CN112669989A - Infectious disease transmission model construction method introducing expert knowledge - Google Patents
Infectious disease transmission model construction method introducing expert knowledge Download PDFInfo
- Publication number
- CN112669989A CN112669989A CN202110299554.6A CN202110299554A CN112669989A CN 112669989 A CN112669989 A CN 112669989A CN 202110299554 A CN202110299554 A CN 202110299554A CN 112669989 A CN112669989 A CN 112669989A
- Authority
- CN
- China
- Prior art keywords
- infectious disease
- model
- disease transmission
- transmission model
- expert knowledge
- 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
Images
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention discloses an infectious disease transmission model construction method introducing expert knowledge, which comprises two parts of constructing an infectious disease transmission model by utilizing the expert knowledge and optimizing the infectious disease transmission model by utilizing data; developers can introduce expert knowledge through an initialization interface of the system platform; defining a function related to the simulation of infectious disease transmission by using expert knowledge, and building an infectious disease transmission model framework; listing parameter sets to be determined in the infectious disease transmission model by using expert knowledge and initializing parameters of the infectious disease transmission model; inputting training data to an infectious disease propagation model; training data is deduced in the constructed infectious disease transmission model; and finally obtaining an optimal infectious disease transmission model. The invention can fuse expert knowledge with historical data and quickly train out an efficient infectious disease transmission model.
Description
Technical Field
The invention relates to an infectious disease propagation model construction method introducing expert knowledge, and belongs to the technical field of artificial intelligence.
Background
With the development of artificial intelligence technology, machine learning models have been widely used in various industries and have achieved great success. However, the deep learning model commonly used in the AI field is a black box model, and although the deep learning model has a very good effect due to the strong representation capability, the internal operation process thereof is difficult to be understood by people, and the knowledge summarized by human is difficult to be injected into the model. How to design a more efficient white box model has become one of the problems to be solved urgently in the development of the artificial intelligence industry. Most machine learning models today are completely data-driven, i.e., learn from existing historical data, without any injection of expert knowledge, which makes the learning process of the model more difficult, time-consuming, and difficult for humans to understand. How to introduce priori knowledge such as expert knowledge and the like into machine learning so as to construct a white-box model is one of important directions for promoting the development of the field of artificial intelligence.
When a machine learning model, especially a deep learning model, learns from historical data, due to lack of prior knowledge, the model is usually initialized randomly at initialization, and the derivation process is difficult to understand by people and reflect a real scene, for example, in a scene of analyzing infectious disease transmission, the machine learning model may randomly try different derivation modes, the result of which is often very different from the real situation, it is difficult to derive a causal relationship which can be understood by human, and a reasonable model may be obtained only after a large amount of training. In fact, in the field of epidemic diseases, various models have been proposed, such as SIR model, SEIR model. These models contain rich expert knowledge, and although they may not be optimal or the most reflective of real-world scenes, they have been widely used and have some rationality, and machine learning models cannot ignore their reference value.
However, it is difficult to combine the data-driven characteristics with expert knowledge in the current machine learning model. Therefore, it is important to design a semi-white-box model or even a white-box model that is more empirical with expert experience and historical data.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the problems and the defects in the prior art, the infectious disease propagation model construction method with the introduction of expert knowledge can combine human expert knowledge and data drive to jointly train to obtain the infectious disease propagation model, solves the problem that the traditional machine learning model cannot be used for both, can obtain the model more efficiently and quickly, and enables the obtained infectious disease propagation model to have more accurate prediction effect.
The technical scheme is as follows: an infectious disease transmission model construction method introducing expert knowledge, the technical contents related in the method are introduced as follows:
the expert knowledge comprises the general experience of human beings, or rules summarized for certain types of tasks and the like, and can be generalized. The machine learning method is obtained by training a constructed model through a machine learning algorithm; the machine learning method can be constructed based on a system platform, and the system comprises a data field and a machine learning field. The default data field comprises a parameter module used for setting parameters to be determined, a variable module used for setting variables in the calculation process and a recording module used for storing data. The default placement loss function module in the machine learning field is used for judging the quality of a model result, and the machine learning algorithm module is used for selecting a learning method. The input of the loss function module is the output result of the model and the real result in the historical data, and the output is the difference of the output result and the real result in a certain metric standard; the machine learning algorithm module can optimize the model according to the output of the loss function module; the system provides a default general learning approach, which generally does not require modification. The system allows developers to select different model structures and training algorithms according to personal experience and knowledge, and finally training is carried out through historical data to obtain the optimal model.
