CN112017790B - Electronic medical record screening method, device, equipment and medium based on countermeasure network - Google Patents
Electronic medical record screening method, device, equipment and medium based on countermeasure network Download PDFInfo
- Publication number
- CN112017790B CN112017790B CN202010941842.2A CN202010941842A CN112017790B CN 112017790 B CN112017790 B CN 112017790B CN 202010941842 A CN202010941842 A CN 202010941842A CN 112017790 B CN112017790 B CN 112017790B
- Authority
- CN
- China
- Prior art keywords
- data
- discrimination
- comprehensive
- discriminator
- initial
- 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.)
- Active
Links
Images
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
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- 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
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Epidemiology (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention relates to the field of machine learning in artificial intelligence, and discloses an electronic medical record screening method, device, equipment and medium based on an countermeasure network, wherein the method comprises the following steps: generating simulated misdiagnosis data through a first generator; generating simulated missed diagnosis data through a second generator; acquiring real normal data, and training an initial comprehensive discriminator by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data; after training, determining the trained initial comprehensive discriminant as a comprehensive discriminant; and processing the medical records to be screened by using the comprehensive discriminator to obtain the processing result of the medical records to be screened. The invention can solve the screening problem of the electronic medical records and reduce the medical record screening cost. The invention can be applied to the construction of smart cities.
Description
Technical Field
The present invention relates to the field of machine learning, and in particular, to a method, apparatus, device, and medium for screening electronic medical records based on an countermeasure network.
Background
The medical record is used for recording the treatment information generated in the diagnosis and treatment process of the patient and has important significance for the diagnosis and treatment of doctors. Along with the popularization of computer technology, medical records are gradually electronized, and the formed electronic medical records are stored in an electronic medical record library of a hospital.
However, due to different levels and experiences of doctors, recorded medical records have uneven quality, so that the electronic medical records are easy to archive wrongly, and hidden medical hidden dangers are generated. Thus, it is necessary to screen the electronic medical records and find out the problem medical records in time. If manual screening is used, high labor and time costs are incurred, greatly aggravating the operating costs of hospitals. Thus, there is a need to find a screening method that is not manually processed to screen the electronic medical records for problematic medical records.
Disclosure of Invention
Based on the foregoing, it is necessary to provide a method, a device, a computer device and a storage medium for screening electronic medical records based on an countermeasure network, so as to solve the problem of screening electronic medical records.
An electronic medical record screening method based on an countermeasure network, comprising:
generating simulated misdiagnosis data through a first generator;
generating simulated missed diagnosis data through a second generator;
acquiring real normal data, and training an initial comprehensive discriminant by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data;
after training, determining the trained initial comprehensive discriminant as a comprehensive discriminant;
And processing the medical record to be screened by using the comprehensive discriminator to obtain a processing result of the medical record to be screened.
An electronic medical record screening apparatus based on an countermeasure network, comprising:
the first generation module is used for generating simulated misdiagnosis data through the first generator;
the second generation module is used for generating simulated missed diagnosis data through a second generator;
the training module is used for acquiring real normal data and training an initial comprehensive discriminator by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data;
the determining discriminator module is used for determining the trained initial comprehensive discriminator as a comprehensive discriminator after training is finished;
and the screening module is used for processing the medical records to be screened by using the comprehensive discriminator to obtain the processing result of the medical records to be screened.
A computer device comprising a memory, a processor, and computer readable instructions stored in the memory and executable on the processor, which when executed by the processor implement the above-described countermeasure network-based electronic medical record screening method.
A computer readable storage medium storing computer readable instructions that when executed by a processor implement the above-described countermeasure network-based electronic medical record screening method.
According to the electronic medical record screening method, device, computer equipment and storage medium based on the countermeasure network, the first generator is used for generating the simulated misdiagnosis data so as to generate a large amount of simulated misdiagnosis data which is close to reality, and the discrimination capability of the initial comprehensive discriminator on the misdiagnosis data is improved. And generating simulated missed diagnosis data through a second generator so as to generate a large amount of simulated missed diagnosis data which is close to reality, and improving the discrimination capability of the initial comprehensive discriminator on the missed diagnosis data. And acquiring real normal data, training an initial comprehensive discriminator by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data, wherein the discrimination capability of the model is greatly improved because the model is trained by enough data. After training, determining the trained initial comprehensive discriminant as a comprehensive discriminant to obtain a discriminant capable of screening medical records. And processing the medical records to be screened by using the comprehensive discriminator to obtain a processing result of the medical records to be screened, and screening whether the medical records are normal or not by using the comprehensive discriminator, so that the accuracy rate and the efficiency of medical record screening can be improved, and the cost of medical record screening is reduced. The invention can solve the screening problem of the electronic medical record. The intelligent medical system can be applied to the intelligent medical field of the smart city, thereby promoting the construction of the smart city.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments of the present invention will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic view of an application environment of an electronic medical record screening method based on an countermeasure network according to an embodiment of the invention;
FIG. 2 is a flow chart of a method for screening electronic medical records based on an countermeasure network according to an embodiment of the invention;
FIG. 3 is a flow chart of a method for screening electronic medical records based on an countermeasure network according to an embodiment of the invention;
FIG. 4 is a flow chart of a method for screening electronic medical records based on an countermeasure network according to an embodiment of the invention;
FIG. 5 is a flow chart of a method for screening electronic medical records based on an countermeasure network according to an embodiment of the invention;
FIG. 6 is a flow chart of a method for screening electronic medical records based on an countermeasure network according to an embodiment of the invention;
FIG. 7 is a flow chart of a method for screening electronic medical records based on an countermeasure network according to an embodiment of the invention;
FIG. 8 is a schematic diagram of an electronic medical record screening apparatus based on an countermeasure network according to an embodiment of the present invention;
FIG. 9 is a schematic diagram of a computer device in accordance with an embodiment of the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The electronic medical record screening method based on the countermeasure network provided by the embodiment can be applied to an application environment as shown in fig. 1, wherein a client communicates with a server. Clients include, but are not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The server may be implemented by a stand-alone server or a server cluster formed by a plurality of servers.
