Eyeball movement training method and device based on artificial intelligence, training instrument and medium

Abstract

The invention relates to an eyeball movement training method and device based on artificial intelligence, a training instrument and a medium, and is applied to the technical field of eyeball movement training. Judging whether a historical training record exists or not based on the user information; if the historical training record exists, generating a user training scheme based on the historical training record; if the historical training record does not exist, determining an associated user based on the user information; acquiring an associated training record of the first training of the associated user; and generating the user training scheme based on the associated training record. The reliability of the training scheme can be improved, so that the training effect is improved.

Description

Technical Field

[0001] This application relates to the technical field of eye movement training, and in particular to an artificial intelligence-based eye movement training method, device, training instrument, and medium. Background Technology

[0002] Eye movement ability includes fixation, saccades, and tracking abilities. It is the foundation for the function of binocular vision and an important factor affecting the academic performance of children and adolescents. Therefore, eye movement ability training is of utmost importance.

[0003] Currently, eye movement training is mainly done manually, with trainers manually developing training plans and recording the training process. This requires not only experience from the trainers but also a high degree of concentration during the training. However, due to limitations in experience and energy, trainers may make oversights in developing training plans and recording data, resulting in lower reliability of the training plans and thus affecting the training effect. Summary of the Invention

[0004] To improve the reliability of training programs and thus enhance training effectiveness, this application provides an artificial intelligence-based eye movement training method, device, training instrument, and medium.

[0005] Firstly, this application provides an eye movement training method based on artificial intelligence, employing the following technical solution: An artificial intelligence-based eye movement training method includes: Obtain user information; Based on the user information, determine whether there are historical training records; If the historical training records exist, a user training plan is generated based on the historical training records; If the historical training record does not exist, the associated user is determined based on the user information; Obtain the associated training records of the first training session of the associated user; The user training scheme is generated based on the associated training records.

[0006] By adopting the above technical solution, before formulating the current user training plan, it is first determined whether there is a historical training record for the user. If there is a historical training record, the current user training plan is automatically generated based on the user's historical training record, which improves the reliability of the user training plan. If there is no historical training record for the user, the associated user is determined based on the user information, and the current user training plan is automatically generated based on the associated user's first training record, which improves the reliability of the user training plan and thus improves the training effect.

[0007] Optionally, the user training scheme includes a gaze ability training scheme, and generating the user training scheme based on the historical training records includes: Based on the historical training records, determine the last character change rate, the last error probability, and the last average delay time for each position; The initial character change rate at each position is determined based on the previous error probability, the previous average delay duration, and the previous character change rate. The positions are sorted from smallest to largest based on the initial character change rate to obtain the position sorting result; An initial gaze ability training scheme is generated based on the position sorting results and the initial character change rate corresponding to each position. Acquire the first user's training data based on a preset duration; The initial gaze ability training scheme is adjusted based on the first user training data to obtain the gaze ability training scheme.

[0008] By adopting the above technical solution, the initial character change rate for this training is first determined based on the previous error probability, the previous average delay time, and the previous character change rate at each user position, thus obtaining the initial gaze ability training scheme. During the training process, the first user's training data is collected every preset time interval, so that the initial gaze ability training scheme can be adjusted to obtain the gaze ability training scheme, making the gaze ability training scheme more reliable.

[0009] Optionally, the user training scheme includes a saccade ability training scheme, and the step of generating the user training scheme based on the historical training records includes: Based on the historical training records, determine the last character interval, the last error probability, and the last completion time; The completion time threshold is determined based on the previous character interval; The current character interval is determined based on the previous error probability, the previous completion time, and the completion time threshold. The saccadic perception training scheme is determined based on the current character interval.

[0010] By adopting the above technical solution, and by fully considering the previous error probability, the previous completion time, and the completion time threshold to determine the current character interval, a saccade training scheme is obtained, which makes the saccade training scheme more reliable.

[0011] Optionally, the user training scheme includes a following ability training scheme, and the step of generating the user training scheme based on the historical training records includes: Based on the historical training records, determine the error probability and average delay time of each character movement trajectory at various character movement speeds; The current character movement speed corresponding to each character movement trajectory is determined based on the error probability and the average delay duration. The initial following ability training scheme is determined based on the trajectory difficulty of each character movement trajectory and the corresponding character movement speed. Obtain the current training iteration, which is the number of times the character movement trajectory has been looped in the current training round; The training data of the second user is obtained based on the current number of training iterations; The initial following ability training scheme is adjusted based on the second user training data to obtain the following ability training scheme.

