CN111191415A - Operation classification coding method based on original operation data - Google Patents
Operation classification coding method based on original operation data Download PDFInfo
- Publication number
- CN111191415A CN111191415A CN201911305459.1A CN201911305459A CN111191415A CN 111191415 A CN111191415 A CN 111191415A CN 201911305459 A CN201911305459 A CN 201911305459A CN 111191415 A CN111191415 A CN 111191415A
- Authority
- CN
- China
- Prior art keywords
- data
- coding
- entity
- surgical
- original
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention relates to the technical field of medical informatization and the technical field of artificial intelligence, in particular to an operation classification coding method and system based on original operation data, which comprises the following steps: receiving raw surgical data; preprocessing input original operation data; performing retrieval and judgment by using international operation standard classification codes ICD-9-CM-3; inquiring admission records, disease course records and discharge records; and (3) coding according to the preprocessed original operation data and the acquired information in the patient medical record by using a bidirectional operation data entity recognition model and a neural entity weight relation discrimination network, and outputting a coding result and an accuracy evaluation result. Compared with the manual coding, the automatic coding method is more efficient, faster and stable in classification, can realize the coding classification of the original medical records in a short time in a large batch, and can quickly prepare and arrange data for large data application and artificial intelligence in the medical field.
Description
Technical Field
The invention relates to the technical field of medical informatization and the technical field of artificial intelligence, in particular to an operation classification coding method and system based on original operation data.
Background
With the continuous deepening of the national medical reform and the generalization of medical layout. The informatization of medical data greatly limits the improvement of medical reform and the whole aspect of medical service. In the period of comprehensive development of medical big data and artificial intelligence, the informatization of medical data begins to be displayed, and the auxiliary decision making, information support and the like are provided for the diagnosis and treatment process, so that the efficiency of doctors can be improved, and the work of the doctors can be greatly reduced. The standardization of medical data has been a difficult problem to be solved and improved in the medical informatization evolution history. The classification standards of diagnosis codes, operation codes, medicine codes and the like are important in the standardization of medical data. The system aims at the operation code and provides technical support for the operation code of the hospital
(1) Hospital coding personnel have a poor understanding of standard surgical codes, resulting in differences in coding results.
The coding and classifying work of the surgical codes in hospitals is manually finished because of the limitation of low manpower (even a plurality of hospitals do not have coders), the quality of coding personnel is uneven, and the degree of understanding of the coding personnel on standard interpretation and huge content of a coding dictionary is uneven. The number of manual codes in hospitals is limited every day and there are some cases of misjudgment. The thickness degree and the judgment standard of the codes of the encoding personnel are different from person to person, the problems can be obviously seen when the data of different hospitals are reported, examined and exchanged together, and even the encoding results of different encoding personnel in the same hospital and different encoding personnel in different periods of the same encoding personnel are different.
(2) The actual operation input by the doctor in a personalized and self-defined way is not completely compatible with the standard operation code library.
Because the operations contained in the operation code library can not completely cover all daily operations in a hospital, and for the same operation, because of the particularity of Chinese, the extreme diversity of the expression modes of the same disease concept and the standard of unified standard medical terms in the industry do not exist, the name writing methods and the structures of the operations used by different doctors are different, and the operation code library has eight flowers and eight doors, and the used aliases have great difference; in addition, the updating of the standard operation dictionary takes a long time (national standard ICD9-CM3, last updating 2011), and the latest operation and the name thereof cannot be covered in time; in addition, many new types of surgeries and new types of surgical auxiliary instruments and the like appear along with the updating of the technology in the years. And for typing of specific diagnosis, the detail degree of the standard codes does not meet the clinical practical requirements of doctors; or the doctor can add some extra detailed information during the writing operation, and the standard diagnosis code library cannot reflect the differential classification brought by the extra information; furthermore, the doctor may use some shorthand abbreviations or abbreviations that are well known in the art when writing the operation, and these are not included in the standard operation code library. Fundamentally, doctors write operations from their perspective, and the doctors can conveniently check the operations by themselves in the future, but not from the perspective of code classification, many doctors cannot know what the standard operations are in the surgical codes and can not remember the standard operations, and if the doctors are forced to write the standard operations, the doctors are undoubtedly in trouble. The above reasons cause great troubles in actual hospital operations for most doctors to perform procedure selection operations when filling out operations or to perform hand-written operations by the doctors themselves.
