A learning difficulties training system based on visual function testing

Abstract

The present application discloses a learning difficulty training system based on visual function testing, which relates to the field of artificial intelligence technology and includes: a personal file recording unit, a visual function testing unit, a level domain positioning unit, a personalized training unit, and a training data uploading unit; the visual function testing unit is used to obtain visual function indicators based on visual function testing, and associate the visual function indicators with personal visual function files to obtain visual function data; the level domain positioning unit is used to obtain the visual ability cluster to which the user belongs, and predict the user's learning difficulty risk level and problem area based on a learning difficulty risk assessment model; the personalized training unit is used to calculate the arrangement order of visual function data required for training, select training items from a training item question bank, and calculate the corresponding training time allocation value and training frequency for each training item; the training data uploading unit is used to generate an analysis report based on the user's historical training result data.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a learning difficulty training system based on visual function detection. Background Art

[0002] As the vision of children and adolescents is receiving more and more attention, routine vision examinations are not enough to reflect the overall visual function. Therefore, visual function examinations are needed. Visual function examinations are an important means of evaluating the overall visual function of the eyes. They include evaluations of color vision, contrast sensitivity, visual field, stereoscopic vision and other aspects. These examinations can comprehensively reflect the state of visual function. Scientific visual function training based on the results of visual function examinations can effectively solve symptoms such as visual fatigue, eye soreness, eye swelling, difficulty in near vision, drowsiness and other symptoms caused by abnormal binocular vision function, as well as learning difficulties caused by some symptoms.

[0003] To address efficiency, management, and user experience issues in visual function testing and training, the industry has developed several improvement methods. These include the introduction of automated testing equipment and software, the establishment of electronic file management systems, the development of visual training apps and games, and the use of virtual reality technology to simulate real-world scenarios. However, these methods still have significant drawbacks.

[0004] First, most existing automated inspection solutions have limited accuracy and applicability, still requiring manual operation and interpretation, and are unable to fully leverage the advantages of intelligence and achieve large-scale application. Secondly, the equipment and systems in each link are relatively independent, data transmission and processing efficiency is low, and there is a lack of information integration and optimization of the entire process. Furthermore, the content of visual function training is single and the format is not interesting enough, making it difficult to truly mobilize patients' initiative and enthusiasm, resulting in poor training compliance and efficacy. In addition, various types of inspection and training data are scattered, lacking system integration and big data analysis, making it impossible to achieve accurate evaluation and personalized program optimization.

[0005] Therefore, an innovative technical solution is urgently needed to build a comprehensive and intelligent visual function inspection and training management platform. Summary of the Invention

[0006] This application aims to solve at least one of the technical problems in the related art to a certain extent. To this end, one purpose of this application is to propose a learning difficulty training system based on visual function testing, which realizes an intelligent visual function testing and training management platform.

[0007] One aspect of the present application provides a learning difficulty training system based on visual function detection, comprising:

[0008] Personal profile recording unit, visual function testing unit, grade field positioning unit, personalized training unit, training data uploading unit;

[0009] The personal profile recording unit is used for user registration and login, to improve the user's personal information indicators and to establish a personal visual function profile;

[0010] The visual function detection unit is used to prompt the user to complete the visual function detection, obtain the user's visual function index according to the visual function detection, and associate the visual function index with the personal visual function file to obtain visual function data;

[0011] The level domain positioning unit is used to classify the user based on the user's visual function data and the trained visual ability clustering clusters, obtain the visual ability clustering cluster to which the current user belongs, and predict the learning difficulty risk level and problem domain of the user's learning difficulty based on a pre-established learning difficulty risk assessment model;

[0012] The personalized training unit is used to calculate the arrangement order of the visual function data required for training based on the visual ability cluster to which the user belongs, select training items from the training item question bank according to the learning difficulty risk level and problem domain of the user's learning difficulty and the arrangement order of the visual function data, calculate the corresponding training time allocation value and training frequency for each training item, and record the actual training time and training score for each training item;

[0013] The training data uploading unit is used to upload the training result data of the user after completing the training project to the data server, calculate the user's score improvement and time reduction on each training project based on the user's historical training result data, calculate the score difference and time difference on each training project between the user and other users based on the historical training result data of other users in the same visual ability cluster, and generate a training result analysis report;

[0014] The user registration and login, improvement of user personal information indicators, and establishment of personal visual function files specifically include:

[0015] Step S110: The user registers using their mobile phone number, and a unique identifier is generated for each user;

[0016] Step S120: The user fills in personal information indicators;

[0017] Step S130: storing the user's personal information indicators in a data server and establishing a personal visual function profile;

[0018] The step of prompting the user to complete a visual function test, obtaining the user's visual function index based on the visual function test, and associating the visual function index with the personal visual function profile to obtain visual function data specifically includes:

[0019] Step S201: Check the user's inhibition level and obtain the number of questions the user answered correctly. Percentage of total questions The percentage of the user's suppression index SI is obtained;

[0020] Step S202: Check the user's eye position and obtain the coordinates of each click position. and the actual center coordinates of the blue circle , calculate the user's perceptual eye position deviation PEP;

[0021] Step S203: Check the convergence range of the user and obtain the position of the rupture point marked in the divergence step and the rupture point locations marked in the collection step , get the maximum displacement of the vector diagram , calculate the user's gathering and dispersion range ;

[0022] Step S204: Perform a stereoscopic inspection on the user to determine the pattern selected by the user and the correct stereoscopic pattern Whether they are consistent, if they are consistent, the stereoscopic accuracy is 1, otherwise it is 0;

[0023] Step S205: Check the user's adjustment sensitivity and calculate the number of E-table directions correctly identified by the user. Total E-table rendering times The percentage of adjustment sensitivity is obtained ;

[0024] Step S206: Perform binocular visual balance check on the user and measure the left eye accommodation sensitivity of the user , right eye adjustment sensitivity , binocular accommodation sensitivity , get the user's binocular balance ;

[0025] Step S207: Perform a color vision test on the user, count the number of patterns correctly identified by the user, and calculate the number of patterns correctly identified by the user. The total number of patterns The ratio of the user's color vision integrity percentage is obtained ;

[0026] Step S208: Check the signal-to-noise ratio of the user, and obtain the average amplitude of the signal based on the different signal-to-noise ratios of the images at different transparency levels. and the average amplitude of the noise , calculate the signal-to-noise ratio of the image identified by the user ;

[0027] Step S209: Check the user's scanning ability and count the user's scanning time. Number of word logo directions that can be correctly identified , calculate the user's scanning ability score ;

[0028] Step S210: Check the user's ability to follow and count the user's time during the check. Number of directions of moving logos that can be correctly identified , calculate the user's followability score ;