[0012] By adopting the above technical solution, an initial following ability training scheme is first generated by analyzing historical training records. Then, the training data of the second user is obtained by judging the current training number, and the initial following ability training scheme is adjusted according to the training data of the second user to obtain the following ability training scheme, which makes the following ability training scheme more reliable.

[0013] Optionally, determining the associated user based on the user information includes: Based on the user information, determine the user's age, gender, eye abnormalities, and cognitive abilities. User similarity is calculated based on the K-nearest neighbor algorithm, the user's age, gender, eye abnormality information, cognitive ability information, and historical user information. Historical users whose similarity to the user is higher than a preset similarity are identified as associated users.

[0014] By adopting the above technical solution, user similarity is calculated by comprehensively considering user age, gender, eye abnormality information, and cognitive ability information, and associated users are identified based on user similarity, thereby improving the reliability of associated users.

[0015] Optionally, the associated training records include associated fixation ability training records, associated saccade ability training records, and associated following ability training records. Generating the user training scheme based on the associated training records includes: The rate of change of the gazed character is determined based on the error probability and average delay duration in the associated gaze ability training record. The interval between scanned characters is determined based on the error probability and completion time in the associated scan ability training records. The movement trajectory of the following character and the corresponding movement speed of the following character are determined based on the error probability and average delay time in the training record of the associated following ability. The user training scheme is determined based on the speed of change of the gaze character, the interval of the saccade character, the trajectory of the following character, and the corresponding speed of the following character.

[0016] Optionally, determining the user training scheme based on the gaze character change rate, the saccade character interval, the following character movement trajectory, and the corresponding following character movement speed includes: The gaze ability training scheme is determined based on the gaze character change speed and position selection algorithm. A saccharification ability training scheme is determined based on the saccharification character interval; A training scheme for the following ability is determined based on the movement trajectory of the following character and the corresponding movement speed of the following character. The user training scheme is determined based on the saccadic ability training scheme, the fixation ability training scheme, and the following ability training scheme.

[0017] Secondly, this application provides an eye movement training device based on artificial intelligence, which adopts the following technical solution: An artificial intelligence-based eye movement training device includes: The user information acquisition module is used to acquire user information; The historical record determination module is used to determine whether there are historical training records based on the user information; The first scheme generation module is used to generate a user training scheme based on the historical training record if the historical training record exists. The associated user determination module is used to determine the associated user based on the user information if the historical training record does not exist. The associated record acquisition module is used to acquire the associated training records of the first training of the associated user; The second scheme generation module is used to generate the user training scheme based on the associated training records.

[0018] By adopting the above technical solution, before formulating the current user training plan, it is first determined whether there is a historical training record for the user. If there is a historical training record, the current user training plan is generated based on the user's historical training record, which improves the reliability of the user training plan. If there is no historical training record for the user, the associated user is determined based on the user information, and the current user training plan is generated based on the associated user's first training record, which improves the reliability of the user training plan and thus improves the training effect.

[0019] Thirdly, this application provides a training device, which adopts the following technical solution: A training device includes a training device body and a printing device for printing the training device data. The training device body includes a processor, a memory, a screen, a microphone, and a power switch. The processor, the memory, the screen, the microphone, and the printing device are all electrically connected to the power switch. The memory, the screen, the microphone, and the printing device are all communicatively connected to the processor. The microphone, the screen, and the printing device are all communicatively connected to the memory. The memory stores a computer program that can be loaded by the processor and executed as described in any one of the first aspects of the artificial intelligence-based eye movement training method.

[0020] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium storing a computer program capable of being loaded by a processor and executing the artificial intelligence-based eye movement training method described in any one of the first aspects. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating an artificial intelligence-based eye movement training method provided in an embodiment of this application.

[0022] Figure 2 This is a structural block diagram of an artificial intelligence-based eye movement training device provided in an embodiment of this application.

[0023] Figure 3 This is a structural block diagram of the training device provided in the embodiments of this application. Detailed Implementation

[0024] The present application will be further described in detail below with reference to the accompanying drawings.

[0025] This application provides an AI-based eye movement training method, which can be executed by an electronic device. The electronic device can be a server or a terminal device. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smartphone, tablet, desktop computer, etc., but is not limited to these.

[0026] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0028] like Figure 1 As shown, an artificial intelligence-based eye movement training method is described in the following steps (S101-S106): Step S101: Obtain user information.

[0029] After the training device is turned on, user information can be obtained through voice dialogue between the device and the user, or the user can manually enter user information on the device. User information includes, but is not limited to, user name, user age, user gender, user ID number, user eye abnormality information, and user cognitive ability information. User eye abnormality information includes, but is not limited to, myopia, hyperopia, astigmatism, strabismus, and amblyopia, while user cognitive ability information includes, but is not limited to, attention deficit and memory problems.