(3) The condition that the standard operation code standards are not uniform exists in each hospital
At present, a plurality of hospitals adopt versions modified by national standards according to the requirements and wishes of the hospitals, and a great amount of manpower and material resources are required to be invested in modifying and revising the versions. All of the hospital surgical code versions are not uniform, and even some hospitals do not adopt national standards for extension and revision. Is not beneficial to the uniform processing and the uniform reporting of the surgical data of the patient.
(4) The situation requirement of big data and medical data standardization, and the surgery code is more standardized and more accurate.
The construction of a medical information platform and a medical information warehouse is greatly limited due to the imperfection and non-uniformity of the medical information. And (4) standard and comprehensive construction of a medical information platform and a medical information warehouse. The further perfection of functions such as medical big data, medical artificial intelligence, auxiliary diagnosis and treatment, medical dynamic monitoring and the like at present also needs more accurate and more standard coding work, and the data volume of coding is also huge.
Disclosure of Invention
Aiming at the four problems, the invention provides an operation classification coding method based on original operation data, which automatically compares a standard diagnosis library ICD-9-CM-3 for coding according to the operation input by a doctor and by combining the analysis word segmentation and semantic understanding of medical records and evaluates the correctness of the operation input by the doctor.
In order to achieve the purpose, the invention adopts the following technical scheme:
a surgery classification coding method based on original surgery data is characterized by comprising the following steps:
(1): receiving input data: the input data bits are original surgical data;
(2): preprocessing input raw surgical data, the preprocessing comprising: removing punctuation marks, converting variant characters into body characters, and converting full-angle characters into half-angle characters;
(3): searching in the international operation standard classification code ICD-9-CM-3 of the preprocessing result obtained in the step (2), judging whether a result is obtained, and if so, directly outputting the coding result; if not, entering the step (4);
(4): inquiring admission records, disease course records and discharge records according to the case number of the patient subjected to the original operation input by the doctor, and acquiring the age, sex, disease and operation process description of the patient;
(5): coding according to the preprocessed original operation data and the acquired information in the patient medical record by using a bidirectional operation data entity recognition model and a neural entity weight relation discrimination network, and outputting a coding result and an accuracy evaluation result;
the calculation mode of the bidirectional operation data entity recognition model is
Wherein p represents probability, D represents the currently input original operation text, L represents the data mark corresponding to D, D represents one piece of data of D, p (L | D) represents the probability of the gold standard data mark output by the model under the condition of knowing the original operation text, and LjRepresenting the data mark corresponding to the jth word in the d, and theta is the set of all parameters of the bidirectional operation data identification and coding model, wherein the set comprises omegak，BmAnd fk，gmThe parameters of the two functions, Z is a normalization factor, ensuring that the result of the model output is a real number between 0 and 1, fkTransfer of discriminant function, ω, to entity classkIs fkWeight of gmFor bi-directional feature mapping of language models, BmFor its corresponding weight, the training process of the model is to randomly initialize ω with a uniform distribution between 0 and 1k，BmAnd fk，gmThe parameter argmax of these two functions is the parameter that maximizes p (L | D) by finding ωk，BmAnd fk，gmThe set of parameters for these two functions, when inferred using the bi-directional surgical data recognition and coding model, the value of the p (LlD) output greater than 0.5 indicates that the current data belongs to a certain entity type.