In an embodiment, as shown in fig. 2, an electronic medical record screening method based on an antagonism network is provided, and the method is applied to the server in fig. 1 for illustration, and includes the following steps:
S10, generating simulated misdiagnosis data through a first generator.
In this embodiment, the first generator is trained by the countermeasure generation network (Generative Adversarial Network, GAN) to obtain a simulation generator for generating simulated misdiagnosis data. The first generator can generate a large amount of simulated misdiagnosis data which is close to reality, so that the initial comprehensive discriminator is ensured to have enough misdiagnosis training data, and the discrimination capability of the misdiagnosis data is improved. The simulated misdiagnosis data is one of medical record data and refers to medical records with misdiagnosis problems.
S20, generating simulation missed diagnosis data through a second generator.
Similarly, the second generator is trained via the countermeasure generation network (Generative Adversarial Network, GAN) to obtain a simulation generator for generating simulated missed diagnosis data. The second generator can generate a large amount of simulated missed diagnosis data which is close to reality, so that the initial comprehensive discriminator is ensured to have enough missed diagnosis training data, and the discrimination capability of the missed diagnosis data is improved. The simulated missed diagnosis data is one of medical record data and refers to medical records with the missed diagnosis problem.
S30, acquiring real normal data, and training an initial comprehensive discriminant by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data.
The true normal data refers to the case history data which is true and has no missed diagnosis and misdiagnosis condition. Here, the initial integrated arbiter is a three-class classifier. When the initial comprehensive discriminant is trained, the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data are used for training, so that the discrimination capability of the initial comprehensive discriminant on the three types of data can be greatly improved, and the types of medical records can be accurately distinguished. Specifically, during training, the discrimination data of the initial comprehensive discriminator can be returned to the first generator and the second generator, so that the association relation (based on the loss function) among the first generator, the second generator and the initial comprehensive discriminator is improved, and the discrimination capability of the initial comprehensive discriminator is further improved.
And S40, after training is finished, determining the trained initial comprehensive discriminant as a comprehensive discriminant.
And after training, the discrimination data of the initial comprehensive discriminator converges. At this time, the trained initial integrated classifier may be determined as an integrated classifier. The comprehensive discriminant can be used to discriminate the type of medical record data. The initial comprehensive discriminant fuses the simulation data (including the simulation missed diagnosis data and the simulation misdiagnosis data) of the first generator and the second generator, and the final comprehensive discriminant has good discrimination capability and can accurately distinguish the category of the electronic medical record.
S50, processing the medical record to be screened by using the comprehensive discriminator to obtain a processing result of the medical record to be screened.
In this embodiment, the medical records to be screened refer to medical records that need to be screened. And processing the medical records to be screened by using the comprehensive discriminator, so that a processing result of the medical records to be screened can be obtained. The processing results are three kinds of medical records, namely medical record of missed diagnosis, medical record of misdiagnosis and normal medical record. Wherein, the missed diagnosis medical record and the misdiagnosis medical record belong to abnormal medical records (also called problem medical records). Because the comprehensive discriminator provided by the embodiment has good discriminating capability, normal medical records and abnormal medical records can be distinguished with high precision, the processing time and the cost of medical record screening are greatly reduced, and the operation cost of hospitals is reduced.
In the steps S10-S50, the first generator is used for generating simulated misdiagnosis data so as to generate a large amount of simulated misdiagnosis data which is close to reality, and the discrimination capability of the initial comprehensive discriminator on the misdiagnosis data is improved. And generating simulated missed diagnosis data through a second generator so as to generate a large amount of simulated missed diagnosis data which is close to reality, and improving the discrimination capability of the initial comprehensive discriminator on the missed diagnosis data. And acquiring real normal data, training an initial comprehensive discriminator by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data, wherein the discrimination capability of the model is greatly improved because the model is trained by enough data. After training, determining the trained initial comprehensive discriminant as a comprehensive discriminant to obtain a discriminant capable of screening medical records. And processing the medical records to be screened by using the comprehensive discriminator to obtain a processing result of the medical records to be screened, and screening whether the medical records are normal or not by using the comprehensive discriminator, so that the accuracy rate and the efficiency of medical record screening can be improved, and the cost of medical record screening is reduced.
Optionally, as shown in fig. 3, step S10, before the generating, by the first generator, the simulated misdiagnosis data further includes:
s101, in a first antagonistic neural network, a first initial generator receives first random noise and generates first analog data;
s102, a first discriminator receives real misdiagnosis data with a label of 1 and generates first real discrimination data; the first discriminator receives the first analog data with the tag of 0 and generates first analog discrimination data;
s103, calculating a first discrimination loss value of the first discriminator according to the first real discrimination data and the first simulation discrimination data; calculating a first generation loss value of the first initial generator according to the first simulation discrimination data;
s104, updating the first discriminator according to the first discrimination loss value, and updating the first initial generator according to the first generation loss value;
s105, repeating the step of updating the first discriminator and the step of updating the first initial generator until a first preset termination condition is met, wherein the first preset termination condition is that the first simulation discrimination data is in a first preset range;
S106, determining the first initial generator meeting a first preset termination condition as the first generator.