[0030] When obtaining user information, it can be done in two steps. First, obtain the user's ID number. If the user's ID number exists in the database corresponding to the historical user information, there is no need to obtain other types of user information again. If the user's ID number does not exist in the database corresponding to the historical user information, then obtain other types of user information again (by voice or manual input).

[0031] Step S102: Determine whether there are historical training records based on user information.

[0032] If the current user information exists in the database corresponding to the historical user information, it means that the user has participated in training in the past and there is a historical training record in the database; if the current user information does not exist in the database corresponding to the historical user information, it means that the user has not participated in training in the past and there is no historical training record in the database.

[0033] Step S103: If historical training records exist, generate a user training plan based on the historical training records.

[0034] If the database contains the user's historical training records, the current user training plan is generated based on the training information in the historical records. The user's eye movement training includes fixation ability training, saccade ability training, and tracking ability training. Therefore, the user training plan includes fixation ability training plan, saccade ability training plan, and tracking ability training plan. When the user is undergoing eye movement training, the training device will divide the screen into a preset number (e.g., 10*10) points (positions). Characters are displayed on the screen of the training device according to certain rules at each point. The user quickly says the characters on the screen while keeping their head still. The training device receives the user's voice through the microphone and processes the received voice through the processor of the training device to generate the user's training performance. The characters include, but are not limited to, graphics (children's pictures, mathematical graphics), Arabic numerals (ones, tens, hundreds), letters (lowercase, uppercase, mixed), Chinese characters (simple Chinese characters - Chinese first grade level, complex Chinese characters - Chinese third grade level), directional arrows (up, down, left, right, upper left, lower left, upper right, lower right), etc.

[0035] When conducting training, the training device can use both artificial intelligence mode and manual mode. In artificial intelligence mode, the device can intelligently generate a training plan for the user, allowing the user to complete the training according to the plan. In manual mode, the trainer manually controls the device to enable the user to complete the training.

[0036] Specifically, generating a user training plan based on historical training records includes: determining the last character change rate, last error probability, and last average latency for each position based on historical training records; determining the initial character change rate for each position based on the last error probability, last average latency, and last character change rate; sorting each position from smallest to largest based on the initial character change rate to obtain a position sorting result; generating an initial fixation ability training plan based on the position sorting result and the initial character change rate corresponding to each position; acquiring the first user training data based on a preset duration; and adjusting the initial fixation ability training plan based on the first user training data to obtain the fixation ability training plan.

[0037] The fixation training involves the training device identifying a position where randomly changing characters are displayed at a certain speed. The user quickly pronounces the characters while keeping their head still. After one training cycle (a cycle of preset duration, e.g., 1 minute, where the character display position and speed of change remain constant), the training device calculates the user's error probability and average latency. If the error probability is less than the preset error probability and the average latency is less than the preset average latency, the training for that cycle is considered successful, and the user moves to the next position to continue training. If the error probability is greater than or equal to the preset error probability and / or... If the average delay is greater than or equal to the preset average delay, it means that the training at that position according to the character change speed is unsuccessful. Training continues at that position according to the character change speed until the training at that position according to the character change speed is successful. That is, after each training cycle, the error probability and average delay are judged. If the training is successful, the position and / or character change speed are changed for training. If the training is unsuccessful, training continues according to the current position and the current character change speed. It is worth noting that the training time for one fixation ability training session is the first preset time (e.g., 3 minutes). When the first preset time is reached, the fixation ability training is stopped.

[0038] The errors that exist in the training process mainly include omission (reading a character less than expected), insertion (reading a character that does not exist), and repetition (reading a character repeatedly). The delay time is the interval between the appearance of a character and the user reading the character. The average delay time is the average of all delay times in a training cycle.

[0039] In this embodiment, during gaze training, a gaze training plan is generated based on the user's historical training records. Specifically, this includes: retrieving the character change rate, error probability, and average latency from the gaze training records in the historical records for each position, i.e., the previous character change rate, the previous error probability, and the previous average latency. If the previous error probability of a position is less than a preset error probability and the previous average latency is less than a preset average latency, then the character change rate after the previous character change rate in the character change rate sequence (pre-set, e.g., 10 seconds / character, 9 seconds / character, ..., 3 seconds / character) is determined as the initial character change rate for that position. Initial character change rate; if the previous error probability of a position is greater than or equal to the preset error probability and / or the previous average delay time is greater than or equal to the preset average delay time, then the previous character change rate is determined as the initial character change rate of that position; if a position does not have a previous character change rate, that is, the position has not yet undergone fixation ability training, then the smallest initial character change rate among the positions with historical training records is determined as the initial character change rate of that position. The positions are sorted from smallest to largest based on the initial character change rate to obtain the position sorting result. The initial fixation ability training scheme is to train each position in the position sorting result and the corresponding initial character change rate of that position in sequence.

