and verification as labeled training sample in manual examination and verification module 6
After block 4, training module 4 can carry out detection model self study after labeled training sample meets strategy.Wherein, so-called " pre-
If tactful " certain threshold value can be met for the quantity of labeled training sample, naturally it is also possible to it is other strategies, it is unlimited herein
It is fixed.
Based on the detection device of above-mentioned medical insurance unlawful practice, the present embodiment additionally provides a kind of medical insurance control charge system, the doctor
Protect the detection device that control charge system includes the medical insurance unlawful practice as provided by the present embodiment.
In the medical insurance control charge system, provided detection device is carried out medical insurance unlawful practice fast through this embodiment
Then speed, accurate detection carry out medical insurance control expense according to testing result, improve the fund that medical insurance control expense can save.
The foregoing is merely a specific embodiment of the invention, but scope of protection of the present invention is not limited thereto, any
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by those familiar with the art, all answers
It is included within the scope of the present invention.Therefore, protection scope of the present invention should be with the scope of protection of the claims
It is quasi-.
Claims (9)
1. a kind of detection method of medical insurance unlawful practice, which is characterized in that described detection method includes the following steps:
S1: medical insurance data and its corresponding medical data are obtained, the medical data is described from multiple angles；
S2: feature extraction is carried out to the medical data and the medical insurance data using machine learning algorithm, obtains feature vector；
S3: being trained described eigenvector and the medical insurance data using deep learning algorithm, generates multidimensional characteristic model；
S4: searching medical data corresponding to medical insurance data to be detected, by the medical insurance data to be detected and its corresponding doctor
It treats data and inputs the multidimensional characteristic model, detect type belonging to the medical insurance behavior that the medical insurance data to be detected are included；
Wherein, the type includes normal, doubtful and violation.
2. detection method according to claim 1, which is characterized in that in step sl, it is described from multiple angles to described
Medical data, which is described, to be specifically included:
Using medical diagnosis on disease associated packets mode, the medical data is described；
By hospital, the essential information and relationship of doctor, patient three, the medical data is described；
From clinical path, Expert Rules direction, the medical data is described；
From the data direction of multiclass different characteristics, the medical data is described.
3. detection method according to claim 1, which is characterized in that step S2 is specifically included:
According to medical diagnosis on disease associated packets, patient is divided into diagnosis relevant group, provides same medical treatment for each diagnosis relevant group
Medical insurance section under resources supplIes；
By hospital, the essential information and relationship of doctor, patient three, the medical data is analyzed, and uses a variety of engineerings
It practises algorithm and data normalization is carried out to the medical data；
It is inputted Expert Rules as feature, obtain the Expert Rules returns the result data, returns the result data for described
It is divided into the data of multiclass different characteristics, and Various types of data is normalized respectively.
4. detection method according to claim 3, which is characterized in that described to carry out data normalization to the medical data
It specifically includes:
Using statistical distribution mode, by expense, length of stay data normalization to [0,1] section；Wherein, using 5 quartiles as just
Beginning minimum value, using 95 quartiles as original maximum, less than 5 quartiles and greater than the number of 95 quartiles in expense, length of stay data
According to setting 1, remainder data sets 0；
Using International Classification of Diseases as group basis, Western medicine medication record, inspection chemical examination record are clustered in same disease, obtained
To Western medicine types of medicines, checking test item, the Western medicine types of medicines in drug white list is labeled as 1, it is otherwise labeled as 0,
It checks that the inspection test item in chemical examination white list is labeled as 1, is otherwise labeled as 0.
5. detection method according to claim 1, which is characterized in that step S3 is specifically included:
The medical data, the medical insurance data and described eigenvector are spliced, the training data of N-dimensional is formed；Wherein,
The medical data is N1 dimension data, and the medical insurance data are N2 dimension data, and described eigenvector is N3 dimension data, N-dimensional=N1
Dimension+N2 dimension+N3 dimension；
Value is carried out on the training data of the N-dimensional with the vector that M is tieed up, default step-length is set as 1, generates 2M+1 M dimension
Vector, the vector of 2M+1 M dimension is handled by the forest of preset kind, generates 2M+1 3 dimensional vectors respectively, will
The 2M+1 3 dimensional vectors are connected to one, generate a G dimensional vector；Wherein, M=N/3, G=3 × (2M+1) × forest
Number；
For the G dimensional vector, all receive characteristic information using every layer of cascade system, combines and carry out by multilayer polymorphic type forest
Feature learning, each layer of output result are spliced with the G dimensional vector, are again passed by every layer of forest combination and are carried out feature
Study saves each layer parameter as multidimensional characteristic model according to frequency of training and convergence index.
6. described in any item detection methods according to claim 1~5, which is characterized in that after step s4, further includes:
S5: manual examination and verification are carried out to doubtful and violation medical insurance behavior is detected as, the result of manual examination and verification is as labeled instruction
Practice sample return step S3.
