CN115844696A - Method and device for generating visual training scheme, terminal equipment and medium - Google Patents
Method and device for generating visual training scheme, terminal equipment and medium Download PDFInfo
- Publication number
- CN115844696A CN115844696A CN202310158987.9A CN202310158987A CN115844696A CN 115844696 A CN115844696 A CN 115844696A CN 202310158987 A CN202310158987 A CN 202310158987A CN 115844696 A CN115844696 A CN 115844696A
- Authority
- CN
- China
- Prior art keywords
- training
- data
- user
- scheme
- medical record
- 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
- Eye Examination Apparatus (AREA)
Abstract
The invention provides a method, a device, terminal equipment and a medium for generating a visual training scheme, wherein the method comprises the following steps: acquiring eye medical record data of a user to be trained; inputting the training data into a training scheme generation model trained in advance so that the model outputs a data sequence of a training project and further generates a corresponding training scheme; the training scheme comprises training time, training items and training categories; the model is obtained by training a sequence prediction model through deep learning based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained through training effect screening. Compared with the prior art that the evaluation is carried out manually, the technical scheme of the specific training project is further determined, subjectivity is reduced, and meanwhile time for evaluation is greatly shortened.
Description
Technical Field
The present invention relates to the field of visual training, and in particular, to a method and an apparatus for generating a visual training scheme, a terminal device, and a computer-readable storage medium.
Background
The visual training is an important means for improving visual function and visual performance, and mainly aims at application scenes including abnormal visual information processing, abnormal visual and motor coordination, rehabilitation of vision after brain trauma or shock and the like. This approach may allow the patient to learn a more effective method so that it performs better visually. In addition, the visual training can effectively improve the visual information processing capability and the capability of watching moving objects for a long time, and is very beneficial to school-age children and reading after adults.
Generally, before performing vision training on a user, an expert or a doctor needs to perform multiple visual function tests on the user, such as binocular vision test, binocular refraction test, gaze property test, contrast sensitivity test, simultaneous vision function test, fusion function test, stereoscopic vision function test, and the like, then a vision training scheme is given through subjective evaluation of the expert according to the results of the tests, and then a training module and a training item in a product are called to realize corresponding vision training. However, the method relies on the evaluation of experts or doctors when determining a specific training scheme, the method has subjectivity and also has the problem of low evaluation and training efficiency, and the accuracy of the evaluation result depends on the experience of the experts or doctors and has a certain knowledge threshold, so that the pertinence of the training scheme determined according to the method is poor, and the training effect is not obvious.
Disclosure of Invention
The invention provides a method and a device for generating a visual training scheme, terminal equipment and a computer readable storage medium, which are used for solving the technical problem of improving the pertinence of the training scheme and improving the matching degree with a user.
In order to solve the above technical problem, an embodiment of the present invention provides a method for generating a visual training scheme, including:
acquiring eye medical record data of a user to be trained;
inputting the eye medical record data of the user to be trained into a training scheme generation model which is trained in advance, so that the training scheme generation model outputs a data sequence of a training item, and further generating a corresponding training scheme through the data sequence; wherein the training scheme comprises training time, training items and training categories for the user to be trained; the training time and the training items are respectively in one-to-one correspondence;
the training scheme generation model is obtained by training by adopting a deep learning sequence prediction model based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained by screening training effects; the training effect is determined according to the change value of the vision level.
Preferably, the medical record data of the eyes comprise sex, age, eyesight, mydriasis, optometry, fixation property, simultaneous vision data, fusion function, stereoscopic vision data, amblyopia type, eye position, nystagmus condition and diagnosis result;
the training process of the training scheme generation model comprises the following steps:
acquiring the user sample data;
inputting the eye medical record data of the users and the corresponding training schemes into a preset transform basic model, performing iterative training on the transform basic model, and taking the trained transform basic model as the training scheme generation model; the training process of the training scheme generation model adopts cross entropy loss as a loss function.
Preferably, the transform base model comprises an embedding process, a standardization process and a full connection layer;
the embedding process is used for converting discrete features in the eye medical records of the users into continuous feature vectors; the standardization process standardizes continuous features in the eye medical record data of the users;
the full connection layer is used for splicing the feature vectors obtained after conversion and the continuous features subjected to standardization processing, and mapping the splicing result; inputting the mapping result into a decoder of the transform.