The infectious disease transmission model building method with introduced expert knowledge consists of mainly building infectious disease transmission model with expert knowledge and optimizing infectious disease transmission model with data. The developer can introduce expert knowledge through the initialization interface of the system platform. The system comprises a data field and a machine learning field. The data field comprises a parameter module used for setting parameters to be determined, a variable module used for setting variables in the calculation process, and a recording module used for storing data. The loss function module is placed in the machine learning field and used for judging the quality of the infectious disease transmission model result, and the machine learning algorithm module is used for selecting an infectious disease transmission model training method. The training procedure is described as follows:
step 101, defining functions related to simulating infectious disease transmission by using expert knowledge, and then building an infectious disease transmission model framework by using the functions.
Step 102, listing a parameter set to be determined in an infectious disease transmission model by using expert knowledge; the set of parameters to be determined is a subset of the set of parameters of the model for the spread of infectious disease, which parameters should be set as the parameters to be determined being decided by expert knowledge.
Step 103, initializing infectious disease transmission model parameters by using expert knowledge. When initializing the model of infectious disease transmission, the expert knowledge is used to set a reasonable initial value for each parameter, the value is usually not an optimal value but has a certain reference value, and the initialization value can be randomly set by the system.
Step 104, inputting training data to the model for infectious disease transmission. The training data comprises features (or sequences of actions) and result values.
And 105, deducing the training data in the constructed infectious disease transmission model, and learning the infectious disease transmission model from the training data. In the first round of optimization, the model of infectious disease transmission is optimized on the basis of parameters with initial values. In subsequent optimization, the model of infectious disease transmission is optimized based on the parameters of the previous round.
Step 106, the system determines whether a training termination condition is reached. If the termination condition is not met, returning to the step 105; if the termination condition is reached, proceed to step 107.
And step 107, finishing the training, and outputting the optimal infectious disease transmission model by the system.
The functions related to the spread of the infectious disease comprise an initialization function, an infection function, a diagnosis function, a latent death function and a disease course development function.
The loss function of the infectious disease propagation model is used for evaluating the error between the output result of the model and the real result, different loss functions need to be selected according to specific tasks and data, and common loss functions comprise a mean square error loss function, a cross entropy loss function and the like.
The model for infectious disease transmission can be solved by a gradient-free optimization algorithm or an evolution calculation method and a gradient-based solution method during optimization. Specifically, when a gradient-free optimization algorithm or an evolution calculation method is used, the system samples in a search space of parameters to be searched, each group of parameters is deduced in the constructed infectious disease transmission model, and errors between an output result and a real result of the infectious disease transmission model are evaluated by using a loss function. And the optimization algorithm continuously reduces the parameter searching space through the error value until the parameter space is an empty set, and the optimal infectious disease propagation model is obtained after the searching is finished. When the gradient-based solving method is used, training data are repeatedly deduced in the model, the error between the output result and the real result of the infectious disease transmission model is evaluated by using a loss function, and the parameter to be searched is iteratively optimized by using gradient information until the infectious disease transmission model is converged to obtain the optimal infectious disease transmission model.