[0029] Step S211: generating visual function indicators based on the test results obtained from the visual function test, wherein the visual function indicators include: inhibition index, perceived eye position deviation, vergence range, stereoscopic accuracy, accommodation sensitivity, binocular balance, color vision integrity percentage, signal-to-noise ratio, saccade ability score, and tracking ability score;

[0030] Step S212: Associating the visual function indicators with the personal visual function profile, collectively referred to as the user's visual function data;

[0031] The method for obtaining the visual ability cluster is:

[0032] Step S301: normalizing the visual function data of N users;

[0033] The normalized calculation formula is: ,in, represents the normalized visual function data of the jth user of the i-th user, represents the original value of the jth visual function data of the i-th user, represents the mean value of the j-th visual function data, represents the standard deviation of the j-th visual function data;

[0034] Step S302: using the normalized visual function data of N users as data points, clustering the data points using a K-means clustering algorithm to obtain K visual ability clusters;

[0035] The specific method for obtaining K visual ability clusters is:

[0036] Step S3021: randomly select K data points as initial cluster centers;

[0037] Step S3022: For each data point , calculate the distance between the data point and each cluster center, and assign it to the cluster with the nearest cluster center;

[0038] Step S3023: for each cluster, recalculate the mean vector of all data points in the cluster as the new cluster center of the cluster;

[0039] Step S3024: Repeat steps S3022 to S3023 until the cluster center of each cluster and the data points within the cluster no longer change, thereby obtaining K visual ability clusters;

[0040] The method for establishing the learning difficulty risk assessment model is:

[0041] Collect data on users diagnosed with learning difficulties, including personal information indicators and visual function indicators, as well as learning difficulty risk levels and problem areas assessed by professionals;

[0042] For each user sample with learning difficulties collected, the risk level of learning difficulties and whether there are obstacles in each problem area are marked based on the professional assessment results. The learning difficulty risk level includes low, medium, and high. The problem areas include but are not limited to reading, writing, and mathematics.

[0043] The visual function data of the learning difficulty user samples are constructed based on personal information indicators and visual function indicators. The learning difficulty user samples are clustered based on the visual ability clustering cluster. The visual ability clustering cluster to which each learning difficulty user sample belongs is determined. The learning difficulty risk score is calculated based on the visual function data and the mean and standard deviation of the visual ability clustering cluster to which it belongs. ;

[0044] A learning difficulty risk classification model is designed based on the learning difficulty risk score, and a first threshold and a second threshold of the learning difficulty risk score are set. When the learning difficulty risk score is less than or equal to the first threshold, the learning difficulty risk level is low; when the learning difficulty risk score is greater than the first threshold and less than or equal to the second threshold, the learning difficulty risk level is medium; and when the learning difficulty risk score is greater than the second threshold, the learning difficulty risk level is high;

[0045] Designing a multi-task learning model, the multi-task learning model comprising a shared feature extraction layer, a task-specific output layer, and a loss function, using samples of users with learning difficulties labeled with whether there are obstacles in each problem area as training samples, using visual function data of the samples of users with learning difficulties as input data, and using binary classification results of whether the samples of users with learning difficulties have obstacles in each problem area as output data, to train the multi-task learning model to obtain a trained multi-task learning model;

[0046] The learning difficulty risk classification model and the multi-task learning model are combined to obtain a learning difficulty risk assessment model;

[0047] The shared feature extraction layer receives visual function data, extracts multimodal input features from different visual function data and performs feature splicing, which serves as the input of the shared feature extraction layer. A deep neural network is used to perform hierarchical abstraction and transformation on the multimodal input features, extracts deep features and outputs them; the output of the deep features is mapped to a low-dimensional latent space to obtain common representations between classification tasks in different problem domains;

[0048] The low-dimensional latent space is realized by a fully connected layer, and its output dimension is smaller than the input dimension, thereby achieving the effect of dimensionality reduction and compression.

[0049] The task feature output layer designs an output sub-network for each problem domain classification task. The problem domain classification task is a binary classification task of whether there is an obstacle in each problem domain. Each sub-network takes the common representation as input and outputs the probability distribution of whether there is an obstacle in the corresponding problem domain classification task.

[0050] The sub-network uses a convolutional neural network, and its output layer generates a probability prediction value of whether there is an obstacle in the corresponding task through a fully connected layer and an activation function;

[0051] The loss function is composed of the independent loss functions of each problem domain classification task, and the independent loss function of each problem domain classification task selects the binary cross entropy loss, and then the independent loss functions are weighted summed to obtain the loss function of the multi-task learning model, wherein the weight coefficient of each independent loss function is set by those skilled in the art based on experience;

[0052] The calculation of the order of visual function data required for training based on the visual ability cluster to which the user belongs, selecting training items from the training item question bank according to the learning difficulty risk level and problem domain of the user's learning difficulty and the order of visual function data, calculating the corresponding training time allocation value and training frequency for each training item, and recording the actual training time and training score for each training item specifically include:

[0053] Step S401: When the user is classified into the kth visual ability cluster, obtain the visual function features of each cluster, which is the mean of the visual function data. and standard deviation , according to the preset visual function characteristic benchmark value , calculate the deviation degree of each visual function data in the cluster ;

[0054] Step S402: sorting the visual function data by priority from large to small according to the degree of deviation, and selecting the arrangement order of the first S visual function features as the arrangement order of the visual function data for training of the visual ability cluster;

[0055] Step S403: The system presets a training item question bank, each training item targets specific visual function data, problem areas, and learning difficulty risk levels;

[0056] Step S404: randomly selecting S training items from the training item database according to the user's learning difficulty risk level and problem areas, as well as the priority of the visual function data;

[0057] Step S405: Calculate the training time allocation value for each training item corresponding to each visual function data based on the deviation degree and the user's learning difficulty risk score and training frequency ;

[0058] Step S406: Allocate a value to each training item according to the calculated training duration and training frequency The system recommends and conducts training for users, and records the actual training time and training score of each training item;

[0059] The training result data of the user after completing the training project is uploaded to the data server, and the score improvement and time reduction of the user on each training project are calculated based on the user's historical training result data. The score difference and time difference of the user compared with other users in the same visual ability cluster are calculated based on the historical training result data of other users in the same visual ability cluster. The training result analysis report is generated specifically including:

[0060] Step S501: After the user completes each training item, the system records the user's score on the training item and obtains the user's training result data, which includes: training item number, training date, actual training time, and training score;

[0061] Step S502: Uploading the user's training result data to the data server to update the user's personal visual function profile;

[0062] Step S503: Extract the user's historical training result data from the data server, compare and analyze the current training result data with the historical training result data, and calculate the user's score improvement on the sth training item. and shortened duration ;

[0063] Step S504: Extract the training result data of other users in the same visual ability cluster as the user from the data server, and calculate the score difference of the user in the kth visual ability cluster on the sth training item relative to other users in the same visual ability cluster. and duration difference ;

[0064] Step S505: The training result analysis report includes a longitudinal training result analysis report and a transverse training result analysis report. The longitudinal training result analysis report of the user is generated by the score improvement and duration reduction of each training item of the user, and the transverse training result analysis report of the user is generated by the score difference and duration difference between the user and other users in the same visual ability cluster.