[0040] During training according to the initial fixation ability training scheme, the first user training data is acquired every preset duration (one training cycle). The first user training data includes the first position, the first character change rate, the first error probability, and the first average delay duration. The fixation ability training scheme is as follows: if the first error probability is less than the preset error probability and the first average delay duration is less than the preset average delay duration, then the next position of the first position in the position sorting result and the corresponding initial character change rate are changed to continue training; if the first error probability is greater than or equal to the preset error probability and / or the first average delay duration is greater than or equal to the preset average delay duration, then training continues at the first position according to the first character change rate until the training at the first position according to the first character change rate is qualified. Then the next position of the first position in the position sorting result and the corresponding initial character change rate are changed to continue training. When the cumulative training time of this fixation ability training reaches the first preset duration, this fixation ability training ends.

[0041] Specifically, generating a user training plan based on historical training records includes: determining the previous character interval, the previous error probability, and the previous completion time based on historical training records; determining a completion time threshold based on the previous character interval; determining the current character interval based on the previous error probability, the previous completion time, and the completion time threshold; and determining a saccadic ability training plan based on the current character interval.

[0042] The saccade training involves the training device determining several positions from various points according to a certain character interval, displaying characters at these positions, and the user quickly saying the characters they see in a preset order (e.g., first from left to right, then from top to bottom, that is, reading one line from left to right, and then reading the next line from top to bottom) while keeping their head still. After reading all the characters, the training device will calculate the user's error probability and completion time.

[0043] The errors that exist in the training process mainly include omission (reading a character less than expected), insertion (reading a character that does not exist), repetition (reading a character repeatedly), and reversal (reversing the positions of adjacent characters).

[0044] In this embodiment, during saccade training, a saccade training plan is generated based on the user's historical training records. Specifically, this includes: retrieving the previous character interval, previous error probability, and previous completion time from the historical training records; different character intervals correspond to different time thresholds; obtaining the time threshold from the database or staff based on the previous character interval; if the previous error probability is less than a preset error probability and the previous completion time is less than the completion time threshold, the previous saccade training is considered successful, and the next character interval in the character interval sequence (pre-set, e.g., intervals of 3, 2, 1, and 0 positions) is determined as the current character interval; if the previous error probability is greater than or equal to a preset error probability and the previous completion time is greater than or equal to the completion time threshold, the previous saccade training is considered unsuccessful, and the previous character interval is determined as the current character interval; the saccade training plan involves displaying characters on the training device screen according to the current character interval.

[0045] Specifically, generating a user training plan based on historical training records includes: determining the error probability and average delay time of each character movement trajectory at various character movement speeds based on historical training records; determining the current character movement speed corresponding to each character movement trajectory based on the error probability and average delay time; determining an initial following ability training plan based on the trajectory difficulty of each character movement trajectory and the corresponding current character movement speed; obtaining the current training count, which is the number of cycles of the character movement trajectory in the current training round; obtaining second user training data based on the current training count; and adjusting the initial following ability training plan based on the second user training data to obtain the following ability training plan.

[0046] The following ability training involves assigning a character movement trajectory (e.g., horizontal, vertical, diagonal, rotational, figure-eight, cloverleaf, random, etc.) and corresponding character movement speed to the training device. The device displays randomly changing characters along the trajectory at the specified speed. The user quickly pronounces the displayed characters while keeping their head still. After one round of training (two consecutive tests using one character movement speed and trajectory constitute one round), the device calculates the user's error probability and average delay. If the error probability is less than the preset error probability and the average delay is less than the preset average delay, the user passes the round and the training continues with the next character movement trajectory. If the error probability is greater than or equal to the preset error probability... If the error probability and / or average latency are greater than or equal to the preset average latency, the training round is considered unsuccessful. Training continues according to the character movement trajectory and speed until a training round using the same character movement trajectory and speed is successful. That is, after each training round, the error probability and average latency are assessed. If the training is successful, the character movement trajectory and / or speed are changed for training. If the training is unsuccessful, training continues according to the current character movement trajectory and speed. It is worth noting that one training round corresponds to one character movement trajectory and one character movement speed. The number of training rounds for one set of following ability training is the first preset number of rounds (e.g., 10 rounds). When the first preset number of rounds is reached, the current following ability training ends.

[0047] The errors mentioned above that exist during the training process mainly include omission (reading a character less than expected), insertion (reading a character that does not exist), and repetition (reading a character repeatedly). The delay time is the interval between the appearance of a character and the user reading the character. The average delay time is the average of all delay times in one training cycle.