Specific g (l)j，dj)＝P(lj|dj) The modeling process is
Where u is the u-th possible mark, | u | is the total number of possible marks, diRepresents the j th word, theta, in dgDenotes g (l)j，dj) The parameter (c) is set, argmax represents the possible P (l)j|dj) A set of maximized parameters;
the calculation mode of the neural entity relationship discrimination network is
Wherein sigma is sigmoid function, F is output of neural entity weight relation discrimination network, h1＝ g(lj，dj)、h2＝g(lj，dj)，
Representing a matrix cascade, linear (x) ═ Φ x + bias where Φ is the mapping parameter, representing the bias intercept, u is the learnable parameter,
e is indicative of belonging to,
is a real space, l is the length of the linguistic sequence, h _ dim is the dimension of the custom hidden layer,
wherein η is a hyperparameter, when η is 8 after a lot of tests and training, the entity weight relationship is judged to be more consistent with the encoding mode of the encoder
Wherein Θ isFRepresenting weights of neural entitiesThe set of parameters of the relationship discriminating network, CE representing the cross entropy, i.e.
I K is the number of categories, K is the current category, γ is the data label,
predicting output for the model;
the neural entity relationship discrimination network is used for calculating entity weight according to each entity identified by the bidirectional operation data entity identification model, finally obtaining a result and outputting the result to an interactive interface or a designated file or a database.
In the operation classification coding method based on the original operation data, the standard operation name includes an operation type, an operation position, disease properties, an operation approach, an auxiliary material for the operation, an operation mode and an operation proper name.
In the operation classification coding method based on the original operation data, the standard operation names are operation positions and operation modes.
Has the advantages that:
1. solves the problem that the original operation of the doctor can only be finished manually corresponding to the standard operation,
2. solves the problem that the doctor is difficult to compare with standard diagnosis because the doctor inputs the terms of the operation randomly.
3. The problem of the standard operation code that each medical institution used is not uniform is solved. After the automatic operation codes are used, the original operation corresponds to the same set of standard diagnosis codes, the classification standards are unified, and the standards are unified in the data exchange process of the medical institution.
4. The problem that corresponding codes before and after the same diagnosis are inconsistent due to non-uniform standards is solved.
5. The automatic coding saves a large amount of human resources, and the diagnosis of the doctor description error can be verified by a coder; and the automatic coding of the program is more efficient than manpower. The encoding speed is faster.
6. Automatic surgical coding is beneficial to ensuring the accuracy of data retrieval for medical treatment, teaching and scientific research, and the development of the group DRGS.
7. Because the automatic coding is quick and stable in classification, the original medical records can be coded and classified in a short time in large batch, the data can be quickly prepared and arranged for large data application and artificial intelligence in the medical field, and the automatic coding has irreplaceable effect as a basic function in the field.
The specific implementation mode is as follows:
the invention relates to a surgery classification coding method based on original surgery data, which comprises the following steps:
step (1): receiving input data: the input data bits are original surgical data;
step (2): preprocessing input raw surgical data, the preprocessing comprising: removing punctuation marks, converting variant characters into body characters, and converting full-angle characters into half-angle characters;
and (3): searching in the international operation standard classification code ICD-9-CM-3 of the preprocessing result obtained in the step (2), judging whether a result is obtained, and if so, directly outputting the coding result; if not, entering the step (4);
and (4): according to the case number of the patient of the original operation input by the doctor, inquiring the admission record, the course record and the discharge record. The patient's age, sex, disease, and surgical procedure description are obtained.
And (5): and (3) coding according to the preprocessed original operation data and the acquired information in the patient medical record by using a bidirectional operation data entity recognition model and a neural entity weight relation discrimination network, and outputting a coding result and an accuracy evaluation result. The results that may be given in step (5) are two types 1. return one or more correct results, 2 return result is null, mainly because the original operation information of the doctor including the incomplete description in the medical record
The standard operation names generally include operation types, operation sites, disease properties, surgical approaches, surgical aids, operation modes, operation proper names (generally, the names of the memorial persons who invented operations, etc.), and the like. Wherein the surgical site, and the surgical modality are the core surgical code's classification axis. The bidirectional operation data identification and coding model summarizes the coding modes of a plurality of coders through the training of a large amount of operation data, and finally shapes.
The system comprises the following modules:
1: original operation acquisition and medical record query module
The module is only used for acquiring original operation data, and acquiring information such as the patient age, sex, disease, operation process and the like in the hospital admission medical record, the medical course record and the patient discharge medical record according to the patient case number if needed, and is used for a module 3 two-way operation data entity identification model behind.