[0065] One aspect of the present application provides a learning difficulty training method based on visual function testing, comprising:

[0066] User registration and login, improvement of user personal information indicators, and establishment of personal visual function profiles;

[0067] Prompt the user to complete a visual function test, obtain the user's visual function index based on the visual function test, and associate the visual function index with the personal visual function file to obtain visual function data;

[0068] According to the user's visual function data, the user is classified based on the trained visual ability clusters to obtain the visual ability cluster to which the current user belongs. The learning difficulty risk level and problem areas of the user's learning difficulty are predicted based on the pre-established learning difficulty risk assessment model;

[0069] Calculate the order of visual function data required for training based on the visual ability cluster to which the user belongs, select training items from the training item question bank based on the learning difficulty risk level and problem domain of the user's learning difficulty and the order of visual function data, calculate the corresponding training time allocation value and training frequency for each training item, and record the actual training time and training score for each training item;

[0070] The training result data of the user after completing the training project is uploaded to the data server. The score improvement and time reduction of the user in each training project are calculated based on the user's historical training result data. Based on the historical training result data of other users in the same visual ability cluster, the score gap and time gap of the user compared with other users in each training project are calculated, and a training result analysis report is generated.

[0071] One aspect of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in a learning difficulty training method based on visual function detection are implemented.

[0072] One aspect of the present application provides a readable storage medium storing a computer program, wherein the computer program is suitable for being loaded by a processor to execute steps in a learning difficulty training method based on visual function detection.

[0073] The learning difficulties training system based on visual function testing proposed in this application has the following advantages over the existing technology:

[0074] This application comprehensively considers multiple visual function indicators and constructs a relatively comprehensive visual function detection and evaluation system, which can more accurately evaluate the user's visual ability level.

[0075] This application adopts a clustering algorithm to cluster users based on multi-dimensional visual function indicators. Users with similar visual abilities can be automatically classified into the same cluster, making it easier to provide personalized training plans for users in different clusters.

[0076] This application establishes a learning difficulty risk assessment model, which integrates personal information and visual function indicators to predict the level of learning difficulty risk and the learning areas where obstacles may exist, which helps to identify learning difficulty risks early and achieve timely intervention.

[0077] This application intelligently prioritizes the visual functions that require training based on the visual function characteristics of the user's cluster and matches them with corresponding training programs, providing personalized, targeted visual training programs. The system can also dynamically and adaptively adjust the training duration and frequency based on the degree of visual deviation and the risk of learning difficulties.

[0078] This application automatically records training data, and on the one hand, analyzes the user's training results improvement vertically, and on the other hand, compares the gap between the user and other users in the same cluster horizontally, and generates an intuitive training result analysis report to help users understand their own progress and the gap with similar people, thereby enhancing training motivation. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 This is a functional module diagram of a learning difficulties training system based on visual function testing provided by this application;

[0080] Figure 2 Schematic diagram of the visual function examination training management platform provided for this application;

[0081] Figure 3 Schematic diagram of the method for obtaining the training result analysis report provided in this application;

[0082] Figure 4 A flowchart of a method for training learning difficulties based on visual function testing provided in this application;

[0083] Figure 5 A schematic diagram of the structure of an electronic device provided in this application;

[0084] Figure 6 This is a schematic diagram of the structure of a readable storage medium provided by this application. DETAILED DESCRIPTION

[0085] To better understand the present application, various aspects of the present application will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are merely descriptions of exemplary embodiments of the present application and are not intended to limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0086] In the accompanying drawings, the size, dimensions, and shapes of the elements have been slightly adjusted for ease of illustration. The accompanying drawings are for illustration only and are not drawn strictly to scale. As used herein, the terms "substantially," "approximately," and similar terms are used to indicate approximate values, not degrees, and are intended to illustrate inherent deviations in measurements or calculations that would be recognized by a person of ordinary skill in the art. In addition, in this application, the order in which the steps are described does not necessarily represent the order in which these steps would occur in actual operation, unless otherwise specified or inferred from the context.

[0087] It should also be understood that expressions such as "including", "comprising", "having", "containing" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just the individual elements in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the term "exemplary" is intended to refer to an example or illustration.

[0088] Unless otherwise defined, all terms used herein (including engineering and scientific terms) have the same meaning as commonly understood by those skilled in the art to which this application belongs. It should also be understood that, unless otherwise specified in this application, words defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0089] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0090] Example 1

[0091] like Figure 1As shown in FIG, a learning difficulty training system based on visual function testing provided by this application includes:

[0092] Personal profile recording unit, visual function testing unit, grade field positioning unit, personalized training unit, training data uploading unit;

[0093] The personal profile recording unit is used for user registration and login, to improve the user's personal information indicators and to establish a personal visual function profile;

[0094] The visual function detection unit is used to prompt the user to complete the visual function detection, obtain the user's visual function index according to the visual function detection, and associate the visual function index with the personal visual function file to obtain visual function data;

[0095] The level domain positioning unit is used to classify the user based on the user's visual function data and the trained visual ability clustering clusters, obtain the visual ability clustering cluster to which the current user belongs, and predict the learning difficulty risk level and problem domain of the user's learning difficulty based on a pre-established learning difficulty risk assessment model;

[0096] The personalized training unit is used to calculate the arrangement order of the visual function data required for training based on the visual ability cluster to which the user belongs, select training items from the training item question bank according to the learning difficulty risk level and problem domain of the user's learning difficulty and the arrangement order of the visual function data, calculate the corresponding training time allocation value and training frequency for each training item, and record the actual training time and training score for each training item;

[0097] The training data uploading unit is used to upload the training result data of the user after completing the training project to the data server, calculate the user's score improvement and time reduction on each training project based on the user's historical training result data, calculate the score difference and time difference on each training project between the user and other users based on the historical training result data of other users in the same visual ability cluster, and generate a training result analysis report;

[0098] like Figure 2 , which is a schematic diagram of the visual function examination training management platform provided by this application;

[0099] The user registration and login, improvement of user personal information indicators, and establishment of personal visual function files specifically include:

[0100] Step S110: The user registers using their mobile phone number, and a unique identifier is generated for each user;

[0101] Step S120: The user fills in personal information indicators;

[0102] Step S130: storing the user's personal information indicators in a data server and establishing a personal visual function profile;

[0103] The personal information indicators include: the user's daily reading time, daily study time, daily electronic device use time, and daily outdoor exercise time;

[0104] The visual function test includes: inhibition test, perceptual eye position test, vergence range test, stereoscopic vision test, accommodation sensitivity test, binocular balance test, color vision test, signal-to-noise ratio test, saccade ability test, and tracking ability test;

[0105] The inhibition check involves displaying Worth's four points on a display screen. Users wear red and green glasses to view the points and select an answer based on the number of points they see. When the selected answer is correct, the number of questions the user has answered correctly is increased by one.