[0048] In this embodiment, during the training of the following ability, a training scheme for the current following ability is generated based on the user's historical training records. Specifically, this includes: finding the error probability and average delay time of each character movement trajectory at various character movement speeds from the historical training records; determining the maximum character movement speed corresponding to each character movement trajectory in the following ability training records as the most recent character movement speed; if the error probability of a character movement trajectory at the most recent character movement speed is less than a preset error probability, and the average delay time is less than a preset average delay time, then the next character movement speed after the most recent character movement speed in the character movement speed sequence is determined as the current character movement speed of that character movement trajectory; if the error probability of a character movement trajectory at the most recent character movement speed is greater than or equal to a preset error probability and / or the average delay time is greater than or equal to a preset average delay time, then the most recent character movement speed is determined as the current character movement speed of that character movement trajectory; if a... If there is no training record for following a character movement trajectory, the minimum current character movement speed corresponding to the character movement trajectory with a training record is determined as the current character movement speed of that character movement trajectory (without a training record). The character movement trajectories are sorted in ascending order according to their corresponding current character movement speeds. If the current character movement speeds are the same, they are sorted in ascending order according to their corresponding trajectory difficulty (pre-set). The initial following ability training scheme is to train the character movement trajectories and their corresponding current character movement speeds in the trajectory sorting results. During the training process, the current training count is recorded. In one round of training, the current training count is incremented by 1 when a character movement trajectory is completed. If the current training count reaches 2, one round of training is completed, the current training count becomes zero, and the second user training data is obtained. The second user training data includes the character movement trajectory, the current character movement speed, the error probability, and the average latency.

[0049] The following ability training scheme is as follows: If the error probability of the current round is less than the preset error probability and the average delay time is less than the preset average delay time, then the next character movement trajectory and the corresponding current character movement speed in the trajectory sorting results are changed and training continues; if the error probability of the current round is greater than or equal to the preset error probability and / or the average delay time is greater than or equal to the preset average delay time, then training continues according to the character movement trajectory and the corresponding current character movement speed until one round of training according to the character movement trajectory and the character movement speed is qualified, then the next character movement trajectory and the corresponding current character movement speed in the trajectory sorting results are changed and training continues. When the number of training rounds of this following ability training reaches the first preset number of rounds, the following ability training ends.

[0050] Step S104: If no historical training record exists, determine the associated user based on the user information.

[0051] If the current user information does not exist in the database corresponding to the historical user information, it means that the user has not participated in training in the past and there is no historical training record in the database. In this case, the associated user of the user information is determined.

[0052] Specifically, determining associated users based on user information includes: determining user age, user gender, user eye abnormality information, and user cognitive ability information based on user information; calculating user similarity based on the K-nearest neighbor algorithm, user age, user gender, user eye abnormality information, user cognitive ability information, and historical user information; and identifying historical users whose user similarity is higher than a preset similarity as associated users.

[0053] In this embodiment, user age, gender, eye abnormality information, and cognitive ability information are retrieved from user information. The similarity of user age, gender, eye abnormality information, and cognitive ability information in user information and historical user information is calculated using the K-nearest neighbor algorithm to obtain the user similarity between the current user and each historical user. Historical users whose user similarity is higher than the preset similarity are identified as associated users.

[0054] Step S105: Obtain the associated training records of the first training of the associated user.

[0055] Retrieve the training records of the first training of the associated user from the database, namely the associated training records. The associated training records include the associated gaze ability training records, the associated saccade ability training records, and the associated following ability training records.

[0056] Step S106: Generate user training scheme based on associated training records.

[0057] Specifically, generating a user training plan based on associated training records includes: determining the gaze character change rate based on the error probability and average delay duration in the associated gaze ability training records; determining the saccade character interval based on the error probability and completion duration in the associated saccade ability training records; determining the following character movement trajectory and corresponding following character movement speed based on the error probability and average delay duration in the associated tracking ability training records; and determining the user training plan based on the gaze character change rate, saccade character interval, following character movement trajectory, and corresponding following character movement speed.

[0058] In this embodiment, if the error probability of all positions in an associated fixation ability training record (in the first training, there is only one character change rate) is less than the preset error probability, and the average delay time of all positions is less than the preset average delay time, then the character change rate in the associated fixation ability training record is qualified. The qualified probability of various character change rates in all associated fixation ability training records is counted, and the character change rate with the highest qualified probability exceeding the first preset qualified probability (e.g., 80%) is determined as the fixation character change rate. If there is no character change rate with a qualified probability exceeding the first preset qualified probability, then the smallest character change rate in the character change rate sequence is determined as the fixation character change rate.