As a preferred scheme, the user sample data is obtained by screening training effects, and specifically comprises the following steps:
determining the training effect of each user sample corresponding to the training scheme according to the change value of the measuring result of the logMAR or the score recording method;
and screening out user samples with training effects within a preset range, and obtaining the corresponding screened user sample data.
Correspondingly, the embodiment of the invention also provides a device for generating the visual training scheme, wherein the device comprises a data acquisition module and a generation module; wherein the content of the first and second substances,
the data acquisition module is used for acquiring eye medical record data of a user to be trained;
the generating module is used for inputting the eye medical record data of the user to be trained into a training scheme generating model which is trained in advance, so that the training scheme generating model outputs a data sequence of a training project, and a corresponding training scheme is generated through the data sequence; wherein the training scheme comprises a training time, a training item and a training category for the user to be trained; the training time and the training items are respectively in one-to-one correspondence;
the training scheme generation model is obtained by training by adopting a deep learning sequence prediction model based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained by screening training effects; the training effect is determined according to the change value of the vision level.
Preferably, the medical record data of the eyes comprise sex, age, eyesight, mydriasis, optometry, fixation property, simultaneous vision data, fusion function, stereoscopic vision data, amblyopia type, eye position, nystagmus condition and diagnosis result;
the training process of the training scheme generation model comprises the following steps:
acquiring the user sample data;
inputting the eye medical record data of the users and the corresponding training schemes into a preset transform basic model, performing iterative training on the transform basic model, and taking the trained transform basic model as the training scheme generation model; the training process of the training scheme generation model adopts cross entropy loss as a loss function.
As a preferred scheme, the transform base model comprises an embedding process, a standardization process and a full connection layer;
the embedding process is used for converting discrete features in the eye medical records of the users into continuous feature vectors; the standardization process standardizes continuous features in the eye medical record data of the users;
the full connection layer is used for splicing the feature vectors obtained after conversion and the continuous features subjected to standardization processing and mapping the splicing result; inputting the mapping result into a decoder of the transform.
As a preferred scheme, the user sample data is obtained by screening training effects, and specifically comprises the following steps:
determining the training effect of each user sample corresponding to the training scheme according to the change value of the measuring result of the logMAR or the score recording method;
and screening out user samples with training effects within a preset range, and obtaining the corresponding screened user sample data.
Correspondingly, the embodiment of the present invention further provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the method for generating the visual training scheme when executing the computer program.
Correspondingly, the embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program, and when the computer program runs, the apparatus on which the computer-readable storage medium is located is controlled to execute the method for generating the visual training scheme.
Compared with the prior art, the embodiment of the invention has the following beneficial effects:
the embodiment of the invention provides a method and a device for generating a visual training scheme, terminal equipment and a computer readable storage medium, wherein the method comprises the following steps: acquiring eye medical record data of a user to be trained; inputting the eye medical record data of the user to be trained into a training scheme generation model which is trained in advance, so that the training scheme generation model outputs a data sequence of a training item, and further generating a corresponding training scheme through the data sequence; wherein the training scheme comprises a training time, a training item and a training category for the user to be trained; the training time and the training items are respectively in one-to-one correspondence; the training scheme generation model is obtained by training by adopting a deep learning sequence prediction model based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained by screening training effects; the training effect is determined according to the change value of the vision level. Compared with the prior art that the training scheme corresponding to the user to be trained is manually evaluated and the specific training scheme is determined, the method and the device have the advantages that subjectivity caused by human is reduced, meanwhile, time for evaluation can be greatly shortened, dependency on expert experience is reduced, and generation efficiency of the training scheme is improved; moreover, the user sample data for model training comprises the eye medical record data of a plurality of users and corresponding training schemes, so that the matching degree of the eye medical record data of the user to be trained and the corresponding training schemes can be effectively improved, and the training effect is effectively improved aiming at the specific eye condition of the user.
Furthermore, discrete features are converted into continuous feature vectors, the continuous feature vectors are spliced with the standardized continuous features, and mapping is performed through a full connection layer, so that eye medical record data of a user and a corresponding training scheme can be effectively matched, the matching degree between the user and the training scheme is further improved, and the method is more targeted.
Furthermore, the degree of change of the vision level of the user is measured through logMAR or a score recording method to determine the training effect of the training scheme, invalid user samples can be removed by screening the user samples according to the training effect, the overall effectiveness of the user samples is improved, and therefore the performance of the training scheme generation model is improved.