[0106] The perceptual eye position test involves displaying a red cross and a blue circle on a display screen. The user, wearing red-green glasses, places the red cross in the center of the blue circle and clicks it three times. The coordinates of each click position and the actual center coordinates of the blue circle are obtained.

[0107] The convergence range check refers to the red and green variable vector Figure 1 The image is restored on the display screen. The user wears red and green glasses to view it. The inspection is divided into two steps: the first step is to gather, and the second step is to disperse. In each step, when the user sees the vector diagram separate, click to mark the fracture point, and record the fracture point position marked in the dispersion step and the fracture point position marked in the gathering step respectively;

[0108] The stereoscopic vision test refers to displaying a stereoscopic vision test image on a display screen, and the user wears red and green glasses to watch and select the pattern they see;

[0109] The adjustment sensitivity test refers to displaying a random E-shaped table on the display screen. The user uses a positive and negative mirror to assist in the test and selects the direction of the E-shaped table to be viewed. Each time the E-shaped table is selected, it is randomly rendered again for one minute. The test is divided into three times: left eye, right eye, and both eyes. The number of correctly identified E-shaped table directions is recorded.

[0110] The binocular visual balance test refers to displaying a random E-chart on a display screen. The user does not need to wear glasses and chooses the direction of the E-chart to view. The test is divided into three parts: left eye, right eye, and both eyes. The user's left eye accommodation sensitivity, right eye accommodation sensitivity, and binocular accommodation sensitivity are measured;

[0111] The color vision test refers to displaying a color vision test pattern on a display screen. The user does not need to wear glasses. If the user can distinguish the graphics in the pattern, the user will proceed to identify the next one. If the user cannot distinguish, the user clicks to end the test and the number of patterns correctly identified by the user is recorded;

[0112] The signal-to-noise ratio check refers to displaying a signal-to-noise ratio check pattern on a display screen, and the user wears red and green glasses to view it. If the user recognizes it correctly, the transparency of the signal-to-noise ratio image is adjusted and checked again. Different signal-to-noise ratios will be obtained according to the different transparency images. If the user cannot distinguish, click to end the check.

[0113] The scanning ability test refers to displaying the test logo in random directions and positions on the display screen. The user does not need to wear glasses. The user selects the direction according to the direction of the logo he or she is viewing. The process is repeated within the test time, and the number of logo directions correctly identified by the user is counted.

[0114] The following ability test refers to displaying the test word logo in random directions and irregularly on the display screen. The user does not need to wear glasses. The user chooses the direction according to the direction of the moving word logo he or she is watching and repeats the process within the test time. The user needs to complete the test of one eye and both eyes, and the number of moving word logo directions correctly identified by the user is counted.

[0115] The step of prompting the user to complete a visual function test, obtaining the user's visual function index based on the visual function test, and associating the visual function index with the personal visual function profile to obtain visual function data specifically includes:

[0116] Step S201: Check the user's inhibition level and obtain the number of questions the user answered correctly. Percentage of total questions The percentage of the user's suppression index SI is obtained;

[0117] The calculation formula of the user's suppression index is: ;

[0118] Step S202: Check the user's eye position and obtain the coordinates of each click position. and the actual center coordinates of the blue circle , calculate the user's perceptual eye position deviation PEP;

[0119] The calculation formula of the perceived eye position deviation is: ;

[0120] Step S203: Check the convergence range of the user and obtain the position of the rupture point marked in the divergence step and the rupture point locations marked in the collection step , get the maximum displacement of the vector diagram , calculate the user's gathering and dispersion range ;

[0121] The calculation formula of the convergence range is: ;

[0122] Step S204: Perform a stereoscopic inspection on the user to determine the pattern selected by the user and the correct stereoscopic pattern Whether they are consistent, if they are consistent, the stereoscopic accuracy is 1, otherwise it is 0;

[0123] The calculation formula for the stereoscopic accuracy is: ;

[0124] Step S205: Check the user's adjustment sensitivity and calculate the number of E-table directions correctly identified by the user. Total E-table rendering times The percentage of adjustment sensitivity is obtained ;

[0125] The calculation formula for adjusting the sensitivity is: ;

[0126] Step S206: Perform binocular visual balance check on the user and measure the left eye accommodation sensitivity of the user , right eye adjustment sensitivity , binocular accommodation sensitivity , get the user's binocular balance ;

[0127] The calculation formula for binocular balance is: ;

[0128] Step S207: Perform a color vision test on the user, count the number of patterns correctly identified by the user, and calculate the number of patterns correctly identified by the user. The total number of patterns The ratio of the user's color vision integrity percentage is obtained ;

[0129] The calculation formula for the color vision integrity percentage is: ;

[0130] Step S208: Check the signal-to-noise ratio of the user, and obtain the average amplitude of the signal based on the different signal-to-noise ratios of the images at different transparency levels. and the average amplitude of the noise , calculate the signal-to-noise ratio of the image identified by the user ;

[0131] The calculation formula of the signal-to-noise ratio is: ;

[0132] Step S209: Check the user's scanning ability and count the user's scanning time. Number of word logo directions that can be correctly identified , calculate the user's scanning ability score ;

[0133] The calculation formula of the scanning ability score is: ;

[0134] Step S210: Check the user's ability to follow and count the user's time during the check. Number of directions of moving logos that can be correctly identified , calculate the user's followability score ;

[0135] The calculation formula of the following ability score is: ;

[0136] Step S211: generating visual function indicators based on the test results obtained from the visual function test, wherein the visual function indicators include: inhibition index, perceived eye position deviation, vergence range, stereoscopic accuracy, accommodation sensitivity, binocular balance, color vision integrity percentage, signal-to-noise ratio, saccade ability score, and tracking ability score;

[0137] Step S212: Associating the visual function indicators with the personal visual function profile, collectively referred to as the user's visual function data;

[0138] The method for obtaining the visual ability cluster is:

[0139] Step S301: normalizing the visual function data of N users;

[0140] The normalized calculation formula is: ,in, represents the normalized visual function data of the jth user of the i-th user, represents the original value of the jth visual function data of the i-th user, represents the mean value of the j-th visual function data, represents the standard deviation of the j-th visual function data;

[0141] Step S302: using the normalized visual function data of N users as data points, clustering the data points using a K-means clustering algorithm to obtain K visual ability clusters;

[0142] The specific method for obtaining K visual ability clusters is:

[0143] Step S3021: randomly select K data points as initial cluster centers;

[0144] Step S3022: For each data point , calculate the distance between the data point and each cluster center, and assign it to the cluster with the nearest cluster center;

[0145] Step S3023: for each cluster, recalculate the mean vector of all data points in the cluster as the new cluster center of the cluster;

[0146] Step S3024: Repeat steps S3022 to S3023 until the cluster center of each cluster and the data points within the cluster no longer change, thereby obtaining K visual ability clusters;

[0147] The method for establishing the learning difficulty risk assessment model is:

[0148] Collect data on users diagnosed with learning difficulties, including personal information indicators and visual function indicators, as well as learning difficulty risk levels and problem areas assessed by professionals;

[0149] For each user sample with learning difficulties collected, the risk level of learning difficulties and whether there are obstacles in each problem area are marked based on the professional assessment results. The learning difficulty risk level includes low, medium, and high. The problem areas include but are not limited to reading, writing, and mathematics.