7. a kind of detection device of medical insurance unlawful practice, which is characterized in that the detection device includes:
Medical insurance data memory module, for storing medical insurance data；
Medical data storage module, for storing medical data corresponding to the medical insurance data, and the medical data be from
The data that multiple angles are described；
The characteristic extracting module being connected with the medical insurance data memory module and the medical data storage module, for using machine
Device learning algorithm carries out feature extraction to the medical data and the medical insurance data, obtains feature vector；
The training module being connected with the characteristic extracting module and the medical insurance data memory module, for being calculated using deep learning
Method is trained described eigenvector and the medical insurance data, generates multidimensional characteristic model；
The detection module being connected with the training module and the medical insurance data memory module, for searching medical insurance data to be detected
The medical insurance data to be detected and its corresponding medical data are inputted the multidimensional characteristic mould by corresponding medical data
Type detects type belonging to the medical insurance behavior that the medical insurance data to be detected are included；Wherein, the type includes normal, doubtful
Sihe is in violation of rules and regulations.
8. detection device according to claim 7, which is characterized in that the detection device further include: with the detection mould
The manual examination and verification module that block is connected with the training module, for manually being examined to being detected as doubtful and violation medical insurance behavior
Core, and the training module is fed back to using the result of manual examination and verification as labeled training sample.
9. a kind of medical insurance control charge system, which is characterized in that the medical insurance control charge system includes inspection as claimed in claim 7 or 8
Survey device.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711367590.1A CN109934719A (en) | 2017-12-18 | 2017-12-18 | The detection method and detection device of medical insurance unlawful practice, medical insurance control charge system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711367590.1A CN109934719A (en) | 2017-12-18 | 2017-12-18 | The detection method and detection device of medical insurance unlawful practice, medical insurance control charge system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109934719A true CN109934719A (en) | 2019-06-25 |
Family
ID=66982827
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711367590.1A Pending CN109934719A (en) | 2017-12-18 | 2017-12-18 | The detection method and detection device of medical insurance unlawful practice, medical insurance control charge system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109934719A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110942396A (en) * | 2019-11-28 | 2020-03-31 | 泰康保险集团股份有限公司 | Data processing method, device and equipment |
| CN111161814A (en) * | 2019-12-18 | 2020-05-15 | 浙江大学 | DRGs automatic grouping method based on convolutional neural network |
| CN111445343A (en) * | 2020-03-23 | 2020-07-24 | 平安医疗健康管理股份有限公司 | Risk medical data identification method and device, computer equipment and storage medium |
| CN112801805A (en) * | 2021-01-21 | 2021-05-14 | 浙江大学山东工业技术研究院 | Medical insurance small card fraud detection method and system based on deep self-supervision neural network |
| CN113642672A (en) * | 2021-08-31 | 2021-11-12 | 平安医疗健康管理股份有限公司 | Feature processing method and device of medical insurance data, computer equipment and storage medium |
| CN113657548A (en) * | 2021-08-31 | 2021-11-16 | 平安医疗健康管理股份有限公司 | Medical insurance abnormity detection method and device, computer equipment and storage medium |
| CN113780792A (en) * | 2021-09-03 | 2021-12-10 | 广州云从鼎望科技有限公司 | Medical insurance violation monitoring method and device and computer readable storage medium |
| CN113869387A (en) * | 2021-09-18 | 2021-12-31 | 平安科技（深圳）有限公司 | Abnormal medical insurance reimbursement identification method and system based on artificial intelligence technology |
| CN115080997A (en) * | 2022-06-02 | 2022-09-20 | 武汉金豆医疗数据科技有限公司 | Mobile checking method and device for medical insurance fund, computer equipment and storage medium |
| CN116070693A (en) * | 2023-04-06 | 2023-05-05 | 北京亚信数据有限公司 | Patient information and medical service relation detection model training and detection method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100794516B1 (en) * | 2007-12-03 | 2008-01-14 | 한국정보통신대학교 산학협력단 | System and method for diagnosis and clinical test selection using case based machine learning inference |
| CN104766127A (en) * | 2014-01-02 | 2015-07-08 | 深圳市永泰中天软件股份有限公司 | Method for predicting the number of hospitalization days of patient after surgery |
| CN105159948A (en) * | 2015-08-12 | 2015-12-16 | 成都数联易康科技有限公司 | Medical insurance fraud detection method based on multiple features |
| CN105279382A (en) * | 2015-11-10 | 2016-01-27 | 成都数联易康科技有限公司 | Medical insurance abnormal data on-line intelligent detection method |
| CN106874658A (en) * | 2017-01-18 | 2017-06-20 | 天津艾登科技有限公司 | A kind of medical insurance fraud recognition methods based on Principal Component Analysis Algorithm |
-
2017
- 2017-12-18 CN CN201711367590.1A patent/CN109934719A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100794516B1 (en) * | 2007-12-03 | 2008-01-14 | 한국정보통신대학교 산학협력단 | System and method for diagnosis and clinical test selection using case based machine learning inference |