In this embodiment, the first antagonistic neural network includes a first initial generator and a first arbiter. The first random noise may be generated by a random algorithm. In the training initial stage, when the first initial generator processes first random noise, the generated first analog data and the true misdiagnosis data have large difference and are easily identified by the first discriminator. At this time, the generator obtains a larger first generation loss value, and adjusts the calculation parameters in the first initial generator according to the first generation loss value, so that the data generated by the first initial generator gradually approaches to the real misdiagnosis data. Meanwhile, if the first discriminator makes a false judgment when judging the first analog data, the first discriminator can also obtain a larger first judging loss value, and the calculation parameters in the first discriminator are adjusted according to the first judging loss value, so that the distinguishing capability of the first analog data and the true false diagnosis data is stronger.
Repeating the step of updating the first arbiter refers to repeating the steps S101-S104 associated with the first arbiter. Repeating the step of updating the first generator refers to repeating the steps of steps S101-S104 associated with the first initial generator. The updating steps of the first arbiter and the first initial generator are performed simultaneously.
When the first simulation discrimination data is in the first preset range, the first antagonistic neural network can be considered to meet the first preset termination condition. For example, when the first discriminator judges that the first analog data of the first analog data is 0.5, it is difficult for the first discriminator to judge whether the first analog data output from the first initial generator is true. That is, the first simulation data generated by the first generator closely approximates the true misdiagnosis data, and the first antagonistic neural network reaches convergence.
In steps S101-S106, in the first antagonistic neural network, the first initial generator receives the first random noise and generates first analog data, where the first initial generator continuously generates new first analog data. The first discriminator receives real misdiagnosis data with a label of 1 and generates first real discrimination data; the first discriminator receives the first analog data with the label of 0 and generates first analog discrimination data, and the first discriminator discriminates the real misdiagnosis data and the first analog data at the same time, so that discrimination capability of the first discriminator on the misdiagnosis data can be improved. Calculating a first discrimination loss value of the first discriminator according to the first real discrimination data and the first simulation discrimination data; and calculating a first generation loss value of the first initial generator according to the first simulation judging data, and calculating a loss value which can be used for updating model parameters. Updating the first discriminator according to the first discrimination loss value, updating the first initial generator according to the first generation loss value, and gradually updating parameters of respective models through the loss value to improve the accuracy of the models. Repeating the step of updating the first discriminator and the step of updating the first initial generator until a first preset termination condition is satisfied, wherein the first preset termination condition is that the first simulation discrimination data is in a first preset range so as to complete training of a model. Determining the first initial generator meeting a first preset termination condition as the first generator to obtain a first generator usable for generating first simulation data.
Optionally, as shown in fig. 4, in step S103, the calculating a first discrimination loss value of the first discriminator according to the first real discrimination data and the first analog discrimination data includes:
s1031, processing the first real discrimination data and the first simulation discrimination data through a first discrimination loss function to generate a first discrimination loss value, wherein the first discrimination loss function is as follows:
wherein x is r1 Representing the real misdiagnosis data D 1 (x) Representing the first real discrimination data, E is a desired calculation symbol, x f1 Representing the first analog data, z 1 Represents the first random noise, G 1 (z 1 ) Representing the first analog data, D 1 (G 1 (z 1 ) A) represents the first discrimination loss value,
representing the first discrimination loss value;
in step S103, the calculating a first generation loss value of the first initial generator according to the first simulation discrimination data includes:
s1032, processing the first simulation discrimination data through a first generation loss function to generate the first generation loss value, wherein the first generation loss function is as follows:
wherein D is 1 (x f1 ) For the first simulation discrimination data, D 3 (x f1 ) Discrimination data generated after discrimination of the first analog data for the initial comprehensive discriminator, alpha being a super parameter, 
A penalty value is generated for the first.
Here, G 1 Refers to a first initial generator, D 1 Refers to a first arbiter. The first initial generator is used for generating first simulation data, and the first discriminator is used for discriminating the first simulation data and generating first simulation discrimination data; and the method is also used for judging the real misdiagnosis data and generating first real judging data. Training of the first initial generator and the first arbiter is a process of mutual antagonism. Alpha is a super parameter and can be set before model training.
After calculating the first generation loss value, the inclusion of D can be increased 3 (initial comprehensive discriminator) discriminating loss term (α·log d) of data 3 (x f1 ))。α·logD 3 (x f1 ) By adding the data, the distribution of the misdiagnosis data generated by the first generator can meet the requirements of actual scenes.
Optionally, as shown in fig. 5, step S20, before the generating, by the second generator, the simulated missed diagnosis data further includes:
s201, in a second antagonistic neural network, a second initial generator receives second random noise and generates second analog data;
s202, a second discriminator receives real missed diagnosis data with a label of 1 and generates second real discrimination data; the second discriminator receives the second analog data with the tag of 0 and generates second analog discrimination data;
S203, calculating a second discrimination loss value of the second discriminator according to the second real discrimination data and the second simulation discrimination data; calculating a second generation loss value of the second initial generator according to the second simulation discrimination data;
s204, updating the second discriminator according to the second discrimination loss value, and updating the second initial generator according to the second generation loss value;
s205, repeating the step of updating the second discriminator and the step of updating the second initial generator until a second preset termination condition is met, wherein the second preset termination condition is that the second simulation discrimination data is in a second preset range;
s206, determining the second initial generator meeting a second preset termination condition as the second generator.