Drawings
FIG. 1: the invention provides a flow chart of an embodiment of a method for generating a visual training scheme.
FIG. 2: is a schematic diagram of an embodiment of a user medical record provided by the invention.
FIG. 3: the invention provides a schematic principle diagram of an embodiment of a transformer basic model.
FIG. 4: a schematic diagram of an embodiment of a user training scenario provided by the present invention.
FIG. 5: the invention provides a schematic structural diagram of an embodiment of a visual training scheme generation device.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The first embodiment is as follows:
according to the related art description: before the vision training of the user, an expert or a doctor needs to perform a plurality of visual function tests on the user, such as binocular vision test, binocular refraction test, fixation property test, contrast sensitivity test, simultaneous vision function test, fusion function test, stereoscopic vision function test and the like, and then a vision training scheme is given through subjective evaluation of the expert according to the results of the tests. However, it should be noted that the visual function of the user is not only a visual function defect, but also a visual function range in which the user feels comfortable can be corrected by means of visual training, although the visual function range is understood as a visual function defect, for example, the visual function of the user is 4.8, which does not represent that the user has a disease in his eyes, and the visual function of the user is low. However, in the prior art, when a specific training scheme is determined, the assessment of experts or doctors is relied on, but different experiences of different experts are considered, so that the method has certain subjectivity, and then the assessment and the determination of the training scheme are performed in a manual mode.
In view of one or more of the above technical problems, referring to fig. 1, fig. 1 is a method for generating a vision training scheme according to an embodiment of the present invention, which includes steps S1 and S2, wherein,
step S1, eye medical record data of a user to be trained are obtained.
In this embodiment, the eye medical record data (eye medical record may refer to fig. 2) of the user to be trained may be obtained from an input terminal (for example, an input terminal for performing visual function examination, or an expert or doctor manually inputs into the terminal), including the gender, age, eyesight (including naked eye eyesight, corrected eyesight), mydriasis (such as mydriasis agent), optometry, gaze property, simultaneous vision data (including dominant eye data, perceptual eye position, and the like, and the perceptual eye position includes horizontal deviation and vertical deviation, and the like), fusion function, stereoscopic vision data, amblyopia type, eye position, nystagmus condition, and diagnosis result of the user. By calling and acquiring eye medical record data of a user to be trained, the specific situation of the eyes of the user can be determined in detail, and a more accurate and targeted training scheme is formulated, so that the training effect is effectively improved.
S2, inputting eye medical record data of the user to be trained into a training scheme generation model which is trained in advance, so that the training scheme generation model outputs a data sequence of a training project, and further generating a corresponding training scheme through the data sequence; wherein the training scheme comprises a training time, a training item and a training category for the user to be trained; the training time and the training items are respectively in one-to-one correspondence.
The training scheme generation model is obtained by training by adopting a deep learning sequence prediction model based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained by screening training effects; the training effect is determined according to the change value of the vision level. Also, the training regimen may be a list of training items, such as the list being training items scheduled daily, the time of training may be scheduled in days, or in hours, or even in minutes.
In this embodiment, the training process of the training scheme generation model includes:
acquiring the user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes. These data may be data recorded by an expert evaluation process, including user samples with various different types of ocular functional defects. Also, since visual training usually requires long-term adherence, each user's training regimen usually has not less than one day of training content, and the training content includes not less than 120 different types of training items, including but not limited to "skiing", "shooting balloon", "mushroom picking", "finding discrepancies", "tetris", "schotter table", and "dragon ball", etc.
Further, screening the user sample data according to the training effect:
before training, the patient can carry out visual function examination to obtain eye medical record data, then confirm specific training scheme according to the medical record data, can carry out visual function examination again after the user training is accomplished, confirm the condition of eyesight, therefore the effect of this embodiment training is mainly confirmed according to the variation of eyesight level.
Specifically, the training effect of each user sample corresponding to the training scheme can be determined according to the change value of the measurement result of the logMAR or the score recording method; when logMAR is used, a smaller value represents better vision level, and when score recording is used, a larger value represents better vision level.
And screening user samples with the training effect within a preset range (for example, sorting the user samples according to the variation degree of the vision level from large to small, and screening data with the training effect of being in the top 20%), and obtaining the corresponding screened user sample data.
Inputting the eye medical record data of the users and the corresponding training schemes into a preset transform basic model, performing iterative training on the transform basic model, and taking the trained transform basic model as the training scheme generation model; the training process of the training scheme generation model adopts cross entropy loss as a loss function.