2: and an original operation pretreatment module.
The module is only used for processing the unconventional characters in the original diagnosis input by a doctor, and if the preprocessing result exists in the standard operation library, the result is directly returned without passing through a subsequent module, so that the coding speed can be greatly increased.
3: standard surgical memory module:
the module stores the current stage standard surgical code set. After the module 2 is preprocessed, the inquiry is carried out, and the weight relation of the neural entity is calculated by the summation module 5.
4: bidirectional operation data entity recognition model
The bidirectional operation data identification and coding model is mainly used for distinguishing and detecting entities in information obtained by inquiring original operation data and medical records.
The calculation mode of the bidirectional operation data identification and coding model is
Wherein p represents probability, D represents the currently input original operation text, L represents the data mark corresponding to D, D represents one piece of data of D, p (L | D) represents the probability of the gold standard data mark output by the model under the condition of knowing the original operation text,ljrepresenting the data mark corresponding to the jth word in the d, and theta is the set of all parameters of the bidirectional operation data identification and coding model, wherein the set comprises omegak，BmAnd fk，gmThe parameters of the two functions, Z is a normalization factor, ensuring that the result of the model output is a real number between 0 and 1, fkTransfer of discriminant function, ω, to entity classkIs fkWeight of gmFor bi-directional feature mapping of language models, BmFor its corresponding weight, the training process of the model is to randomly initialize ω with a uniform distribution between 0 and 1k，BmAnd fk，gmThe parameter argmax of these two functions is the parameter that maximizes p (L | D) by finding ωk，BmAnd fk，gmThe set of parameters for these two functions, when inferred using the bi-directional surgical data recognition and coding model, the value of the p (LlD) output greater than 0.5 indicates that the current data belongs to a certain entity type.
Specific g (l)j，dj)＝P(lj|dj) The modeling process is
Where u is the u-th possible mark, | u | is the total number of possible marks, djRepresents the j th word, theta, in dgDenotes g (l)j，dj) The parameter (c) is set, argmax represents the possible P (l)j|dj) A set of maximized parameters.
5: neural entity relationship discrimination network
The neural entity relationship discrimination network is used for calculating entity weight according to each entity identified by the module 4 and finally obtaining a result.
The neural entity relationship discrimination network is calculated in the way that
Wherein sigma is sigmoid function, F is output of neural entity weight relation discrimination network, h1＝ g(lj，dj)、h2＝g(lj，dj)，
Representing a matrix cascade, linear (x) ═ Φ x + bias where Φ is the mapping parameter, representing the bias intercept, u is the learnable parameter,
e is indicative of belonging to,
is a real space, l is the length of the linguistic sequence, h _ dim is the dimension of the custom hidden layer,
wherein η is a hyperparameter, when η is 8 after a lot of tests and training, the entity weight relationship is judged to be more consistent with the encoding mode of the encoder
Wherein Θ isFSet of parameters representing neural entity weight relationship discriminant networks, CE represents cross entropy, i.e.
I K is the number of categories, K is the current category, γ is the data label,
and (5) predicting output for the model.
6: a result output module:
and the result output module is used for outputting the result to an interactive interface or a designated file or a database.
The following is a detailed description of the embodiments with reference to the examples
Example 1: assuming that the operation input by the doctor in the step one is cesarean section, the lower uterine segment is passed through, the result is obtained in the step two, namely the lower uterine segment of cesarean section (in order to eliminate the influence of punctuation marks on matching, the punctuation marks are replaced in the step two, and the punctuation marks are also removed in the standard operation used for training in the step four), the processing result in the step two is passed through the step three, the corresponding standard operation is not found, and the step four is entered. The calculation result of the bidirectional operation data entity identification model in the fourth step is 0.5831, and then the result of the standard operation with the highest relation calculated by the neural entity relation judgment network is 0.999974.1x00 lower uterine segment cesarean section. The present example shows that the doctor can write randomly and write a few more words for some operations, and the coding result of the system is not affected.