In this embodiment, the second antagonistic neural network includes a second initial generator and a second arbiter. The second random noise may be generated by a random algorithm. In the training initial stage, when the second initial generator processes the second random noise, the generated second simulation data has larger difference from the real missed diagnosis data, and is easily identified by the second discriminator. At this time, the generator can obtain two larger second generation loss values, and adjust the calculation parameters in the second initial generator according to the second generation loss values, so that the data generated by the second initial generator gradually approaches to the real missed diagnosis data. Meanwhile, if the second discriminator makes a false judgment when judging the second analog data, the second discriminator can also obtain two larger second judging loss values, and the calculation parameters in the second discriminator are adjusted according to the second judging loss values, so that the second analog data and the real missed diagnosis data can be distinguished more strongly.
Repeating the step of updating the second arbiter refers to repeating the steps of steps S201-S204 associated with the second arbiter. Repeating the step of updating the second generator refers to repeating the steps of steps S201-S204 associated with the second initial generator. The updating steps of the second arbiter and the second initial generator are performed simultaneously.
When the second simulation discrimination data is in the second preset range, the second antagonistic neural network can be considered to meet the second preset termination condition. For example, when the second discriminator judges that the second analog data of the second analog data is 0.5, it is difficult for the second discriminator to judge whether the second analog data output from the second initial generator is true. That is, the second simulation data generated by the second generator closely approximates the true missed data, at which time the second antagonistic neural network has reached convergence.
In steps S201-S206, in the second antagonistic neural network, the second initial generator receives the second random noise and generates second analog data, where the second initial generator continuously generates new second analog data. The second discriminator receives the real missed diagnosis data with the label of 1 and generates second real discrimination data; the second discriminator receives the second analog data with the label of 0 and generates second analog discrimination data, and the second discriminator discriminates the real missed diagnosis data and the second analog data at the same time, so that discrimination capability of the second discriminator on the missed diagnosis data can be improved. Calculating a second discrimination loss value of the second discriminator according to the second real discrimination data and the second simulation discrimination data; and calculating a second generation loss value of the second initial generator according to the second simulation discrimination data, and calculating a loss value which can be used for updating model parameters. Updating the second discriminator according to the second discrimination loss value, updating the second initial generator according to the second generation loss value, and gradually updating parameters of respective models through the loss value to improve the accuracy of the models. Repeating the step of updating the second discriminator and the step of updating the second initial generator until a second preset termination condition is satisfied, wherein the second preset termination condition is that the second simulation discrimination data is in a second preset range, so as to complete training of the model. And determining the second initial generator meeting a second preset termination condition as the second generator to obtain a second generator which can be used for generating second simulation data.
Optionally, as shown in fig. 6, in step S203, the calculating a second discrimination loss value of the second discriminator according to the second real discrimination data and the second analog discrimination data includes:
s2031, processing the second real discrimination data and the second simulation discrimination data through a second discrimination loss function to generate the second discrimination loss value, wherein the second discrimination loss function is as follows:
wherein x is r2 Representing the real missed diagnosis data D 2 (x) Representing the second real discrimination data, E is a desired calculation symbol, x f2 Representing the second analog data, z 2 Representing the second random noise, G 2 (z 2 ) Representing the second analog data, D 2 (G 2 (z 2 ) A) represents the second discrimination loss value,
representing the second discrimination loss value;
in step S203, the calculating a second generation loss value of the second initial generator according to the second simulation discrimination data includes:
s2032, processing the second simulation discrimination data through a second generation loss function to generate a second generation loss value, wherein the second generation loss function is as follows:
wherein D is 2 (x f2 ) D for the second simulation discrimination data 3 (x f2 ) Discrimination data generated after discrimination of the second analog data is performed for the initial comprehensive discriminator, beta is a super parameter, 
A loss value is generated for the second.
Here, G 2 Refers to a second initial generator, D 2 Refers to a second arbiter. The second initial generator is used for generating second simulation data, and the second discriminator is used for discriminating the second simulation data and generating second simulation discrimination data; and the method is also used for judging the real missed diagnosis data and generating second real judgment data. Training of the second initial generator and the second arbiter is two mutually opposing processes. Beta is a super parameter, and can be set before model training.
After calculating the second generation loss value, the inclusion of D can be increased 3 (initial comprehensive discriminator) discriminating loss term (β -log D) of data 3 (x f2 ))。β·logD 3 (x f2 ) The distribution of missed diagnosis data generated by the second generator can be made to meet the requirements of actual scenes.
Optionally, as shown in fig. 7, step S30, that is, the acquiring real normal data, trains the initial comprehensive discriminant using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data, includes:
s301, judging the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data by using the initial comprehensive judging device to generate comprehensive judging data, wherein the comprehensive judging data comprises a missed diagnosis rate, a misdiagnosis rate and a normal rate;
S302, calculating a comprehensive discrimination loss value of the initial comprehensive discriminator according to the missed diagnosis rate, the misdiagnosis rate and the normal rate;
s303, updating the initial comprehensive discriminant according to the comprehensive discriminant loss value;
s304, repeating the step of updating the initial comprehensive discriminant until the comprehensive discriminant loss value meets the preset convergence condition.