After the filtered user sample data is obtained, a data set D = (X, Y) may be obtained, where X is case data before patient training, and Y is a training scheme corresponding to the case data, specifically, a sequence composed of training items.
Considering that the input data is not sequence data, in the transform basic model (refer to fig. 3), it specifically includes an embedding process, a normalization process, and a full connection layer; the embedding process is used for converting discrete features in the eye medical record data of the users into continuous feature vectors; the normalization process normalizes (e.g., normalizes) consecutive features in the ocular medical record data of the users; it should be noted that information such as age, sex, mydriasis, gaze quality, eye position, nystagmus, diagnosis result, etc. are discrete features, while visual acuity, optometry, contrast sensitivity, simultaneous vision data, stereoscopic vision data, fusion function, etc. are continuous features.
After converting the discrete features into continuous feature vectors and standardizing the continuous features, the full-connection layer is used for splicing the feature vectors obtained after conversion and the continuous features subjected to the standardized processing and mapping splicing results; inputting a result of an encoder layer, the mapping result of which is a transform, into a decoder of the transform. The decoder structure and the loss function of the transform remain unchanged, and the parameter setting in the training process also remains unchanged.
Finally, the output of the model is mainly a training item sequence, and after the training items are indexed, the training scheme is changed into an index sequence. In this embodiment, the cross entropy loss is used as a loss function in the training process of the training scheme generation model, and when a preset convergence condition is met, for example, the loss function is in a certain range, or the loss function variation is smaller than a preset threshold, the model is judged to be converged, so as to obtain the trained training scheme generation model. And user sample data can be updated in the training process of the model, for example, when new user sample data is obtained in channels such as an input terminal and the like, the user sample can be used for training the model after being screened, so that the model is continuously optimized, and the performance of the model is improved.
After the trained training scheme generation model is obtained, the eye medical record data of the user to be trained is input into the model, so that the training scheme corresponding to the user to be trained can be obtained, and fig. 4 is referred to. The specific training scheme comprises a training category, a training item, training time and the like, so that the user can be subjected to targeted visual training.
Correspondingly, referring to fig. 5, an embodiment of the present invention further provides a generation apparatus for a visual training scheme, where the generation apparatus includes a data acquisition module 101 and a generation module 102; wherein the content of the first and second substances,
the data acquisition module 101 is configured to acquire eye medical record data of a user to be trained;
the generating module 102 is configured to input the eye medical record data of the user to be trained into a training scheme generating model trained in advance, so that the training scheme generating model outputs a data sequence of a training project, and a corresponding training scheme is generated through the data sequence; wherein the training scheme comprises a training time, a training item and a training category for the user to be trained; the training time and the training items are respectively in one-to-one correspondence;
the training scheme generation model is obtained by training by adopting a deep learning sequence prediction model based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained by screening training effects; the training effect is determined according to the change value of the vision level.
As a preferred scheme, the eye medical record data comprises sex, age, eyesight, mydriasis, optometry, fixation properties, simultaneous vision data, fusion function, stereoscopic vision data, amblyopia type, eye position, nystagmus condition and diagnosis results;
the training process of the training scheme generation model comprises the following steps:
acquiring the user sample data;
inputting the eye medical record data of the users and the corresponding training schemes into a preset transform basic model, performing iterative training on the transform basic model, and taking the trained transform basic model as the training scheme generation model; the training process of the training scheme generation model adopts cross entropy loss as a loss function.
Preferably, the transform base model comprises an embedding process, a standardization process and a full connection layer;
the embedding process is used for converting discrete features in the eye medical records of the users into continuous feature vectors; the standardization process standardizes continuous features in the eye medical record data of the users;
the full connection layer is used for splicing the feature vectors obtained after conversion and the continuous features subjected to standardization processing and mapping the splicing result; inputting the mapping result into a decoder of the transform.
As a preferred scheme, the user sample data is obtained by screening training effects, and specifically comprises the following steps:
determining the training effect of each user sample corresponding to the training scheme according to the change value of the measuring result of the logMAR or the score recording method;
and screening out user samples with training effects within a preset range, and obtaining the corresponding screened user sample data.
Correspondingly, the embodiment of the present invention further provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the method for generating the visual training scheme when executing the computer program.
The Processor may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable gate array (FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware component, etc. The general-purpose processor may be a microprocessor or the processor may be any conventional processor or the like, which is the control center of the terminal and connects the various parts of the overall terminal using various interfaces and lines.