The parameter to be optimized in the model for the transmission of infectious diseases is combined intoθUsing the set of parameters to be optimizedθThe model is constructed ash θ Including expert knowledge. The historical data of infectious disease transmission is characterized in thatx i()Corresponding to a result ofy i()WhereiniTo representmThe number of individual data samples. Recording the loss function asLThen the output value and data of model prediction are trueThe error in the value of the error is,
；
if a loss of mean square error function is used, the error is
；
From this error, minimizing the error using a machine learning optimization algorithm (using either a gradient-based optimization algorithm or a non-gradient optimization algorithm as appropriate) can be expressed as
；
Finally obtaining the optimal parameter setθ * Optimized machine learning modelh θ* 。
Has the advantages that: the current machine learning model is difficult to have the ability of utilizing expert knowledge and data driving at the same time. Compared with the prior art, the infectious disease propagation model construction method with the introduction of expert knowledge provided by the invention can fuse the expert knowledge with historical data and quickly train a high-efficiency infectious disease propagation model. The method can be used by developers in the field to integrate expert knowledge in a simple and convenient way, and simultaneously, the machine learning model can be trained by directly inputting data.
Drawings
FIG. 1 is a schematic flow diagram of the process of the present invention;
FIG. 2 is a schematic diagram of training an infectious disease transmission model;
FIG. 3 is an interface diagram of an infectious disease transmission model system incorporating expert knowledge.
Detailed Description
The present invention is further illustrated by the following examples, which are intended to be purely exemplary and are not intended to limit the scope of the invention, as various equivalent modifications of the invention will occur to those skilled in the art upon reading the present disclosure and fall within the scope of the appended claims.
The infectious disease propagation model building method with introduced expert knowledge can fuse the expert knowledge with infectious disease historical data, make up for deficiencies, and quickly train an efficient infectious disease propagation model learned by a (semi-) white-box machine through the expert knowledge and the data. The current machine learning model is difficult to have the ability of utilizing expert knowledge and data driving at the same time. As shown in fig. 3, the interface diagram of the infectious disease transmission model system with the introduction of expert knowledge can introduce the expert knowledge through the initialization interface of the system platform, and a model framework with the introduction of expert knowledge is quickly built through the system. The system comprises a data field and a machine learning field. The data field comprises a parameter module used for setting parameters to be determined, a variable module used for setting variables in the calculation process, and a recording module used for storing data. The loss function module is placed in the machine learning field and used for judging the quality of the infectious disease transmission model result, and the machine learning algorithm module is used for selecting an infectious disease transmission model training method.
As shown in FIG. 1, the method mainly comprises two parts of constructing an infectious disease transmission model by using expert knowledge and optimizing the infectious disease transmission model by using data. The method comprises the following steps:
step 101, defining functions related to simulating infectious disease transmission by using expert knowledge, and then building an infectious disease transmission model framework by using the functions. Functions related to the spread of infectious diseases include an initialization function, an infection function, a diagnosis function, a latent death function, and a disease course development function.
Step 102, listing a parameter set to be determined in an infectious disease transmission model by using expert knowledge; the set of parameters to be determined is a subset of the set of parameters of the model for the spread of infectious disease, which parameters should be set as the parameters to be determined being decided by expert knowledge.
Step 103, initializing infectious disease transmission model parameters by using expert knowledge. When the infectious disease propagation model is initialized, expert knowledge is used for setting a reasonable initial value for each parameter, the value is usually not an optimal value but has a certain reference value, the model can learn by taking the value as a starting point to accelerate the convergence speed of the model, and the initialization value can also be randomly set by the system in the process.
Step 104, inputting training data to the model for infectious disease transmission. The training data contains features (or sequences of actions) and result values. The training data here refers to: and in a period of time, a data set formed by newly adding the number of confirmed persons, the number of cured persons and the number of dead persons every day.
And 105, deducing the training data in the constructed infectious disease transmission model, and learning the infectious disease transmission model from the training data. In the first round of optimization, the model of infectious disease transmission is optimized on the basis of parameters with initial values. In subsequent optimization, the model of infectious disease transmission is optimized based on the parameters of the previous round.
Step 106, the system determines whether a training termination condition is reached. If the termination condition is not met, returning to the step 105; if the termination condition is reached, proceed to step 107.
And step 107, finishing the training, and outputting the optimal infectious disease transmission model by the system.
Fig. 2 is a schematic diagram of a model for training infectious disease transmission according to the present invention.