In this embodiment, the missed diagnosis rate in the comprehensive discrimination data refers to the ratio of the number of missed diagnosis medical records determined by the initial comprehensive discriminator to the total number of discrimination medical records. For example, if the total number of medical records is 100 and the number of missed medical records determined by the initial comprehensive discriminator is 4, the missed diagnosis rate in the comprehensive discrimination data is 4%. Likewise, the misdiagnosis rate in the comprehensive discrimination data refers to the ratio of the number of misdiagnosed medical records judged by the initial comprehensive discriminator to the total number of discriminated medical records; the normal rate in the comprehensive discrimination data refers to the ratio of the number of normal medical records judged by the initial comprehensive discriminator to the total number of discriminated medical records.
When calculating the comprehensive discrimination loss value, besides the initial comprehensive discriminator generates comprehensive discrimination data, the true missed diagnosis rate, false diagnosis rate and normal rate are combined.
The step of repeatedly updating the initial comprehensive discriminant refers to repeatedly performing steps S301-S303.
The preset convergence condition may refer to that the integrated discrimination loss value approaches a certain value.
Optionally, step S302, that is, calculating the integrated discrimination loss value of the initial integrated discriminator according to the missed diagnosis rate, the misdiagnosis rate and the normal rate includes:
processing the missed diagnosis rate, the misdiagnosis rate and the normal rate through a comprehensive loss function to generate the comprehensive discrimination loss value, wherein the comprehensive loss function is as follows:
where, when k=1, y 1 For the true rate of missed diagnosis,
the missed diagnosis rate in the comprehensive discrimination data is determined; when k=2, y 2 For true misdiagnosis rate, +.>
The misdiagnosis rate in the comprehensive discrimination data is used; when k=3, y 3 For true normal rate, ++>
The normal rate in the comprehensive discrimination data is determined; />
And judging the loss value for the synthesis.
Here, the integrated discrimination loss value can be calculated by the integrated loss function. When the initial comprehensive discriminant converges, the comprehensive discriminant loss value approaches a certain constant value.
It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic, and should not limit the implementation process of the embodiment of the present invention.
In an embodiment, an electronic medical record screening device based on a countermeasure network is provided, where the electronic medical record screening device based on the countermeasure network corresponds to the electronic medical record screening method based on the countermeasure network in the foregoing embodiment one by one. As shown in fig. 8, the electronic medical record screening apparatus based on the countermeasure network includes a first generating module 10, a second generating module 20, a training module 30, a determination discriminator module 40, and a screening module 50. The functional modules are described in detail as follows:
a first generation module 10, configured to generate simulated misdiagnosis data through a first generator;
a second generation module 20, configured to generate simulated missed diagnosis data through a second generator;
the training module 30 is configured to acquire real normal data, and train the initial comprehensive discriminant using the real normal data, the simulated misdiagnosis data, and the simulated missed diagnosis data;
a determining discriminant module 40, configured to determine the trained initial comprehensive discriminant as a comprehensive discriminant after training is completed;
and the screening module 50 is used for processing the medical records to be screened by using the comprehensive discriminator to obtain the processing result of the medical records to be screened.
Optionally, the first generating module 10 includes:
Generating a first analog data unit for receiving a first random noise and generating first analog data in a first antagonistic neural network;
the first judging unit is used for receiving the real misdiagnosis data with the label of 1 by the first judging device and generating first real judging data; the first discriminator receives the first analog data with the tag of 0 and generates first analog discrimination data;
a first loss value calculation unit configured to calculate a first loss value of the first discriminator according to the first real discrimination data and the first analog discrimination data; calculating a first generation loss value of the first initial generator according to the first simulation discrimination data;
a first updating unit, configured to update the first arbiter according to the first discrimination loss value, and update the first initial generator according to the first generation loss value;
a first iteration updating unit, configured to repeat the step of updating the first arbiter and the step of updating the first initial generator until a first preset termination condition is satisfied, where the first preset termination condition is that the first simulation discrimination data is in a first preset range;
A first generator unit is determined for determining the first initial generator satisfying a first preset termination condition as the first generator.
Optionally, calculating the first loss value unit includes:
a first discrimination loss value calculating unit, configured to process the first real discrimination data and the first analog discrimination data through a first discrimination loss function, and generate the first discrimination loss value, where the first discrimination loss function is:
wherein x is r1 Representing the real misdiagnosis data D 1 (x) Representing the first real discrimination data, E is a desired calculation symbol, x f1 Representing the first analog data, z 1 Represents the first random noise, G 1 (z 1 ) Representing the first analog data, D 1 (G 1 (z 1 ) A) represents the first discrimination loss value,
representing the first discrimination loss value;
a first generation loss value calculating unit, configured to process the first simulation discrimination data through a first generation loss function, and generate the first generation loss value, where the first generation loss function is:
wherein D is 1 (x f1 ) For the first simulation discrimination data, D 3 (x f1 ) Discrimination data generated after discrimination of the first analog data for the initial comprehensive discriminator, alpha being a super parameter, 
A penalty value is generated for the first.
Optionally, the second generating module 20 includes:
generating a second analog data unit for receiving a second random noise and generating second analog data in a second antagonistic neural network;
the second judging unit is used for receiving the real missed diagnosis data with the label of 1 by the second judging device and generating second real judging data; the second discriminator receives the second analog data with the tag of 0 and generates second analog discrimination data;
a second loss value calculating unit configured to calculate a second discrimination loss value of the second discriminator based on the second real discrimination data and the second analog discrimination data; calculating a second generation loss value of the second initial generator according to the second simulation discrimination data;
a second updating unit, configured to update the second arbiter according to the second discrimination loss value, and update the second initial generator according to the second generation loss value;
a second iteration updating unit, configured to repeat the step of updating the second identifier and the step of updating the second initial generator until a second preset termination condition is satisfied, where the second preset termination condition is that the second simulation discrimination data is in a second preset range;
And a second generator unit for determining the second initial generator satisfying a second preset termination condition as the second generator.