[0059] If the error probability in a training record of associated saccade ability is less than the preset error probability, and the completion time is less than the time threshold corresponding to the character interval in the training record of associated saccade ability, then the character interval in the training record of associated saccade ability is qualified. The qualified probability of various character intervals in all training records of associated saccade ability is counted, and the character interval with the smallest qualified probability exceeding the second preset qualified probability (e.g., 90%) is determined as the saccade character interval; if there is no character interval with a qualified probability exceeding the second preset qualified probability, then the largest character interval in the character interval sequence is determined as the saccade character interval.

[0060] If the error probability of a trajectory speed combination in a training record of the association following ability (containing at least one trajectory speed combination) is less than a preset error probability, and the average delay time of such trajectory speed combination is less than a preset average delay time, then such trajectory speed combination in the training record of the association following ability is qualified. The qualified probability of various trajectory speed combinations in all training records of the association following ability is calculated. One trajectory speed combination corresponds to one character movement trajectory and one character movement speed. The trajectory speed combination with a qualified probability exceeding a third preset qualified probability (e.g., 70%) is determined as a candidate combination. The character movement speed with the largest value in the candidate combination is determined as the following character movement speed. The character movement trajectory with the highest trajectory difficulty in the candidate combination of the following character movement speed is determined as the following character movement trajectory. If there is no candidate combination, the smallest character movement speed in the character movement speed sequence is determined as the following character movement speed. The character movement trajectory with the lowest trajectory difficulty is determined as the following character movement trajectory.

[0061] The user training scheme is determined based on the speed of character gaze, the interval of character scanning, the trajectory of character tracking, and the corresponding speed of character tracking.

[0062] Specifically, the user training scheme is determined based on the gaze character change rate, saccade character interval, following character movement trajectory, and corresponding following character movement speed. This includes: determining the gaze ability training scheme based on the gaze character change rate and position selection algorithm; determining the saccade ability training scheme based on the saccade character interval; determining the following ability training scheme based on the following character movement trajectory and corresponding following character movement speed; and determining the user training scheme based on the saccade ability training scheme, the gaze ability training scheme, and the following ability training scheme.

[0063] In this embodiment, the first quantity = first preset duration / preset duration, where the first preset duration is the total duration of gaze ability training. The position is changed every preset duration. The first quantity of positions is selected by a position selection algorithm (each position can be represented by a numerical value and the position is selected by a random number algorithm). The gaze ability training scheme is to perform gaze ability training for a preset duration at each selected position according to the order of the positions, and record the training data. It is worth noting that after the training at each position is completed, there is no need to make a judgment and immediately change to the next position to continue training.

[0064] The saccadic recognition training scheme involves displaying characters on the screen at saccadic character intervals and recording the user's training data.

[0065] The training trajectories are defined as those that follow the movement of characters and those that are more difficult than those that follow the movement of characters. The training scheme is to select the training trajectories in order of difficulty from easy to hard, while following the movement of characters at different speeds. Each training trajectory is performed twice in succession until the training of all training trajectories is completed. It is worth noting that after each training trajectory has been performed twice in succession, no judgment is required, and the next training trajectory is immediately changed to continue the training.

[0066] The training programs for saccade awareness, fixation, and following ability were jointly determined as the user training program.

[0067] Figure 2 This is a structural block diagram of an artificial intelligence-based eye movement training device 200 provided in an embodiment of this application.

[0068] like Figure 2 As shown, the AI-based eye movement training device 200 mainly includes: User information acquisition module 201 is used to acquire user information; The historical record judgment module 202 is used to determine whether there are historical training records based on user information. The first scheme generation module 203 is used to generate a user training scheme based on the historical training records if historical training records exist. The associated user determination module 204 is used to determine the associated user based on user information if no historical training record exists. The associated record acquisition module 205 is used to acquire the associated training records of the associated user's first training session; The second scheme generation module 206 is used to generate user training schemes based on associated training records.

[0069] As an optional implementation of this embodiment, the user training scheme includes a gaze ability training scheme. The first scheme generation module 203 is specifically used to generate a user training scheme based on historical training records, including: determining the last character change rate, the last error probability, and the last average delay duration for each position based on historical training records; determining the initial character change rate for each position based on the last error probability, the last average delay duration, and the last character change rate; sorting each position from smallest to largest based on the initial character change rate to obtain a position sorting result; generating an initial gaze ability training scheme based on the position sorting result and the initial character change rate corresponding to each position; acquiring first user training data based on a preset duration; and adjusting the initial gaze ability training scheme based on the first user training data to obtain the gaze ability training scheme.