[0150] The visual function data of the learning difficulty user samples are constructed based on personal information indicators and visual function indicators. The learning difficulty user samples are clustered based on the visual ability clustering cluster. The visual ability clustering cluster to which each learning difficulty user sample belongs is determined. The learning difficulty risk score is calculated based on the visual function data and the mean and standard deviation of the visual ability clustering cluster to which it belongs. ;

[0151] The calculation formula for the learning difficulty risk score is: ,in, Indicates the Visual function data, is the weight coefficient of the j-th visual function data, and They represent the mean and standard deviation of the jth visual function data in the kth visual ability cluster where the user is located, and M represents the total number of indicators of the visual function data;

[0152] The value of the weight coefficient of the visual function data is determined by expert knowledge;

[0153] A learning difficulty risk classification model is designed based on the learning difficulty risk score, and a first threshold and a second threshold of the learning difficulty risk score are set. When the learning difficulty risk score is less than or equal to the first threshold, the learning difficulty risk level is low; when the learning difficulty risk score is greater than the first threshold and less than or equal to the second threshold, the learning difficulty risk level is medium; and when the learning difficulty risk score is greater than the second threshold, the learning difficulty risk level is high;

[0154] Preferably, the first threshold is 0.3 and the second threshold is 0.7;

[0155] The learning difficulty risk classification model classifies the user's learning difficulty risk level according to the calculated learning difficulty risk score value according to the first threshold and the second threshold;

[0156] Designing a multi-task learning model, the multi-task learning model comprising a shared feature extraction layer, a task-specific output layer, and a loss function, using samples of users with learning difficulties labeled with whether there are obstacles in each problem area as training samples, using visual function data of the samples of users with learning difficulties as input data, and using binary classification results of whether the samples of users with learning difficulties have obstacles in each problem area as output data, to train the multi-task learning model to obtain a trained multi-task learning model;

[0157] The learning difficulty risk classification model and the multi-task learning model are combined to obtain a learning difficulty risk assessment model;

[0158] The shared feature extraction layer receives visual function data, extracts multimodal input features from different visual function data and performs feature splicing, which serves as the input of the shared feature extraction layer. A deep neural network is used to perform hierarchical abstraction and transformation on the multimodal input features, extracts deep features and outputs them; the output of the deep features is mapped to a low-dimensional latent space to obtain common representations between classification tasks in different problem domains;

[0159] The low-dimensional latent space is realized by a fully connected layer, and its output dimension is smaller than the input dimension, thereby achieving the effect of dimensionality reduction and compression.

[0160] The task feature output layer designs an output sub-network for each problem domain classification task. The problem domain classification task is a binary classification task of whether there is an obstacle in each problem domain. Each sub-network takes the common representation as input and outputs the probability distribution of whether there is an obstacle in the corresponding problem domain classification task.

[0161] The sub-network uses a convolutional neural network, and its output layer generates a probability prediction value of whether there is an obstacle in the corresponding task through a fully connected layer and an activation function;

[0162] The loss function is composed of the independent loss functions of each problem domain classification task. The independent loss function of each problem domain classification task selects binary cross entropy loss, and then the independent loss functions are weighted summed to obtain the loss function of the multi-task learning model. Among them, the weight coefficient of each independent loss function is set by technical personnel in this field based on experience.

[0163] The calculation of the order of visual function data required for training based on the visual ability cluster to which the user belongs, selecting training items from the training item question bank according to the learning difficulty risk level and problem domain of the user's learning difficulty and the order of visual function data, calculating the corresponding training time allocation value and training frequency for each training item, and recording the actual training time and training score for each training item specifically include:

[0164] Step S401: When the user is classified into the kth visual ability cluster, obtain the visual function features of each cluster, which is the mean of the visual function data. and standard deviation , according to the preset visual function characteristic benchmark value , calculate the deviation degree of each visual function data in the cluster;

[0165] The calculation formula for the degree of deviation is: ,in, is the visual function characteristic benchmark value of the j-th visual function data in the k-th visual ability cluster;

[0166] The visual function characteristic reference value is set by those skilled in the art according to actual needs;

[0167] Step S402: sorting the visual function data by priority from large to small according to the degree of deviation, and selecting the arrangement order of the first S visual function features as the arrangement order of the visual function data for training of the visual ability cluster;

[0168] Step S403: The system presets a training item question bank, each training item targets specific visual function data, problem areas, and learning difficulty risk levels;

[0169] Step S404: randomly selecting S training items from the training item database according to the user's learning difficulty risk level and problem areas, as well as the priority of the visual function data;

[0170] Step S405: Calculate the training time allocation value for each training item corresponding to each visual function data based on the deviation degree and the user's learning difficulty risk score and training frequency ;

[0171] The calculation formula of the training duration allocation value is: ,in, The duration of basic training, is the maximum value among all deviation values, Score the user's learning difficulty risk, and are the influence coefficients of the deviation value item and the learning difficulty risk score item on the training time allocation value;

[0172] The training time allocation value is the training time of the training project allocated by the system to the user, and its value is not necessarily equal to the training time actually consumed by the user;

[0173] The weight coefficients of the deviation value item and the learning difficulty risk score item are set by those skilled in the art based on experience;

[0174] The calculation formula for the training frequency is: ,in, Indicates the frequency of basic training, 、 are the influence coefficients of the deviation value item and the learning difficulty risk score item on the training frequency;

[0175] The influence coefficients of the deviation value item and the learning difficulty risk score item on the training frequency are set by those skilled in the art based on experience;

[0176] Step S406: Allocate a value to each training item according to the calculated training duration and training frequency The system recommends and conducts training for users, and records the actual training time and training score of each training item;

[0177] The training result data of the user after completing the training project is uploaded to the data server, and the score improvement and time reduction of the user on each training project are calculated based on the user's historical training result data. The score difference and time difference of the user compared with other users in the same visual ability cluster are calculated based on the historical training result data of other users in the same visual ability cluster. The training result analysis report is generated specifically including:

[0178] Step S501: After the user completes each training item, the system records the user's score on the training item and obtains the user's training result data, which includes: training item number, training date, actual training time, and training score;

[0179] Step S502: Uploading the user's training result data to the data server to update the user's personal visual function profile;

[0180] Step S503: Extract the user's historical training result data from the data server, compare and analyze the current training result data with the historical training result data, and calculate the user's score improvement on the sth training item. and shortened duration ;

[0181] The calculation formula for the score improvement is: ,in, represents the current training score of the s-th training item, represents the bth historical training score of the sth training item, Indicates the total number of historical training times;

[0182] The calculation formula for the shortened duration is: ,in, Indicates the current actual training time of the sth training item, represents the bth historical actual training time of the sth training item;

[0183] Step S504: extracting training result data of other users in the same visual ability cluster as the user from the data server, and calculating the score difference and time difference of the user in the kth visual ability cluster on the sth training item relative to other users in the same visual ability cluster;

[0184] The calculation formula for the score gap is: ,in, represents the total number of users in the k-th visual ability cluster, represents the average training score of the gth other user in the kth visual ability cluster on the sth training item;

[0185] The calculation formula for the time difference is: ,in, represents the average actual training time of the gth other user in the kth visual ability cluster on the sth training item;

[0186] Step S505: The training result analysis report includes a longitudinal training result analysis report and a transverse training result analysis report. The longitudinal training result analysis report of the user is generated by the score improvement and duration reduction of each training item of the user, and the transverse training result analysis report of the user is generated by the score difference and duration difference between the user and other users in the same visual ability cluster.

[0187] like Figure 3 , which is a schematic diagram of a method for obtaining a training result analysis report provided by this application;

[0188] For example, taking the training score of training item s as an example, the longitudinal training result analysis report is:

[0189] First training score: 60 points;

[0190] Second training score: 70 points, improvement value 10 points, improvement rate 16.7%;

[0191] The third training score: 75 points, improvement value 5 points, improvement rate 7.1%;

[0192] …

[0193] Cumulative score improvement: 15 points, average improvement rate: 11.9%;

[0194] …

[0195] For example, taking the score difference of the user's training item s in the kth visual ability cluster as an example, the horizontal training result analysis report is as follows:

[0196] The difference between the user's score and other user 1: 5 points;

[0197] The difference between the user's score and other user 2: -4 points;

[0198] …

[0199] User score ranking: 5th, top 10%;

[0200] …

[0201] The above steps, through the combination of vertical and horizontal reports, comprehensively evaluate the user's training effects, understand their progress over time, and their strengths and weaknesses compared with similar users, which helps to enhance the user's training motivation and confidence.

[0202] Example 2

[0203] like Figure 4 As shown, this application provides a learning difficulty training method based on visual function testing, including:

[0204] Step S100: User registration and login, completing user personal information indicators, and establishing a personal visual function profile;

[0205] Step S200: prompting the user to complete a visual function test, obtaining the user's visual function index based on the visual function test, and associating the visual function index with the personal visual function profile to obtain visual function data;

[0206] Step S300: Based on the user's visual function data, the user is classified based on the trained visual ability clusters to obtain the visual ability cluster to which the current user belongs, and the learning difficulty risk level and problem areas of the user's learning difficulty are predicted based on the pre-established learning difficulty risk assessment model;

[0207] Step S400: Calculate the order of visual function data required for training based on the visual ability cluster to which the user belongs, select training items from the training item database based on the learning difficulty risk level and problem domain of the user's learning difficulty and the order of visual function data, calculate the corresponding training time allocation value and training frequency for each training item, and record the actual training time and training score for each training item;

[0208] Step S500: Upload the training result data after the user completes the training project to the data server, calculate the user's score improvement and time reduction in each training project based on the user's historical training result data, calculate the user's score difference and time difference in each training project compared with other users based on the historical training result data of other users in the same visual ability cluster, and generate a training result analysis report.

[0209] Example 3

[0210] Figure 5 This is a schematic diagram of the electronic device structure provided by an embodiment of the present application. Figure 5 According to another aspect of the present application, an electronic device is provided. The electronic device may include one or more processors and one or more memories. The memories may store computer-readable code, which, when executed by the one or more processors, may execute the above-described method for training learning difficulties based on visual function testing.

[0211] The method or system according to the embodiment of the present application can also be used by Figure 5 The electronic device architecture shown in FIG. Figure 5As shown, the electronic device may include a bus, one or more CPUs, a read-only memory (ROM), a random access memory (RAM), a communication port connected to a network, an input / output component, a hard disk, etc. The storage device in the electronic device, such as a ROM or a hard disk, can store a learning difficulty training method based on visual function detection provided in the present application. A learning difficulty training method based on visual function detection may, for example, include: user registration and login, improving user personal information indicators, and establishing a personal visual function profile; prompting the user to complete the visual function test, obtaining the user's visual function indicators based on the visual function test, associating the visual function indicators with the personal visual function profile to obtain visual function data; based on the user's visual function data, classifying the user based on the trained visual ability clustering cluster to obtain the visual ability clustering cluster to which the current user belongs, and predicting the user's learning difficulty risk level and problem area based on a pre-established learning difficulty risk assessment model; calculating the visual function that needs to be trained based on the visual ability clustering cluster to which the user belongs. According to the arrangement order of the visual ability data, the learning difficulty risk level and problem areas of the user's learning difficulties and the arrangement order of the visual function data, training items are selected from the training item question bank, and the corresponding training time allocation value and training frequency of each training item are calculated, and the actual training time and training score of each training item are recorded; the training result data after the user completes the training item is uploaded to the data server, and the score improvement and time reduction of the user on each training item are calculated based on the user's historical training result data, and the score gap and time gap of the user compared with other users on each training item are calculated based on the historical training result data of other users in the same visual ability cluster, and a training result analysis report is generated. Furthermore, the electronic device may also include a user interface. Of course, Figure 5 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 5 One or more components of an electronic device are shown.

[0212] Example 4

[0213] Figure 6 This is a schematic diagram of the structure of a readable storage medium provided by an embodiment of the present application. Figure 6 , a computer-readable storage medium according to one embodiment of the present application. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by a processor, the method for training learning difficulties based on visual function testing according to an embodiment of the present application described with reference to the above figures can be executed. The storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), a hard disk, flash memory, etc.