According to general experience, we make an assumption about the model of the transmission of infectious diseases (note: this assumption lacks the support of experts in the infectious disease field and is not authoritative, and here, this is taken as an example only and should not be applied at all): from a certain starting point, the 0 th infected person appears; the infected person has a latent period, wherein the maximum latent period is 20 days (without authority, and only the example is taken here), and the latent period has infectivity; latent infected patients have a certain probability of onset (according to media reports, a considerable number of infected patients without any symptoms are present), and patients with onset have a certain probability of being diagnosed in hospitals (some patients are reported not to go to the hospitals); the patient in the hospital is well isolated and no longer has the ability of infection. We add an infection function, a diagnosis function, a latent death function, and a disease progression function in the system platform. Since it is assumed that the diagnosed patient is isolated and the number of deaths therein has no effect on the disease transmission, it is omitted. In addition, the real accumulated diagnosis data of a certain area is added in the data area. The implementation process of each module for building the infectious disease model is described as follows:
step 201, designing and building a latent period development model frame according to expert knowledge, wherein the latent period development model frame comprises an initialization function, an infection function, a diagnosis function, a latent period death function and a disease course development function.
Step 202, listing parameters to be determined in a parameter module according to expert knowledge, wherein the parameters comprise the number of latent infection people before 20 days of the first confirmed patient, the number of people all contacting each day, the average infection rate, the confirmed diagnosis probability after 5 days of infection and the average death rate, and setting reasonable value ranges for the parameters. In this example, the number of latent infections before 20 days in the first confirmed patient is in the range of 1 to 10, the number of persons who are all in daily contact is in the range of 1 to 50, the average infection rate is in the range of 0 to 1, the probability of confirmed diagnosis 5 days after infection is in the range of 0 to 1, and the average daily mortality rate is in the range of 0 to 1.
In the initialization function, we mainly set the initial infected person number, step 203. In this example, based on general human experience, assuming a maximum latency of 20 days, the latency array should be filled with the number of latently infected persons for the first 20 days when the first confirmed patient is observed. However, the number of latent infections is unknown, so we set a parameter to be determined to indicate the number of latent infections 20 days before the first diagnosed patient.
Step 204, the system executes an initialization function once, sets an initial value for each parameter according to expert knowledge, and then iteratively executes the model for multiple times.
Step 205, the disease progression function moves the number of people in the incubation period of each day one day backward.
In step 206, the infection function calculates the number of infected persons for a day. The parameters to be searched are set to be 'the number of people per day who contact the human body' and 'the contact average infection rate'.
And step 207, calculating the number of confirmed persons by the confirmed function. According to general human experience, a certain probability is assumed to be confirmed after 5 days of infection, the probability is set as a parameter to be searched, namely the probability of confirmed diagnosis after 5 days of infection, and if the probability of confirmed diagnosis is removed from a latent queue.
And step 208, calculating the death number by using the latent death function, and setting a parameter to be searched, namely the daily average death rate, for the function.
Step 209, inputting the historical data into the constructed model containing expert knowledge, and outputting a result by the model, wherein the result is the number of confirmed persons in the example.
And step 210, calculating the error between the output result of the model and the real result of the data by using the mean square error loss function. In this example, the error between the number of confirmed persons obtained by model inference and the actual number of confirmed persons is obtained.
In step 211, the optimization algorithm determines the parameter value to be optimized for the next iteration according to the size of the loss function.
Step 212, repeatedly executing step 205-step 211 until after a training stopping command is triggered (for example, the maximum training times are reached, a satisfactory optimization result is obtained, and the parameter searching space is empty), the model stops training, and returns the optimal model parameters searched at this time. Because the model incorporates expert knowledge, it can be considered as a (semi-) white-box model, which is more interpretable than a traditional machine-learned black-box model, easily understood by humans. Meanwhile, the input data finely adjusts the model built by the experts, so that the practical situation of infectious disease transmission can be reflected more truly.