[0070] As an optional implementation of this embodiment, the user training scheme includes a saccade ability training scheme. The first scheme generation module 203 is specifically used to generate a user training scheme based on historical training records, including: determining the previous character interval, the previous error probability, and the previous completion time based on historical training records; determining a completion time threshold based on the previous character interval; determining the current character interval based on the previous error probability, the previous completion time, and the completion time threshold; and determining the saccade ability training scheme based on the current character interval.

[0071] As an optional implementation of this embodiment, the user training scheme includes a following ability training scheme. The first scheme generation module 203 is specifically used to generate a user training scheme based on historical training records, including: determining the error probability and average delay time of each character movement trajectory at various character movement speeds based on historical training records; determining the current character movement speed corresponding to each character movement trajectory based on the error probability and average delay time; determining an initial following ability training scheme based on the trajectory difficulty of each character movement trajectory and the corresponding current character movement speed; obtaining the current training count, which is the number of cycles of the character movement trajectory in the current training round; obtaining second user training data based on the current training count; and adjusting the initial following ability training scheme based on the second user training data to obtain the following ability training scheme.

[0072] As an optional implementation of this embodiment, the associated user determination module 204 is specifically used to determine associated users based on user information, including: determining user age, user gender, user eye abnormality information, and user cognitive ability information based on user information; calculating user similarity based on the K-nearest neighbor algorithm, user age, user gender, user eye abnormality information, user cognitive ability information, and historical user information; and determining historical users corresponding to historical user information whose user similarity is higher than a preset similarity as associated users.

[0073] As an optional implementation of this embodiment, the associated training records include associated fixation ability training records, associated saccade ability training records, and associated tracking ability training records. The second scheme generation module 206 is specifically used to generate a user training scheme based on the associated training records, including: determining the fixation character change speed based on the error probability and average delay duration in the associated fixation ability training records; determining the saccade character interval based on the error probability and completion duration in the associated saccade ability training records; determining the tracking character movement trajectory and corresponding tracking character movement speed based on the error probability and average delay duration in the associated tracking ability training records; and determining the user training scheme based on the fixation character change speed, saccade character interval, tracking character movement trajectory, and corresponding tracking character movement speed.

[0074] As an optional implementation of this embodiment, the second scheme generation module 206 is specifically used to determine a user training scheme based on the gaze character change speed, saccade character interval, following character movement trajectory, and corresponding following character movement speed, including: determining a gaze ability training scheme based on the gaze character change speed and position selection algorithm; determining a saccade ability training scheme based on the saccade character interval; determining a following ability training scheme based on the following character movement trajectory and corresponding following character movement speed; and determining a user training scheme based on the saccade ability training scheme, the gaze ability training scheme, and the following ability training scheme.

[0075] In one example, the module in any of the above devices may be one or more integrated circuits configured to implement the above methods, such as one or more application-specific integrated circuits (ASICs), or one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.

[0076] For example, when modules in a device can be implemented via a processing element scheduler, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor capable of calling programs. Alternatively, these modules can be integrated together as a system-on-a-chip (SOC).

[0077] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device and module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0078] Figure 3 This is a structural block diagram of the training device 300 provided in an embodiment of this application.

[0079] like Figure 3 As shown, the training device includes a main body and a printing device for printing training data. The printing device is communicatively connected to the main body. The main body includes a processor, a memory, a screen, a microphone, and a power switch (not shown in the figure). The processor, memory, screen, microphone, and printing device are all electrically connected to the power switch. The memory, screen, microphone, and printing device are all communicatively connected to the processor. The microphone, screen, and printing device are all communicatively connected to the memory.

[0080] The processor controls the overall operation of the training device 300 to complete all or part of the steps of the aforementioned AI-based eye movement training method. The memory stores various types of data to support the operation of the training device 300. This data may include, for example, instructions for any application or method operating on the training device 300, and application-related data. The memory can be implemented using any type of volatile or non-volatile storage device or a combination thereof, such as one or more of Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0081] The screen can be any display screen used to display data and has touch screen control function. The microphone is a sound-receiving device with sound-receiving function and adjustable length and direction, used to collect the user's voice data, and can be adjusted in length and direction according to the user's position. The power switch controls the opening and closing of the entire training device. The printing device is used to print the training data of the training device. A stylus can also be detachably installed on the training device at any position for convenient control of the screen.

[0082] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described artificial intelligence-based eye movement training method.

[0083] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0084] The terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0085] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing application concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions claimed in this application.

Claims

1. An artificial intelligence-based eye movement training method, characterized in that, include: Obtain user information; Based on the user information, determine whether there are historical training records; If the historical training records exist, a user training plan is generated based on the historical training records; If the historical training record does not exist, the associated user is determined based on the user information; Obtain the associated training records of the first training session of the associated user; The user training scheme is generated based on the associated training records.