Optionally, calculating the second loss value unit includes:
a second discrimination loss value calculating unit, configured to process the second real discrimination data and the second simulated discrimination data through a second discrimination loss function, and generate the second discrimination loss value, where the second discrimination loss function is:
wherein x is r2 Representing the real missed diagnosis data D 2 (x) Representing the second real discrimination data, E is a desired calculation symbol, x f2 Representing the second analog data, z 2 Representing the second random noise, G 2 (z 2 ) Representing the second analog data, D 2 (G 2 (z 2 ) A) represents the second discrimination loss value,
representing the second discrimination loss value;
a second generation loss value calculating unit, configured to process the second simulation discrimination data through a second generation loss function, and generate the second generation loss value, where the second generation loss function is:
wherein D is 2 (x f2 ) D for the second simulation discrimination data 3 (x f2 ) Discrimination data generated after discrimination of the second analog data is performed for the initial comprehensive discriminator, beta is a super parameter, 
A loss value is generated for the second.
Optionally, training module 30 includes:
generating comprehensive judging data unit, which is used for judging the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data by using the initial comprehensive judging device, and generating comprehensive judging data, wherein the comprehensive judging data comprises a missed diagnosis rate, a misdiagnosis rate and a normal rate;
a comprehensive discrimination loss value generation unit for calculating a comprehensive discrimination loss value of the initial comprehensive discriminator according to the missed diagnosis rate, the misdiagnosis rate and the normal rate;
an updating initial comprehensive discriminant unit for updating the initial comprehensive discriminant according to the comprehensive discriminant loss value;
and the iterative updating initial comprehensive discriminator unit is used for repeatedly updating the initial comprehensive discriminator until the comprehensive discriminating loss value meets the preset convergence condition.
Optionally, the integrated discrimination loss value generating unit is further configured to process the missed diagnosis rate, the misdiagnosis rate and the normal rate through an integrated loss function, and generate the integrated discrimination loss value, where the integrated loss function is:
where, when k=1, y 1 For the true rate of missed diagnosis,
for the comprehensive discrimination data The missed diagnosis rate in the process; when k=2, y 2 For true misdiagnosis rate, +.>
The misdiagnosis rate in the comprehensive discrimination data is used; when k=3, y 3 For true normal rate, ++>
The normal rate in the comprehensive discrimination data is determined; />
And judging the loss value for the synthesis.
Specific limitations regarding the countermeasure network-based electronic medical record screening apparatus may be found in the above description of the countermeasure network-based electronic medical record screening method, and will not be described in detail herein. The modules in the electronic medical record screening device based on the countermeasure network can be realized in whole or in part by software, hardware and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, and the internal structure of which may be as shown in fig. 9. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer readable instructions, and a database. The internal memory provides an environment for the execution of an operating system and computer-readable instructions in a non-volatile storage medium. The database of the computer equipment is used for storing the data related to the electronic medical record screening method. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer readable instructions, when executed by a processor, implement a method of electronic medical record screening based on an countermeasure network.
In one embodiment, a computer device is provided that includes a memory, a processor, and computer readable instructions stored on the memory and executable on the processor, when executing the computer readable instructions, performing the steps of:
generating simulated misdiagnosis data through a first generator;
generating simulated missed diagnosis data through a second generator;
acquiring real normal data, and training an initial comprehensive discriminant by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data;
after training, determining the trained initial comprehensive discriminant as a comprehensive discriminant;
and processing the medical record to be screened by using the comprehensive discriminator to obtain a processing result of the medical record to be screened.
In one embodiment, one or more computer-readable storage media are provided having computer-readable instructions stored thereon, the readable storage media provided by the present embodiment including non-volatile readable storage media and volatile readable storage media. The readable storage medium has stored thereon computer readable instructions which when executed by one or more processors perform the steps of:
generating simulated misdiagnosis data through a first generator;
Generating simulated missed diagnosis data through a second generator;
acquiring real normal data, and training an initial comprehensive discriminant by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data;
after training, determining the trained initial comprehensive discriminant as a comprehensive discriminant;
and processing the medical record to be screened by using the comprehensive discriminator to obtain a processing result of the medical record to be screened.
Those skilled in the art will appreciate that implementing all or part of the processes of the methods of the embodiments described above may be accomplished by instructing the associated hardware by computer readable instructions stored on a non-transitory computer readable storage medium, which when executed may comprise processes of embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.
Claims (10)
1. An electronic medical record screening method based on an countermeasure network is characterized by comprising the following steps:
generating simulated misdiagnosis data through a first generator;
generating simulated missed diagnosis data through a second generator;
Acquiring real normal data, and training an initial comprehensive discriminant by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data;
after training, determining the trained initial comprehensive discriminant as a comprehensive discriminant;
and processing the medical record to be screened by using the comprehensive discriminator to obtain a processing result of the medical record to be screened.