Specifically, sampling is performed in a search space of parameters to be searched, each group of parameters is deduced once in the constructed model, and the loss function is used for evaluating errors between the output result of the model and the real result. And the optimization algorithm continuously reduces the parameter search space through the error value until the parameter space is an empty set, and the optimal infectious disease propagation model parameter is obtained after the search is finished.
In summary, the present invention can be applied to machine learning model training involving experience with human expert knowledge and historical data, including but not limited to virus propagation models, population mobility models, economic models, etc. For example, in predicting infection transmission, an expert-summarized empirical model such as SIR model, SEIR model, etc. may be combined with currently acquired infection transmission data to train an easily understandable predictive model.
Claims (5)
1. An infectious disease transmission model construction method introducing expert knowledge is characterized by comprising two parts of constructing an infectious disease transmission model by utilizing the expert knowledge and optimizing the infectious disease transmission model by utilizing data; developers can introduce expert knowledge through an initialization interface of the system platform; the system comprises a data field and a machine learning field; the data field comprises a parameter module for setting parameters to be determined, a variable module for setting variables in the calculation process, and a recording module for storing data; the machine learning field placement loss function module is used for judging the quality of an infectious disease transmission model result, and the machine learning algorithm module is used for selecting an infectious disease transmission model training method; the training procedure is described as follows:
101, defining a function related to the simulation of infectious disease transmission by using expert knowledge, and building an infectious disease transmission model framework;
step 102, listing a parameter set to be determined in an infectious disease transmission model by using expert knowledge;
103, initializing infectious disease transmission model parameters by using expert knowledge;
step 104, inputting training data to an infectious disease propagation model;
105, deducing training data in the constructed infectious disease transmission model;
step 106, judging whether a training termination condition is reached by the system; if the termination condition is not met, returning to the step 105; if the termination condition is reached, go to step 107;
and step 107, finishing the training, and outputting the optimal infectious disease transmission model by the system.
2. An infectious disease propagation model building method with expert knowledge introduced according to claim 1, wherein the functions related to infectious disease propagation include an initialization function, an infection function, a diagnosis function, a latent death function, and a disease progression function.
3. The method of claim 1, wherein the loss function of the model is used to estimate the error between the output result and the actual result, and different loss functions are selected according to specific tasks and data, and the loss functions include mean square error loss function and cross entropy loss function.
4. The method of claim 1, wherein the model is optimized using a non-gradient optimization algorithm or an evolutionary computation method and a gradient-based solution method; when a gradient-free optimization algorithm or an evolution calculation method is used, the system samples in a search space of parameters to be searched, each group of parameters is deduced once in the constructed infectious disease transmission model, and errors between the output result and the real result of the infectious disease transmission model are evaluated by using a loss function; the optimization algorithm continuously reduces the parameter searching space through the error value until the parameter space is an empty set, and an optimal infectious disease propagation model is obtained after the searching is finished; when the gradient-based solving method is used, training data are repeatedly deduced in the model, the error between the output result and the real result of the infectious disease transmission model is evaluated by using a loss function, and the parameter to be searched is iteratively optimized by using gradient information until the infectious disease transmission model is converged to obtain the optimal infectious disease transmission model.