Example 2: assuming that the surgery input by the doctor obtained in the step one is a femoral fracture closed reduction PFNA internal fixation surgery, the femoral fracture closed reduction PFNA internal fixation surgery is obtained through the step two (no special characters or punctuations exist, so the surgery is not processed through the step two), and after the step three, a corresponding standard surgery is not found, and the operation enters the step four. The calculation result of the bidirectional operation data entity identification model in the fourth step is 0.8013, and then the neural entity relationship judgment network is entered to calculate the result of the standard operation with the highest relationship to be 0.999979.1500x006 femur fracture closed reduction intramedullary pin internal fixation. The PFNA mentioned in this example is a new type of intramedullary needle, from which it can be found that the system can automatically identify concepts not mentioned in detail in some standard diagnoses written by the physician during surgery.
Surgical automated code deployment approach:
the operation automatic coding deployment is simple, a hospital only needs one computer as a front-end processor, original operation data are placed in a fixed database every day, and every day, the operation automatic coding service automatically acquires the original operation data, performs automatic coding and returns results to the database.
Claims (3)
1. A surgery classification coding method based on original surgery data is characterized by comprising the following steps:
(1): receiving input data: the input data bits are original surgical data;
(2): preprocessing input raw surgical data, the preprocessing comprising: removing punctuation marks, converting variant characters into body characters, and converting full-angle characters into half-angle characters;
(3): searching in the international operation standard classification code ICD-9-CM-3 of the preprocessing result obtained in the step (2), judging whether a result is obtained, and if so, directly outputting the coding result; if not, entering the step (4);
(4): inquiring admission records, disease course records and discharge records according to the case number of the patient subjected to the original operation input by the doctor, and acquiring the age, sex, disease and operation process description of the patient;
(5): coding according to the preprocessed original operation data and the acquired information in the patient medical record by using a bidirectional operation data entity recognition model and a neural entity weight relation discrimination network, and outputting a coding result and an accuracy evaluation result;
the calculation mode of the bidirectional operation data entity recognition model is
Wherein p represents probability, D represents the currently input original operation text, L represents the data mark corresponding to D, D represents one piece of data of D, p (L | D) represents the probability of the gold standard data mark output by the model under the condition of knowing the original operation text, and LjRepresenting the data mark corresponding to the jth word in the d, and theta is the set of all parameters of the bidirectional operation data identification and coding model, wherein the set comprises omegak，BmAnd fk，gmThe parameters of the two functions, Z is a normalization factor, ensuring that the result of the model output is a real number between 0 and 1, fkTransfer of discriminant function, ω, to entity classkIs fkWeight of gmFor bi-directional feature mapping of language models, BmFor its corresponding weight, the training process of the model is to randomly initialize ω with a uniform distribution between 0 and 1k，BmAnd fk，gmThe parameter argmax of these two functions is the parameter that maximizes p (L | D) by finding ωk，BmAnd fk，gmThe set of parameters for these two functions, when inferred using the bi-directional surgical data recognition and coding model, the value of the p (LlD) output being greater than 0.5 indicates that the current data belongs to a certain entity type;
specific g (l)j，dj)＝P(lj|dj) The modeling process is
Where u is the u-th possible mark, | u | is the total number of possible marks, djRepresents the j th word, theta, in dgDenotes g (l)j，dj) The parameter (c) is set, argmax represents the possible P (l)j|dj) A set of maximized parameters;
the calculation mode of the neural entity relationship discrimination network is
Wherein sigma is sigmoid function, F is output of neural entity weight relation discrimination network, h1＝g(lj，dj)、h2＝g(li，dj)，
Representing a matrix cascade, linear (x) ═ Φ x + bias where Φ is the mapping parameter, representing the bias intercept, u is the learnable parameter,
e is indicative of belonging to,
is a real space, l is the length of the linguistic sequence, h _ dim is the dimension of the custom hidden layer,
wherein η is a hyperparameter, when η is 8 after a lot of tests and trainings, the entity weight relationship is judged to be more consistent with the encoding mode of the encoder, and the error of F and data mark C is minimized in the modeling process, that is, the error is minimized
Wherein Θ isFSet of parameters representing neural entity weight relationship discriminant networks, CE represents cross entropy, i.e.