2. The method for screening an electronic medical record based on an countermeasure network of claim 1, further comprising, prior to generating the simulated misdiagnosis data by the first generator:
in the first antagonistic neural network, a first initial generator receives first random noise and generates first analog data;
the first discriminator receives real misdiagnosis data with a label of 1 and generates first real discrimination data; the first discriminator receives the first analog data with the tag of 0 and generates first analog discrimination data;
calculating a first discrimination loss value of the first discriminator according to the first real discrimination data and the first simulation discrimination data; calculating a first generation loss value of the first initial generator according to the first simulation discrimination data;
Updating the first discriminator according to the first discrimination loss value, and updating the first initial generator according to the first generation loss value;
repeating the step of updating the first discriminator and the step of updating the first initial generator until a first preset termination condition is satisfied, wherein the first preset termination condition is that the first simulation discrimination data is in a first preset range;
determining the first initial generator meeting a first preset termination condition as the first generator.
3. The method of claim 2, wherein the calculating a first discrimination loss value for the first discriminator based on the first true discrimination data and the first simulated discrimination data comprises:
processing the first real discrimination data and the first simulation discrimination data through a first discrimination loss function to generate the first discrimination loss value, wherein the first discrimination loss function is as follows:
wherein x is r1 Representing the real misdiagnosis data D 1 (x) Representing the first real discrimination data, E is a desired calculation symbol, x f1 Representing the first analog data, z 1 Represents the first random noise, G 1 (z 1 ) Representing the first analog data, D 1 (G 1 (z 1 ) A) represents the first discrimination loss value,
representing the first discrimination loss value;
the calculating a first generation loss value of the first initial generator according to the first simulation discrimination data includes:
processing the first simulation discrimination data through a first generation loss function to generate the first generation loss value, wherein the first generation loss function is as follows:
wherein D is 1 (x f1 ) For the first simulation discrimination data, D 3 (x f1 ) Discrimination data generated after discrimination of the first analog data for the initial comprehensive discriminator, alpha being a super parameter,
a penalty value is generated for the first.
4. The method for screening an electronic medical record based on an countermeasure network of claim 1, wherein prior to generating the simulated missed diagnosis data by the second generator, further comprising:
in a second antagonistic neural network, a second initial generator receives a second random noise and generates second analog data;
the second discriminator receives the real missed diagnosis data with the label of 1 and generates second real discrimination data; the second discriminator receives the second analog data with the tag of 0 and generates second analog discrimination data;
Calculating a second discrimination loss value of the second discriminator according to the second real discrimination data and the second simulation discrimination data; calculating a second generation loss value of the second initial generator according to the second simulation discrimination data;
updating the second discriminator according to the second discrimination loss value, and updating the second initial generator according to the second generation loss value;
repeating the step of updating the second discriminator and the step of updating the second initial generator until a second preset termination condition is satisfied, wherein the second preset termination condition is that the second simulation discrimination data is in a second preset range;
and determining the second initial generator meeting a second preset termination condition as the second generator.
5. The method of claim 4, wherein calculating a second discrimination loss value for the second discriminator based on the second true discrimination data and the second simulated discrimination data, comprises:
processing the second real discrimination data and the second simulation discrimination data through a second discrimination loss function to generate a second discrimination loss value, wherein the second discrimination loss function is as follows:
Wherein x is r2 Representing the real missed diagnosis data D 2 (x) Representing the second real discrimination data, E is a desired calculation symbol, x f2 Representing the second analog data, z 2 Representing the second random noise, G 2 (z 2 ) Representing the second analog data, D 2 (G 2 (z 2 ) A) represents the second discrimination loss value,
representing the second discrimination loss value;
the calculating a second generation loss value of the second initial generator according to the second simulation discrimination data includes:
processing the second simulation discrimination data through a second generation loss function to generate a second generation loss value, wherein the second generation loss function is as follows:
wherein D is 2 (x f2 ) D for the second simulation discrimination data 3 (x f2 ) Discrimination data generated after discrimination of the second analog data is performed for the initial comprehensive discriminator, beta is a super parameter,
a loss value is generated for the second.
6. The method of claim 1, wherein the obtaining real normal data, training the initial comprehensive discriminant using the real normal data, the simulated misdiagnosis data, and the simulated missed diagnosis data, comprises:
the initial comprehensive discriminator is used for discriminating the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data to generate comprehensive discrimination data, wherein the comprehensive discrimination data comprises a missed diagnosis rate, a misdiagnosis rate and a normal rate;
Calculating the comprehensive discrimination loss value of the initial comprehensive discriminator according to the missed diagnosis rate, the misdiagnosis rate and the normal rate;
updating the initial comprehensive discriminant according to the comprehensive discriminant loss value;
repeating the step of updating the initial comprehensive discriminant until the comprehensive discriminant loss value meets the preset convergence condition.
7. The method for screening an electronic medical record based on an countermeasure network of claim 6, wherein the calculating the integrated discrimination loss value of the initial integrated discriminator based on the missed diagnosis rate, the misdiagnosis rate, and the normal rate includes:
processing the missed diagnosis rate, the misdiagnosis rate and the normal rate through a comprehensive loss function to generate the comprehensive discrimination loss value, wherein the comprehensive loss function is as follows:
where, when k=1, y 1 For the true rate of missed diagnosis,
the missed diagnosis rate in the comprehensive discrimination data is determined; when k=2, y 2 For true misdiagnosis rate, +.>
The misdiagnosis rate in the comprehensive discrimination data is used; when k=3, y 3 For true normal rate, ++>
The normal rate in the comprehensive discrimination data is determined; />
And judging the loss value for the synthesis.