The memory may be used to store the computer program, and the processor may implement various functions of the terminal by executing or executing the computer program stored in the memory and calling data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data, a phonebook, etc.) created according to the use of the cellular phone, and the like. In addition, the memory may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other volatile solid state storage device.
Correspondingly, the embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program, and when the computer program runs, a device on which the computer-readable storage medium is located is controlled to execute the method for generating the visual training scheme.
Wherein, the module integrated by the visual training scheme generating device can be stored in a computer readable storage medium if the module is realized in the form of a software functional unit and sold or used as a stand-alone product. Based on such understanding, all or part of the flow of the method according to the embodiments of the present invention may also be implemented by a computer program, which may be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the method embodiments may be implemented. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U.S. disk, removable hard disk, magnetic diskette, optical disk, computer Memory, read-Only Memory (ROM), random Access Memory (RAM), electrical carrier wave signal, telecommunications signal, and software distribution medium, etc.
Compared with the prior art, the embodiment of the invention has the following beneficial effects:
the embodiment of the invention provides a method and a device for generating a visual training scheme, a terminal device and a computer readable storage medium, wherein the method comprises the following steps: acquiring eye medical record data of a user to be trained; inputting the eye medical record data of the user to be trained into a pre-trained training scheme generation model, so that the training scheme generation model outputs a data sequence of a training project, and further generating a corresponding training scheme through the data sequence; wherein the training scheme comprises a training time, a training item and a training category for the user to be trained; the training time and the training items are respectively in one-to-one correspondence; the training scheme generation model is obtained by training by adopting a deep learning sequence prediction model based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained by screening training effects; the training effect is determined according to the change value of the vision level. Compared with the prior art that the training scheme corresponding to the user to be trained is manually evaluated and the specific training scheme is determined, the method and the device have the advantages that subjectivity caused by human is reduced, meanwhile, time for evaluation can be greatly shortened, dependency on expert experience is reduced, and generation efficiency of the training scheme is improved; moreover, the user sample data for model training comprises the eye medical record data of a plurality of users and the corresponding training scheme, so that the matching degree of the eye medical record data of the user to be trained and the corresponding training scheme can be effectively improved, and the training effect is effectively improved aiming at the specific eye condition of the user.
Furthermore, discrete features are converted into continuous feature vectors, the continuous feature vectors are spliced with the standardized continuous features, and mapping is performed through a full connection layer, so that eye medical record data of a user and a corresponding training scheme can be effectively matched, the matching degree between the user and the training scheme is further improved, and the method is more targeted.
Furthermore, the degree of change of the vision level of the user is measured through logMAR or a score recording method to determine the training effect of the training scheme, invalid user samples can be removed by screening the user samples according to the training effect, the overall effectiveness of the user samples is improved, and therefore the performance of the training scheme generation model is improved.
The above-mentioned embodiments are provided to further explain the objects, technical solutions and advantages of the present invention in detail, and it should be understood that the above-mentioned embodiments are only examples of the present invention and are not intended to limit the scope of the present invention. It should be understood that any modifications, equivalents, improvements and the like, which come within the spirit and principle of the invention, may occur to those skilled in the art and are intended to be included within the scope of the invention.
Claims (10)
1. A method for generating a vision training solution, comprising:
acquiring eye medical record data of a user to be trained;
inputting the eye medical record data of the user to be trained into a training scheme generation model which is trained in advance, so that the training scheme generation model outputs a data sequence of a training item, and further generating a corresponding training scheme through the data sequence; wherein the training scheme comprises training time, training items and training categories for the user to be trained; the training time and the training items are respectively in one-to-one correspondence;
the training scheme generation model is obtained by training by adopting a deep learning sequence prediction model based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained by screening training effects; the training effect is determined according to the change value of the vision level.
2. The method of claim 1, wherein the eye medical record data comprises gender, age, vision, mydriasis, optometry, gaze quality, simultaneous vision data, fusion function, stereoscopic vision data, amblyopia type, eye position, nystagmus condition, and diagnosis;
the training process of the training scheme generation model comprises the following steps:
acquiring the user sample data;
inputting the eye medical record data of the users and the corresponding training schemes into a preset transform basic model, performing iterative training on the transform basic model, and taking the trained transform basic model as the training scheme generation model; the training process of the training scheme generation model adopts cross entropy loss as a loss function.