[0214] In addition, according to the implementation mode of the present application, the process described in the above reference flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, which stores machine-readable instructions, and the machine-readable instructions can be run by a processor to execute instructions corresponding to the method steps provided by the present application, such as: user registration and login, improving user personal information indicators, and establishing a personal visual function profile; prompting the user to complete a visual function test, obtaining the user's visual function indicators based on the visual function test, associating the visual function indicators with the personal visual function profile to obtain visual function data; based on the user's visual function data, classifying the user based on the trained visual ability clustering cluster, obtaining the visual ability clustering cluster to which the current user belongs, and predicting the user's learning difficulty risk based on a pre-established learning difficulty risk assessment model, etc. level and problem domain; calculate the order of visual function data required for training based on the visual ability cluster to which the user belongs, select training items from the training item question bank according to the learning difficulty risk level and problem domain of the user's learning difficulty and the order of visual function data, calculate the corresponding training time allocation value and training frequency for each training item, and record the actual training time and training score of each training item; upload the training result data after the user completes the training item to the data server, calculate the score improvement and time reduction of the user on each training item based on the user's historical training result data, calculate the score difference and time difference of the user compared with other users on each training item based on the historical training result data of other users in the same visual ability cluster, and generate a training result analysis report. When the computer program is executed by the central processing unit (CPU), it performs the above functions defined in the method of this application.

[0215] The methods, apparatus, and devices of the present application may be implemented in many ways. For example, the methods, apparatus, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the methods according to the present application. Therefore, the present application also covers recording media that store programs for executing the methods according to the present application.

[0216] In addition, the parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0217] The above-described specific embodiments further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is merely a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A learning difficulties training system based on visual function testing, characterized in that: include: Personal profile recording unit, visual function testing unit, grade field positioning unit, personalized training unit, training data uploading unit; The personal profile recording unit is used for user registration and login, to improve the user's personal information indicators and to establish a personal visual function profile; The visual function detection unit is used to prompt the user to complete the visual function detection, obtain the user's visual function index according to the visual function detection, and associate the visual function index with the personal visual function file to obtain visual function data; The level domain positioning unit is used to classify the user based on the user's visual function data and the trained visual ability clustering clusters, obtain the visual ability clustering cluster to which the current user belongs, and predict the learning difficulty risk level and problem domain of the user's learning difficulty based on a pre-established learning difficulty risk assessment model; The personalized training unit is used to calculate the arrangement order of the visual function data required for training based on the visual ability cluster to which the user belongs, select training items from the training item question bank according to the learning difficulty risk level and problem domain of the user's learning difficulty and the arrangement order of the visual function data, calculate the corresponding training time allocation value and training frequency for each training item, and record the actual training time and training score for each training item; The training data uploading unit is used to upload the training result data of the user after completing the training project to the data server, calculate the user's score improvement and time reduction on each training project based on the user's historical training result data, calculate the score difference and time difference on each training project between the user and other users based on the historical training result data of other users in the same visual ability cluster, and generate a training result analysis report; The method for establishing the learning difficulty risk assessment model is: Collect data on users diagnosed with learning difficulties, including personal information indicators and visual function indicators, as well as learning difficulty risk levels and problem areas assessed by professionals; For each sample of users with learning difficulties collected, the risk level of learning difficulties is marked based on the professional assessment results, as well as whether there are obstacles in each problem area. The learning difficulty risk level includes low, medium, and high; The visual function data of the learning difficulty user samples are constructed based on personal information indicators and visual function indicators. The learning difficulty user samples are clustered based on the visual ability clustering cluster. The visual ability clustering cluster to which each learning difficulty user sample belongs is determined. The learning difficulty risk score is calculated based on the visual function data and the mean and standard deviation of the visual ability clustering cluster to which it belongs. ; A learning difficulty risk classification model is designed based on the learning difficulty risk score, and a first threshold and a second threshold of the learning difficulty risk score are set. When the learning difficulty risk score is less than or equal to the first threshold, the learning difficulty risk level is low; when the learning difficulty risk score is greater than the first threshold and less than or equal to the second threshold, the learning difficulty risk level is medium; and when the learning difficulty risk score is greater than the second threshold, the learning difficulty risk level is high; Designing a multi-task learning model, the multi-task learning model comprising a shared feature extraction layer, a task-specific output layer, and a loss function, using samples of users with learning difficulties labeled with whether there are obstacles in each problem area as training samples, using visual function data of the samples of users with learning difficulties as input data, and using binary classification results of whether the samples of users with learning difficulties have obstacles in each problem area as output data, to train the multi-task learning model to obtain a trained multi-task learning model; The learning difficulty risk classification model and the multi-task learning model are combined to obtain a learning difficulty risk assessment model.

2. A learning difficulties training system based on visual function testing as claimed in claim 1, characterized in that: The step of prompting the user to complete a visual function test, obtaining the user's visual function index based on the visual function test, and associating the visual function index with the personal visual function profile to obtain visual function data specifically includes: Perform a suppression check on the user and obtain the number of questions the user answered correctly Percentage of total questions The percentage of the user's suppression index SI is obtained; Perform a perceptual eye position check on the user to obtain the coordinates of each click position and the actual center coordinates of the blue circle , calculate the user's perceptual eye position deviation PEP; Check the convergence range of the user and obtain the location of the rupture point marked in the divergence step and the rupture point locations marked in the collection step , get the maximum displacement of the vector diagram , calculate the user's gathering and dispersion range ; Perform stereoscopic vision checks on users to determine the pattern they select and the correct stereoscopic pattern Whether they are consistent, if they are consistent, the stereoscopic accuracy is 1, otherwise it is 0; Check the user's adjustment sensitivity and count the number of E-table directions correctly identified by the user Total E-table rendering times The percentage of adjustment sensitivity is obtained ; Perform binocular visual balance check on the user and measure the user's left eye accommodation sensitivity , right eye adjustment sensitivity , binocular accommodation sensitivity , get the user's binocular balance ; Perform color vision tests on users, count the number of patterns they correctly identify, and calculate the number of patterns they correctly identify. The total number of patterns The ratio of the user's color vision integrity percentage is obtained ; Perform a signal-to-noise ratio check on the user, and obtain the average amplitude of the signal based on the different signal-to-noise ratios of images at different transparency levels. and the average amplitude of the noise , calculate the signal-to-noise ratio of the image identified by the user ; Check the user's scanning ability and count the user's inspection time Number of word logo directions that can be correctly identified , calculate the user's scanning ability score ; Check the user's follow-up ability and count the user's time during the check Number of directions of moving logos that can be correctly identified , calculate the user's followability score ; Generating visual function indices based on the test results obtained from the visual function test, the visual function indices include: inhibition index, perceived eye position deviation, vergence range, stereoscopic accuracy, accommodation sensitivity, binocular balance, color vision integrity percentage, signal-to-noise ratio, saccade ability score, and tracking ability score; The visual function indicators are associated with the personal visual function profile, collectively referred to as the user's visual function data.