8. An electronic medical record screening apparatus based on an countermeasure network, comprising:
The first generation module is used for generating simulated misdiagnosis data through the first generator;
the second generation module is used for generating simulated missed diagnosis data through a second generator;
the training module is used for acquiring real normal data and training an initial comprehensive discriminator by using the real normal data, the simulated misdiagnosis data and the simulated missed diagnosis data;
the determining discriminator module is used for determining the trained initial comprehensive discriminator as a comprehensive discriminator after training is finished;
and the screening module is used for processing the medical records to be screened by using the comprehensive discriminator to obtain the processing result of the medical records to be screened.
9. A computer device comprising a memory, a processor, and computer readable instructions stored in the memory and executable on the processor, wherein the processor, when executing the computer readable instructions, implements the countermeasure network-based electronic medical record screening method of any one of claims 1 to 7.
10. One or more readable storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the countermeasure network-based electronic medical record screening method of any of claims 1 to 7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010941842.2A CN112017790B (en) | 2020-09-09 | 2020-09-09 | Electronic medical record screening method, device, equipment and medium based on countermeasure network |
| PCT/CN2020/124219 WO2021151326A1 (en) | 2020-09-09 | 2020-10-28 | Electronic medical record screening method and apparatus based on adversarial network, and device and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010941842.2A CN112017790B (en) | 2020-09-09 | 2020-09-09 | Electronic medical record screening method, device, equipment and medium based on countermeasure network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112017790A CN112017790A (en) | 2020-12-01 |
| CN112017790B true CN112017790B (en) | 2023-06-20 |
Family
ID=73521424
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010941842.2A Active CN112017790B (en) | 2020-09-09 | 2020-09-09 | Electronic medical record screening method, device, equipment and medium based on countermeasure network |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112017790B (en) |
| WO (1) | WO2021151326A1 (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109003678A (en) * | 2018-06-12 | 2018-12-14 | 清华大学 | A kind of generation method and system emulating text case history |
| CN109073630A (en) * | 2016-02-23 | 2018-12-21 | 诺尔有限公司 | Diagnostic method and the equipment for executing diagnostic method |
| CN110808095A (en) * | 2019-09-18 | 2020-02-18 | 平安科技（深圳）有限公司 | Method for identifying diagnosis result, method for training model, computer device and storage medium |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160180022A1 (en) * | 2014-12-18 | 2016-06-23 | Fortinet, Inc. | Abnormal behaviour and fraud detection based on electronic medical records |
| CN110060774B (en) * | 2019-04-29 | 2022-12-09 | 赵蕾 | Thyroid nodule identification method based on generative confrontation network |
| CN110910976A (en) * | 2019-10-12 | 2020-03-24 | 平安国际智慧城市科技股份有限公司 | Medical record detection method, device, equipment and storage medium |
-
2020
- 2020-09-09 CN CN202010941842.2A patent/CN112017790B/en active Active
- 2020-10-28 WO PCT/CN2020/124219 patent/WO2021151326A1/en active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109073630A (en) * | 2016-02-23 | 2018-12-21 | 诺尔有限公司 | Diagnostic method and the equipment for executing diagnostic method |
| CN109003678A (en) * | 2018-06-12 | 2018-12-14 | 清华大学 | A kind of generation method and system emulating text case history |
| CN110808095A (en) * | 2019-09-18 | 2020-02-18 | 平安科技（深圳）有限公司 | Method for identifying diagnosis result, method for training model, computer device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112017790A (en) | 2020-12-01 |
| WO2021151326A1 (en) | 2021-08-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113593611B (en) | Voice classification network training method and device, computing equipment and storage medium | |
| CN109102797B (en) | Speech recognition test method, device, computer equipment and storage medium | |
| CN112016318B (en) | Triage information recommendation method, device, equipment and medium based on interpretation model | |
| KR101901307B1 (en) | Method, apparatus and computer-readable recording medium for classifying classes using deep neural network with weighted fuzzy membership function | |
| CN112232951B (en) | Credit evaluation method, device, equipment and medium based on multi-dimensional cross feature | |
| CN115985513A (en) | Data processing method, device and equipment based on multigroup cancer typing | |
| CN110808095A (en) | Method for identifying diagnosis result, method for training model, computer device and storage medium | |
| CN109800147A (en) | A kind of test cases generation method and terminal device | |
| CN109871866B (en) | Model training method, device, equipment and medium for hospital infection prediction | |
| CN113627159B (en) | Training data determining method, device, medium and product of error correction model | |
| CN115062734A (en) | Wind control modeling method, device, equipment and medium capable of outputting explanatory information | |
| CN113254673B (en) | Data management method, system, equipment and medium based on knowledge graph | |
| CN107292320A (en) | System and its index optimization method and device | |
| CN112017790B (en) | Electronic medical record screening method, device, equipment and medium based on countermeasure network | |
| CN111275059A (en) | Image processing method and device and computer readable storage medium | |
| CN111191692B (en) | Data calculation method and device based on decision tree and computer equipment | |
| CN113782187A (en) | Index data processing method, related device and medium | |
| CN114170000A (en) | Credit card user risk category identification method, device, computer equipment and medium | |
| CN115222049A (en) | Problem solving layer selection method, system and device of multi-modal data model | |
| CN110780850B (en) | Requirement case auxiliary generation method and device, computer equipment and storage medium | |
| CN110533158A (en) | Model Construction method, system and non-volatile readable in computer recording medium | |
| CN117495191A (en) | Quality of service evaluation method, device, computer equipment and storage medium | |
| CN111656370A (en) | Detection method and verification platform of accelerator | |
| CN116578701A (en) | Text classification method, device, equipment and medium | |
| Toda et al. | Outlier Elimination Technique Using Deletion‐Imputation Iteration for Fault‐Prone Module Detection |
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 |