3. The method of claim 2, wherein the transform base model comprises an embedding process, a normalization process, and a full join layer;
the embedding process is used for converting discrete features in the eye medical record data of the users into continuous feature vectors; the standardization process standardizes continuous features in the eye medical record data of the users;
the full connection layer is used for splicing the feature vectors obtained after conversion and the continuous features subjected to standardization processing, and mapping the splicing result; inputting the mapping result into a decoder of the transform.
4. The method of claim 1, wherein the user sample data is obtained by screening training effects, and specifically comprises:
determining the training effect of each user sample corresponding to the training scheme according to the change value of the measuring result of the logMAR or the score recording method;
and screening out user samples with training effects within a preset range, and obtaining the corresponding screened user sample data.
5. The generation device of the vision training scheme is characterized by comprising a data acquisition module and a generation module; wherein the content of the first and second substances,
the data acquisition module is used for acquiring eye medical record data of a user to be trained;
the generating module is used for inputting the eye medical record data of the user to be trained into a training scheme generating model which is trained in advance, so that the training scheme generating model outputs a data sequence of a training project, and a corresponding training scheme is generated through the data sequence; wherein the training scheme comprises a training time, a training item and a training category for the user to be trained; the training time and the training items are respectively in one-to-one correspondence;
the training scheme generation model is obtained by training by adopting a deep learning sequence prediction model based on user sample data; the user sample data comprises eye medical record data of a plurality of users and corresponding training schemes, and is obtained by screening training effects; the training effect is determined according to the change value of the vision level.
6. The apparatus for generating a vision training solution as claimed in claim 5, wherein said eye medical record data includes sex, age, eyesight, mydriasis, optometry, gaze property, simultaneous vision data, fusion function, stereoscopic vision data, amblyopia type, eye position, nystagmus condition and diagnosis result;
the training process of the training scheme generation model comprises the following steps:
acquiring the user sample data;
inputting the eye medical record data of the users and the corresponding training schemes into a preset transform basic model, performing iterative training on the transform basic model, and taking the trained transform basic model as the training scheme generation model; the training process of the training scheme generation model adopts cross entropy loss as a loss function.
7. The apparatus for generating a visual training scheme as claimed in claim 6, wherein said transform base model comprises an embedding process, a normalization process and a full join layer;
the embedding process is used for converting discrete features in the eye medical record data of the users into continuous feature vectors; the standardization process standardizes continuous features in the eye medical record data of the users;
the full connection layer is used for splicing the feature vectors obtained after conversion and the continuous features subjected to standardization processing, and mapping the splicing result; inputting the mapping result into a decoder of the transform.
8. The apparatus for generating a visual training scheme according to claim 5, wherein the user sample data is obtained by training effect screening, and specifically comprises:
determining the training effect of each user sample corresponding to the training scheme according to the change value of the measuring result of the logMAR or the score recording method;
and screening out user samples with training effects within a preset range, and obtaining the corresponding screened user sample data.
9. A terminal device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the method of generating a vision training scheme as claimed in any one of claims 1 to 4 when executing the computer program.
10. A computer-readable storage medium, comprising a stored computer program, wherein the computer program, when executed, controls an apparatus in which the computer-readable storage medium is located to perform the method for generating a vision training scenario of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310158987.9A CN115844696A (en) | 2023-02-24 | 2023-02-24 | Method and device for generating visual training scheme, terminal equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310158987.9A CN115844696A (en) | 2023-02-24 | 2023-02-24 | Method and device for generating visual training scheme, terminal equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115844696A true CN115844696A (en) | 2023-03-28 |
Family
ID=85658810