| CN104766127A (en) * | 2014-01-02 | 2015-07-08 | 深圳市永泰中天软件股份有限公司 | Method for predicting the number of hospitalization days of patient after surgery |
| CN105159948A (en) * | 2015-08-12 | 2015-12-16 | 成都数联易康科技有限公司 | Medical insurance fraud detection method based on multiple features |
| CN105279382A (en) * | 2015-11-10 | 2016-01-27 | 成都数联易康科技有限公司 | Medical insurance abnormal data on-line intelligent detection method |
| CN106874658A (en) * | 2017-01-18 | 2017-06-20 | 天津艾登科技有限公司 | A kind of medical insurance fraud recognition methods based on Principal Component Analysis Algorithm |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110942396A (en) * | 2019-11-28 | 2020-03-31 | 泰康保险集团股份有限公司 | Data processing method, device and equipment |
| CN111161814A (en) * | 2019-12-18 | 2020-05-15 | 浙江大学 | DRGs automatic grouping method based on convolutional neural network |
| CN111445343A (en) * | 2020-03-23 | 2020-07-24 | 平安医疗健康管理股份有限公司 | Risk medical data identification method and device, computer equipment and storage medium |
| CN112801805A (en) * | 2021-01-21 | 2021-05-14 | 浙江大学山东工业技术研究院 | Medical insurance small card fraud detection method and system based on deep self-supervision neural network |
| CN113642672A (en) * | 2021-08-31 | 2021-11-12 | 平安医疗健康管理股份有限公司 | Feature processing method and device of medical insurance data, computer equipment and storage medium |
| CN113657548A (en) * | 2021-08-31 | 2021-11-16 | 平安医疗健康管理股份有限公司 | Medical insurance abnormity detection method and device, computer equipment and storage medium |
| CN113642672B (en) * | 2021-08-31 | 2024-05-14 | 平安医疗健康管理股份有限公司 | Feature processing method and device of medical insurance data, computer equipment and storage medium |
| CN113780792A (en) * | 2021-09-03 | 2021-12-10 | 广州云从鼎望科技有限公司 | Medical insurance violation monitoring method and device and computer readable storage medium |
| CN113869387A (en) * | 2021-09-18 | 2021-12-31 | 平安科技（深圳）有限公司 | Abnormal medical insurance reimbursement identification method and system based on artificial intelligence technology |
| CN115080997A (en) * | 2022-06-02 | 2022-09-20 | 武汉金豆医疗数据科技有限公司 | Mobile checking method and device for medical insurance fund, computer equipment and storage medium |
| CN115080997B (en) * | 2022-06-02 | 2024-01-09 | 武汉金豆医疗数据科技有限公司 | Mobile checking method and device for medical insurance fund, computer equipment and storage medium |
| CN116070693A (en) * | 2023-04-06 | 2023-05-05 | 北京亚信数据有限公司 | Patient information and medical service relation detection model training and detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109934719A (en) | The detection method and detection device of medical insurance unlawful practice, medical insurance control charge system | |
| Jagannatha et al. | Bidirectional RNN for medical event detection in electronic health records | |
| Tougui et al. | Impact of the choice of cross-validation techniques on the results of machine learning-based diagnostic applications | |
| CN110136788B (en) | Medical record quality inspection method, device, equipment and storage medium based on automatic detection | |
| AU2019240633A1 (en) | System for automated analysis of clinical text for pharmacovigilance | |
| US11328220B2 (en) | Prediction characterization for black box machine learning models | |
| CN108614885A (en) | Knowledge mapping analysis method based on medical information and device | |
| US11651252B2 (en) | Prognostic score based on health information | |
| Halpern et al. | A comparison of dimensionality reduction techniques for unstructured clinical text | |
| KR20200023689A (en) | The method of artificial intelligence(AI)-based adverse drug reactions detection and the system thereof | |
| CN109545317A (en) | The method and Related product of behavior in hospital are determined based on prediction model in hospital | |
| CN114026651A (en) | Automatic generation of structured patient data records | |
| CN109346145A (en) | A kind of actively monitoring method and system of adverse drug reaction | |
| CN106919794A (en) | Towards the drug class entity recognition method and device of multi-data source | |
| De Bruijn et al. | Automated information extraction of key trial design elements from clinical trial publications | |
| US20230086217A1 (en) | Multimodal Cell Complex Neural Networks for Prediction of Multiple Drug Side Effects Severity and Frequency | |
| Trenta et al. | Extraction of evidence tables from abstracts of randomized clinical trials using a maximum entropy classifier and global constraints | |
| CN111291554A (en) | Labeling method, relation extracting method, storage medium, and computing device | |
| CN113707253A (en) | Medical scheme recommendation method, device, equipment and medium | |
| CN115050442A (en) | Disease category data reporting method and device based on mining clustering algorithm and storage medium | |
| Chiang et al. | A large language model–based generative natural language processing framework fine‐tuned on clinical notes accurately extracts headache frequency from electronic health records | |
| CN112071431B (en) | Clinical path automatic generation method and system based on deep learning and knowledge graph | |
| CN113724830A (en) | Medicine taking risk detection method based on artificial intelligence and related equipment | |
| Bodenreider et al. | Comparing the representation of anatomy in the FMA and SNOMED CT | |
| US20220051114A1 (en) | Inference process visualization system for medical scans |
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 |