5. The method of claim 1, wherein the set of parameters to be optimized in the model is combined asθUsing the set of parameters to be optimizedθThe model is constructed ash θ (ii) a The historical data of infectious disease transmission is characterized in thatx i()Corresponding to a result ofy i()WhereiniTo representmThe fourth of the individual data samples; recording the loss function asLThen the error between the output value predicted by the model and the true value of the data is,
(ii) a If a loss of mean square error function is used, the error is
(ii) a From this error, minimizing the error using a machine learning optimization algorithm can be expressed as
(ii) a Finally obtaining the optimal parameter setθ * Optimized machine learning modelh θ* 。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110299554.6A CN112669989A (en) | 2021-03-22 | 2021-03-22 | Infectious disease transmission model construction method introducing expert knowledge |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110299554.6A CN112669989A (en) | 2021-03-22 | 2021-03-22 | Infectious disease transmission model construction method introducing expert knowledge |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112669989A true CN112669989A (en) | 2021-04-16 |
Family
ID=75399533
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110299554.6A Pending CN112669989A (en) | 2021-03-22 | 2021-03-22 | Infectious disease transmission model construction method introducing expert knowledge |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112669989A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113611429A (en) * | 2021-05-12 | 2021-11-05 | 中国人民解放军军事科学院军事医学研究院 | Infectious disease propagation deduction method and device and electronic equipment |
| CN113611429B (en) * | 2021-05-12 | 2024-06-07 | 中国人民解放军军事科学院军事医学研究院 | Infectious disease transmission deduction method and device and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111540478A (en) * | 2020-04-22 | 2020-08-14 | 第四范式（北京）技术有限公司 | Epidemic situation deduction simulation system and simulation method |
| CN112331357A (en) * | 2020-09-18 | 2021-02-05 | 广州优飞信息科技有限公司 | Parameter fitting method of infectious disease transmission model and related device |
-
2021
- 2021-03-22 CN CN202110299554.6A patent/CN112669989A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111540478A (en) * | 2020-04-22 | 2020-08-14 | 第四范式（北京）技术有限公司 | Epidemic situation deduction simulation system and simulation method |
| CN112331357A (en) * | 2020-09-18 | 2021-02-05 | 广州优飞信息科技有限公司 | Parameter fitting method of infectious disease transmission model and related device |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113611429A (en) * | 2021-05-12 | 2021-11-05 | 中国人民解放军军事科学院军事医学研究院 | Infectious disease propagation deduction method and device and electronic equipment |
| CN113611429B (en) * | 2021-05-12 | 2024-06-07 | 中国人民解放军军事科学院军事医学研究院 | Infectious disease transmission deduction method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Naeem et al. | A gentle introduction to reinforcement learning and its application in different fields | |
| CN110147450B (en) | Knowledge complementing method and device for knowledge graph | |
| CN114860893B (en) | Intelligent decision-making method and device based on multi-mode data fusion and reinforcement learning | |
| CN102682210B (en) | Self-adaptive frog cluster evolutionary tree designing method used for electronic medical record attribute reduction | |
| CN110014428B (en) | Sequential logic task planning method based on reinforcement learning | |
| Puranic et al. | Learning from demonstrations using signal temporal logic in stochastic and continuous domains | |
| CN116051574A (en) | Semi-supervised segmentation model construction and image analysis method, device and system | |
| CN111178486A (en) | Hyper-parameter asynchronous parallel search method based on population evolution | |
| Rosenbloom et al. | Towards emotion in sigma: from appraisal to attention | |
| CN113894780B (en) | Multi-robot cooperation countermeasure method, device, electronic equipment and storage medium | |
| Jiang et al. | CU. POKer: placing DNNs on wafer-scale AI accelerator with optimal kernel sizing | |
| CN112669989A (en) | Infectious disease transmission model construction method introducing expert knowledge | |
| CN110941962B (en) | Answer sentence selection method and device based on graph network | |
| CN112669186A (en) | Optimal intervention method and system for positive emotion and negative emotion in crowd evacuation | |
| CN116208399A (en) | Network malicious behavior detection method and device based on metagraph | |
| CN114818682B (en) | Document level entity relation extraction method based on self-adaptive entity path perception | |
| Namasivayam et al. | Learning neuro-symbolic programs for language guided robot manipulation | |
| US20220395975A1 (en) | Demonstration-conditioned reinforcement learning for few-shot imitation | |
| KR102559605B1 (en) | Method and apparatus for function optimization | |
| Fortier et al. | Parameter estimation in Bayesian networks using overlapping swarm intelligence | |
| WO2022127603A1 (en) | Model processing method and related device | |
| CN107480768A (en) | Bayesian network structure adaptive learning method and device, storage device and terminal device | |
| Iqbal | Improving the scalability of XCS-based learning classifier systems | |
| James | Learning portable symbolic representations | |
| Lu | Learning causal representations for generalization and adaptation in supervised, imitation, and reinforcement learning |
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 |
Application publication date: 20210416 |
|
| RJ01 | Rejection of invention patent application after publication |