I K is the number of categories, K is the current category, γ is the data label,
predicting output for the model;
the neural entity relationship discrimination network is used for calculating entity weight according to each entity identified by the bidirectional operation data entity identification model, finally obtaining a result and outputting the result to an interactive interface or a designated file or a database.
2. The surgical classification encoding method based on raw surgical data as claimed in claim 1, characterized in that: the standard operation name comprises operation type, operation position, disease property, operation approach, auxiliary equipment for operation, operation mode and operation proper name.
3. The surgical classification encoding method based on raw surgical data as claimed in claim 1, characterized in that: the standard surgical names are surgical site and surgical type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911305459.1A CN111191415A (en) | 2019-12-16 | 2019-12-16 | Operation classification coding method based on original operation data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911305459.1A CN111191415A (en) | 2019-12-16 | 2019-12-16 | Operation classification coding method based on original operation data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111191415A true CN111191415A (en) | 2020-05-22 |
Family
ID=70709830
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911305459.1A Pending CN111191415A (en) | 2019-12-16 | 2019-12-16 | Operation classification coding method based on original operation data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111191415A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112560400A (en) * | 2020-12-30 | 2021-03-26 | 杭州依图医疗技术有限公司 | Medical data processing method and device and storage medium |
| CN112700825A (en) * | 2020-12-30 | 2021-04-23 | 杭州依图医疗技术有限公司 | Medical data processing method and device and storage medium |
| CN112735545A (en) * | 2020-12-31 | 2021-04-30 | 杭州依图医疗技术有限公司 | Self-training method, model, processing method, device and storage medium |
| CN112735544A (en) * | 2020-12-30 | 2021-04-30 | 杭州依图医疗技术有限公司 | Medical record data processing method and device and storage medium |
| CN112749307A (en) * | 2020-12-30 | 2021-05-04 | 杭州依图医疗技术有限公司 | Medical data processing method and device and storage medium |
| CN112861535A (en) * | 2021-01-18 | 2021-05-28 | 山东众阳健康科技集团有限公司 | Surgery classification coding method and system based on diagnosis and treatment data |
| CN114445129A (en) * | 2022-01-13 | 2022-05-06 | 湖北国际物流机场有限公司 | BIM coding system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110035210A1 (en) * | 2009-08-10 | 2011-02-10 | Benjamin Rosenfeld | Conditional random fields (crf)-based relation extraction system |
| CN107577826A (en) * | 2017-10-25 | 2018-01-12 | 山东众阳软件有限公司 | Classification of diseases coding method and system based on raw diagnostic data |
| CN108829662A (en) * | 2018-05-10 | 2018-11-16 | 浙江大学 | A kind of conversation activity recognition methods and system based on condition random field structuring attention network |
| CN109145112A (en) * | 2018-08-06 | 2019-01-04 | 北京航空航天大学 | A kind of comment on commodity classification method based on global information attention mechanism |
| CN109471895A (en) * | 2018-10-29 | 2019-03-15 | 清华大学 | The extraction of electronic health record phenotype, phenotype name authority method and system |
-
2019
- 2019-12-16 CN CN201911305459.1A patent/CN111191415A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110035210A1 (en) * | 2009-08-10 | 2011-02-10 | Benjamin Rosenfeld | Conditional random fields (crf)-based relation extraction system |
| CN107577826A (en) * | 2017-10-25 | 2018-01-12 | 山东众阳软件有限公司 | Classification of diseases coding method and system based on raw diagnostic data |
| CN108829662A (en) * | 2018-05-10 | 2018-11-16 | 浙江大学 | A kind of conversation activity recognition methods and system based on condition random field structuring attention network |
| CN109145112A (en) * | 2018-08-06 | 2019-01-04 | 北京航空航天大学 | A kind of comment on commodity classification method based on global information attention mechanism |
| CN109471895A (en) * | 2018-10-29 | 2019-03-15 | 清华大学 | The extraction of electronic health record phenotype, phenotype name authority method and system |