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310158987.9A Pending CN115844696A (en) | 2023-02-24 | 2023-02-24 | Method and device for generating visual training scheme, terminal equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115844696A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116775996A (en) * | 2023-06-21 | 2023-09-19 | 广州视景医疗软件有限公司 | Visual training project recommending method and device based on user feedback |
| CN117593783A (en) * | 2023-11-20 | 2024-02-23 | 广州视景医疗软件有限公司 | Visual training scheme generation method and device based on self-adaptive SMOTE |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101673301A (en) * | 2009-09-11 | 2010-03-17 | 广州视景医疗软件有限公司 | Automatic generation method of visual training proposal and system thereof |
| CN113077866A (en) * | 2021-03-19 | 2021-07-06 | 国家康复辅具研究中心 | Automatic recommendation method and system for limb exercise training scheme |
| CN114969511A (en) * | 2022-05-05 | 2022-08-30 | 中国电子科技集团公司第十研究所 | Content recommendation method, device and medium based on fragments |
| CN115187512A (en) * | 2022-06-10 | 2022-10-14 | 珠海市人民医院 | Hepatocellular carcinoma great vessel invasion risk prediction method, system, device and medium |
| CN115423016A (en) * | 2022-08-31 | 2022-12-02 | 聚好看科技股份有限公司 | Training method of multi-task prediction model, multi-task prediction method and device |
-
2023
- 2023-02-24 CN CN202310158987.9A patent/CN115844696A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101673301A (en) * | 2009-09-11 | 2010-03-17 | 广州视景医疗软件有限公司 | Automatic generation method of visual training proposal and system thereof |
| CN113077866A (en) * | 2021-03-19 | 2021-07-06 | 国家康复辅具研究中心 | Automatic recommendation method and system for limb exercise training scheme |
| CN114969511A (en) * | 2022-05-05 | 2022-08-30 | 中国电子科技集团公司第十研究所 | Content recommendation method, device and medium based on fragments |
| CN115187512A (en) * | 2022-06-10 | 2022-10-14 | 珠海市人民医院 | Hepatocellular carcinoma great vessel invasion risk prediction method, system, device and medium |
| CN115423016A (en) * | 2022-08-31 | 2022-12-02 | 聚好看科技股份有限公司 | Training method of multi-task prediction model, multi-task prediction method and device |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116775996A (en) * | 2023-06-21 | 2023-09-19 | 广州视景医疗软件有限公司 | Visual training project recommending method and device based on user feedback |
| CN117593783A (en) * | 2023-11-20 | 2024-02-23 | 广州视景医疗软件有限公司 | Visual training scheme generation method and device based on self-adaptive SMOTE |
| CN117593783B (en) * | 2023-11-20 | 2024-04-05 | 广州视景医疗软件有限公司 | Visual training scheme generation method and device based on self-adaptive SMOTE |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115844696A (en) | Method and device for generating visual training scheme, terminal equipment and medium | |
| US20190258930A1 (en) | Apparatus for ascertaining predicted subjective refraction data or predicted correction values, and computer program | |
| US11890184B2 (en) | Systems, apparatuses, and methods for intraocular lens selection using artificial intelligence | |
| JP7097012B2 (en) | Kansei estimation device, Kansei estimation system, Kansei estimation method and program | |
| CN1762299A (en) | Method and apparatus for processing bio-information | |
| Pundlik et al. | Development and preliminary evaluation of a smartphone app for measuring eye alignment | |
| CN109766007B (en) | Method and device for compensating fixation point of display equipment and display equipment | |
| TW201013248A (en) | Method for designing eyeglass lens | |
| CN112102940B (en) | Refraction detection method, refraction detection device, computer equipment and storage medium | |
| CN112472089A (en) | System and method for judging reliability of psychological test based on eye movement technology | |
| CN114496243A (en) | Data processing method, data processing device, storage medium and electronic equipment | |
| CN114305418B (en) | Data acquisition system and method for intelligent assessment of depression state | |
| CN116563932A (en) | Eye image recognition method and related equipment based on multitask learning | |
| CN111568412A (en) | Method and device for reconstructing visual image by utilizing electroencephalogram signal | |
| CN113850804A (en) | Retina image generation system and method based on generation countermeasure network | |
| CN112381012A (en) | Method and device for identifying target region in eye image and electronic equipment | |
| CN111640097A (en) | Skin mirror image identification method and equipment | |
| CN116775996A (en) | Visual training project recommending method and device based on user feedback | |
| JP2023060172A (en) | Blood pressure measurement device, model setting device, and blood pressure measurement method | |
| JP7307507B2 (en) | Pathological condition analysis system, pathological condition analyzer, pathological condition analysis method, and pathological condition analysis program | |
| US20240148245A1 (en) | Method, device, and computer program product for determining a sensitivity of at least one eye of a test subject | |
| CN113469274B (en) | Method, device and equipment for evaluating reading capability | |
| CN115223232A (en) | Eye health comprehensive management system | |
| WO2021139446A1 (en) | Anti-vascular endothelial growth factor (vegf) curative effect prediction apparatus and method | |
| CN112674714A (en) | Mobile phone image examination optometry method combining filter and peripheral equipment |
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 |