3. A learning difficulties training system based on visual function testing as claimed in claim 2, characterized in that: The method for obtaining the visual ability cluster is: Normalize the visual function data of N users; The normalized calculation formula is: ,in, represents the normalized visual function data of the jth user of the i-th user, represents the original value of the jth visual function data of the i-th user, represents the mean value of the j-th visual function data, represents the standard deviation of the j-th visual function data; The normalized visual function data of N users are used as data points, and the data points are clustered using the K-means clustering algorithm to obtain K visual ability clusters; The specific method for obtaining K visual ability clusters is: Step S3021: randomly select K data points as initial cluster centers; Step S3022: For each data point , calculate the distance between the data point and each cluster center, and assign it to the cluster with the nearest cluster center; Step S3023: for each cluster, recalculate the mean vector of all data points in the cluster as the new cluster center of the cluster; Step S3024: Repeat steps S3022 to S3023 until the cluster center of each cluster and the data points within the cluster no longer change, thereby obtaining K visual ability clusters.

4. A learning difficulties training system based on visual function testing as claimed in claim 3, characterized in that: The shared feature extraction layer receives visual function data, extracts multimodal input features of different visual function data and performs feature splicing as input to the shared feature extraction layer, uses a deep neural network to perform hierarchical abstraction and transformation on the multimodal input features, extracts deep features and outputs them; maps the output of the deep features to a low-dimensional latent space to obtain common representations between classification tasks in different problem domains.

5. The learning difficulties training system based on visual function testing according to claim 4, characterized in that: The calculation of the order of visual function data required for training based on the visual ability cluster to which the user belongs, selecting training items from the training item question bank according to the learning difficulty risk level and problem domain of the user's learning difficulty and the order of visual function data, calculating the corresponding training time allocation value and training frequency for each training item, and recording the actual training time and training score for each training item specifically include: When the user is classified into the kth visual ability cluster, the visual function features of each cluster are obtained, and the visual function features are the mean values ​​of the visual function data. and standard deviation , according to the preset visual function characteristic benchmark value , calculate the deviation degree of each visual function data in the cluster ; Prioritize each visual function data according to the degree of deviation from large to small, and select the arrangement order of the first S visual function features as the arrangement order of the visual function data for training of the visual ability cluster; The system has a preset training project question bank, each of which targets specific visual function data, problem areas, and learning difficulty risk levels; According to the user's learning difficulty risk level and problem area, as well as the priority of visual function data, S training items are randomly selected from the training item database; Calculate the training time allocation value for each training item corresponding to each visual function data based on the degree of deviation and the user's learning difficulty risk score and training frequency ; Each training item is assigned a value based on the calculated training duration and training frequency The system recommends and conducts training for users, and records the actual training time and training score of each training item for users.

6. A learning difficulties training system based on visual function testing as claimed in claim 5, characterized in that: The training result data of the user after completing the training project is uploaded to the data server, and the score improvement and time reduction of the user on each training project are calculated based on the user's historical training result data. The score difference and time difference of the user compared with other users in the same visual ability cluster are calculated based on the historical training result data of other users in the same visual ability cluster. The training result analysis report is generated specifically including: After the user completes each training project, the system records the user's score on the training project and obtains the user's training result data, which includes: training project number, training date, actual training time, and training score; Upload the user's training result data to the data server and update the user's personal visual function profile; Extract the user's historical training result data from the data server, compare and analyze the current training result data with the historical training result data, and calculate the user's score improvement on the sth training item. and shortened duration ; Extract the training result data of other users in the same visual ability cluster as the user from the data server, and calculate the score difference of the user in the kth visual ability cluster on the sth training item relative to other users in the same visual ability cluster. and duration difference ; The training result analysis report includes a longitudinal training result analysis report and a transverse training result analysis report. The longitudinal training result analysis report of the user is generated by the score improvement and duration reduction of each training item of the user, and the transverse training result analysis report of the user is generated by the score difference and duration difference between the user and other users in the same visual ability cluster.

7. A learning difficulties training method based on visual function testing, characterized in that: include: User registration and login, improvement of user personal information indicators, and establishment of personal visual function profiles; Prompt the user to complete a visual function test, obtain the user's visual function index based on the visual function test, and associate the visual function index with the personal visual function file to obtain visual function data; According to the user's visual function data, the user is classified based on the trained visual ability clusters to obtain the visual ability cluster to which the current user belongs. The learning difficulty risk level and problem areas of the user's learning difficulty are predicted based on the pre-established learning difficulty risk assessment model; Calculate the order of visual function data required for training based on the visual ability cluster to which the user belongs, select training items from the training item question bank based on the learning difficulty risk level and problem domain of the user's learning difficulty and the order of visual function data, calculate the corresponding training time allocation value and training frequency for each training item, and record the actual training time and training score for each training item; Upload the training result data of the user after completing the training project to the data server, calculate the user's score improvement and time reduction in each training project based on the user's historical training result data, and calculate the score difference and time difference between the user and other users in the same visual ability cluster based on the historical training result data of other users in each training project, and generate a training result analysis report; The method for establishing the learning difficulty risk assessment model is: Collect data on users diagnosed with learning difficulties, including personal information indicators and visual function indicators, as well as learning difficulty risk levels and problem areas assessed by professionals; For each sample of users with learning difficulties collected, the risk level of learning difficulties is marked based on the professional assessment results, as well as whether there are obstacles in each problem area. The learning difficulty risk level includes low, medium, and high; The visual function data of the learning difficulty user samples are constructed based on personal information indicators and visual function indicators. The learning difficulty user samples are clustered based on the visual ability clustering cluster. The visual ability clustering cluster to which each learning difficulty user sample belongs is determined. The learning difficulty risk score is calculated based on the visual function data and the mean and standard deviation of the visual ability clustering cluster to which it belongs. ; A learning difficulty risk classification model is designed based on the learning difficulty risk score, and a first threshold and a second threshold of the learning difficulty risk score are set. When the learning difficulty risk score is less than or equal to the first threshold, the learning difficulty risk level is low; when the learning difficulty risk score is greater than the first threshold and less than or equal to the second threshold, the learning difficulty risk level is medium; and when the learning difficulty risk score is greater than the second threshold, the learning difficulty risk level is high; Designing a multi-task learning model, the multi-task learning model comprising a shared feature extraction layer, a task-specific output layer, and a loss function, using samples of users with learning difficulties labeled with whether there are obstacles in each problem area as training samples, using visual function data of the samples of users with learning difficulties as input data, and using binary classification results of whether the samples of users with learning difficulties have obstacles in each problem area as output data, to train the multi-task learning model to obtain a trained multi-task learning model; The learning difficulty risk classification model and the multi-task learning model are combined to obtain a learning difficulty risk assessment model.

8. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the learning difficulty training method based on visual function detection as claimed in claim 7 are implemented.

9. A readable storage medium, characterized in that The readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps of the learning difficulty training method based on visual function testing as claimed in claim 7.

Images