Non-Patent Citations (1)
| Title |
|---|
| 李世超: "基于Hadoop平台和隐马尔可夫模型的生物医学命名实体识别方法研究", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112560400A (en) * | 2020-12-30 | 2021-03-26 | 杭州依图医疗技术有限公司 | Medical data processing method and device and storage medium |
| CN112700825A (en) * | 2020-12-30 | 2021-04-23 | 杭州依图医疗技术有限公司 | Medical data processing method and device and storage medium |
| CN112735544A (en) * | 2020-12-30 | 2021-04-30 | 杭州依图医疗技术有限公司 | Medical record data processing method and device and storage medium |
| CN112749307A (en) * | 2020-12-30 | 2021-05-04 | 杭州依图医疗技术有限公司 | Medical data processing method and device and storage medium |
| CN112700825B (en) * | 2020-12-30 | 2024-03-05 | 杭州依图医疗技术有限公司 | Medical data processing method, device and storage medium |
| CN112735545A (en) * | 2020-12-31 | 2021-04-30 | 杭州依图医疗技术有限公司 | Self-training method, model, processing method, device and storage medium |
| CN112861535A (en) * | 2021-01-18 | 2021-05-28 | 山东众阳健康科技集团有限公司 | Surgery classification coding method and system based on diagnosis and treatment data |
| CN112861535B (en) * | 2021-01-18 | 2023-11-14 | 众阳健康科技集团有限公司 | Surgical classification coding method and system based on diagnosis and treatment data |
| CN114445129A (en) * | 2022-01-13 | 2022-05-06 | 湖北国际物流机场有限公司 | BIM coding system |
| CN114445129B (en) * | 2022-01-13 | 2024-03-19 | 湖北国际物流机场有限公司 | BIM coding system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111191415A (en) | Operation classification coding method based on original operation data | |
| CN108831559B (en) | Chinese electronic medical record text analysis method and system | |
| CN107731269B (en) | Disease coding method and system based on original diagnosis data and medical record file data | |
| CN107705839B (en) | Disease automatic coding method and system | |
| CN111540468B (en) | ICD automatic coding method and system for visualizing diagnostic reasons | |
| CN111986770B (en) | Prescription medication auditing method, device, equipment and storage medium | |
| EP3567605A1 (en) | Structured report data from a medical text report | |
| CN106919793A (en) | A kind of data standardization processing method and device of medical big data | |
| CN111222340B (en) | Breast electronic medical record entity recognition system based on multi-standard active learning | |
| CN108305690A (en) | The nervous system disease point examines the system for distribution of out-patient department of method and the nervous system disease | |
| CN109378066A (en) | A kind of control method and control device for realizing disease forecasting based on feature vector | |
| CN110335653A (en) | Non-standard case history analytic method based on openEHR case history format | |
| CN109411062A (en) | A method of diabetes and its complication are judged automatically using physiological parameter based on depth learning technology | |
| CN115293161A (en) | Reasonable medicine taking system and method based on natural language processing and medicine knowledge graph | |
| CN111191456A (en) | Method for identifying text segmentation by using sequence label | |
| CN114186070A (en) | Medical term normalization method and device based on artificial intelligence model | |
| CN111477320A (en) | Construction system of treatment effect prediction model, treatment effect prediction system and terminal | |
| CN113611401A (en) | Perioperative blood management system and method | |
| CN115910263A (en) | PET/CT image report conclusion auxiliary generation method and device based on knowledge graph | |
| CN111524570A (en) | Ultrasonic follow-up patient screening method based on machine learning | |
| CN111128388A (en) | Value domain data matching method and device and related products | |
| CN112071431B (en) | Clinical path automatic generation method and system based on deep learning and knowledge graph | |
| CN115762721A (en) | Medical image quality control method and system based on computer vision technology | |
| CN112861535B (en) | Surgical classification coding method and system based on diagnosis and treatment data | |
| CN115458135A (en) | Bio-NER intelligent medical treatment triage system based on BGRU-Attention-CRF |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200522 |