2. The method according to claim 1, characterized in that, The user training scheme includes a gaze ability training scheme, and the generation of the user training scheme based on the historical training records includes: Based on the historical training records, determine the last character change rate, the last error probability, and the last average delay time for each position; The initial character change rate at each position is determined based on the previous error probability, the previous average delay duration, and the previous character change rate. The positions are sorted from smallest to largest based on the initial character change rate to obtain the position sorting result; An initial gaze ability training scheme is generated based on the position sorting results and the initial character change rate corresponding to each position. Acquire the first user's training data based on a preset duration; The initial gaze ability training scheme is adjusted based on the first user training data to obtain the gaze ability training scheme.

3. The method according to claim 1, characterized in that, The user training scheme includes a saccade ability training scheme, and the step of generating the user training scheme based on the historical training records includes: Based on the historical training records, determine the last character interval, the last error probability, and the last completion time; The completion time threshold is determined based on the previous character interval; The current character interval is determined based on the previous error probability, the previous completion time, and the completion time threshold. The saccadic perception training scheme is determined based on the current character interval.

4. The method according to claim 1, characterized in that, The user training scheme includes a following ability training scheme, and the step of generating the user training scheme based on the historical training records includes: Based on the historical training records, determine the error probability and average delay time of each character movement trajectory at various character movement speeds; The current character movement speed corresponding to each character movement trajectory is determined based on the error probability and the average delay duration. The initial following ability training scheme is determined based on the trajectory difficulty of each character movement trajectory and the corresponding character movement speed. Obtain the current training iteration, which is the number of times the character movement trajectory has been looped in the current training round; The training data of the second user is obtained based on the current number of training iterations; The initial following ability training scheme is adjusted based on the second user training data to obtain the following ability training scheme.

5. The method according to claim 1, characterized in that, The step of determining associated users based on the user information includes: Based on the user information, determine the user's age, gender, eye abnormalities, and cognitive abilities. User similarity is calculated based on the K-nearest neighbor algorithm, the user's age, gender, eye abnormality information, cognitive ability information, and historical user information. Historical users whose similarity to the user is higher than a preset similarity are identified as associated users.

6. The method according to claim 1, characterized in that, The associated training records include associated fixation ability training records, associated saccade ability training records, and associated following ability training records. Generating the user training plan based on the associated training records includes: The rate of change of the gazed character is determined based on the error probability and average delay duration in the associated gaze ability training record. The interval between scanned characters is determined based on the error probability and completion time in the associated scan ability training records. The movement trajectory of the following character and the corresponding movement speed of the following character are determined based on the error probability and average delay time in the training record of the associated following ability. The user training scheme is determined based on the speed of change of the gaze character, the interval of the saccade character, the trajectory of the following character, and the corresponding speed of the following character.

7. The method according to claim 6, characterized in that, The process of determining the user training scheme based on the gaze character change rate, the saccade character interval, the following character movement trajectory, and the corresponding following character movement speed includes: The gaze ability training scheme is determined based on the gaze character change speed and position selection algorithm. A saccharification ability training scheme is determined based on the saccharification character interval; A training scheme for the following ability is determined based on the movement trajectory of the following character and the corresponding movement speed of the following character. The user training scheme is determined based on the saccadic ability training scheme, the fixation ability training scheme, and the following ability training scheme.

8. An eye movement training device based on artificial intelligence, characterized in that, include: The user information acquisition module is used to acquire user information; The historical record determination module is used to determine whether there are historical training records based on the user information; The first scheme generation module is used to generate a user training scheme based on the historical training record if the historical training record exists. The associated user determination module is used to determine the associated user based on the user information if the historical training record does not exist. The associated record acquisition module is used to acquire the associated training records of the first training of the associated user; The second scheme generation module is used to generate the user training scheme based on the associated training records.

9. A training device, characterized in that, The device includes a training instrument body and a printing device for printing the training instrument data. The training instrument body includes a processor, a memory, a screen, a microphone, and a power switch. The processor is used to execute a computer program stored in the memory to cause the training instrument to perform the method as described in any one of claims 1 to 7. The processor, the memory, the screen, the microphone, and the printing device are all electrically connected to the power switch. The memory, the screen, the microphone, and the printing device are all communicatively connected to the processor. The microphone, the screen, and the printing device are all communicatively connected to the memory.

10. A computer-readable storage medium, characterized in that, It includes a computer program or instructions that, when run on a computer, cause the computer to perform the method as described in any one of claims 1 to 7.

Images