An interactive vision training rehabilitation system based on virtual reality (VR)

Abstract

The application belongs to the technical field of vision training, and particularly relates to an interactive vision training rehabilitation system based on virtual reality (VR), which comprises: a progressive depth of field training module, which drives virtual focal length dynamic adjustment to induce periodic accommodation of ciliary muscles by a nonlinear progressive depth of field change function; an accommodation convergence collaborative training module, which generates convergence accommodation coordination scores by training convergence reserves through virtual prism deflection; a psychological fatigue monitoring and intervention module, which implements three-level grading differentiated intervention and pushes psychological treatment content; an active relaxation recovery module, which adaptively adjusts training relaxation time length proportion according to fatigue accumulation rate to induce active relaxation of ciliary muscles; a multi-modal interactive feedback module, which synchronously outputs differentiated tactile vibration pulses and corrects parallax parameters in real time according to head posture; and a closed-loop adaptive optimization module, which drives progressive adjustment of training parameters based on dynamic weight evaluation of four-dimensional training indexes and maintains a time-weighted user benchmark profile.

Description

Technical Field

[0001] This invention belongs to the technical field of vision training, specifically an interactive vision training and rehabilitation system based on virtual reality (VR). Background Technology

[0002] Vision training and rehabilitation technology is an important research direction in the field of optometry. With the increasing prevalence of visual impairments such as myopia, amblyopia, accommodative dysfunction, and accommodative fatigue, digital vision rehabilitation training systems based on virtual reality technology have gradually become a research hotspot. Virtual reality technology has inherent advantages such as precise control of visual stimulus parameters, provision of immersive training experiences, and real-time data acquisition, providing a technological foundation for building a systematic and personalized vision rehabilitation training platform.

[0003] However, existing technologies are mainly limited by the lack of personalized adaptive mechanisms in depth-of-field variation design and the dual architectural bottleneck of separation between accommodation training and ensemble training when implementing systematic vision rehabilitation training. This makes it difficult to balance the safety of users with weaker accommodation abilities with the continuous training benefits of users with stronger accommodation abilities in practical applications. It is also difficult to balance the synchronous matching of accommodation training stimuli and ensemble training stimuli and the personalized adaptation under the different AC / A values ​​of different individuals. This affects the effect of accommodation-ensemble collaborative training and may aggravate visual discomfort symptoms. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an interactive vision training and rehabilitation system based on virtual reality (VR) to solve the problems in existing technologies that make it difficult to balance the safety of users with weaker accommodative abilities with the continuous training benefits of users with stronger accommodative abilities, as well as the difficulty in synchronizing the matching of accommodative training stimuli and ensemble training stimuli and the personalized adaptation under the differences in AC / A values ​​among different individuals.

[0005] An interactive vision training and rehabilitation system based on virtual reality (VR) includes: a progressive depth-of-field training module, which is used to construct a multi-layer depth target distribution structure covering the refractive power range of 0.25D to 4D in a virtual three-dimensional space, and generate a personalized progressive nonlinear depth-of-field change function with a focus quality index as feedback signal based on the user's initial accommodation ability assessment results, driving the dynamic adjustment of virtual focal length parameters to induce the ciliary muscle to produce periodic contraction and relaxation;

[0006] The accommodation-accommodation coordination training module is used to automatically estimate the individual accommodation-accommodation ratio (AC / A) of a user through a standardized measurement task. Based on the AC / A value, a dual-channel synchronous control mechanism for accommodation and convergence channels is constructed. The focal length parameter adjustment and binocular disparity parameter adjustment are output synchronously according to the individual AC / A value. The module also simulates the application of additional convergence load through virtual prism deflection to train convergence reserve capacity. At the same time, the module calculates the accommodation lead-lag value between the user's actual accommodation response point and the target focal length, and generates a convergence accommodation coordination score.

[0007] The psychological fatigue monitoring and intervention module is used to collect six-dimensional fatigue-sensitive indicators, including pupillary oscillation amplitude, gaze stability decline rate, task response delay growth rate, heart rate variability low-to-high frequency ratio, blink rate change rate, and subjective fatigue score. It constructs a psychological fatigue state vector, generates a comprehensive psychological fatigue score through weighted fusion, classifies fatigue state into three levels: mild, moderate, and severe, and triggers differentiated intervention strategies. In the severe fatigue state, psychological therapy content is pushed out.

[0008] The active relaxation and recovery module is used to adaptively adjust the duration ratio of the training segment and the relaxation segment according to the fatigue accumulation rate. In the relaxation segment, a low spatial frequency, low contrast open natural virtual scene with an optical equivalent distance of not less than 6 meters is constructed. The ciliary muscle is induced to move to the maximum relaxation state through a progressive visual guidance trajectory from near to far. At the same time, low frequency vibration tactile feedback and environmental sound scene are output simultaneously to form a multi-sensory collaborative relaxation environment.

[0009] The multimodal interactive feedback module is used to synchronously output differentiated tactile vibration pulses that are consistent with the depth-of-field switching rhythm when the virtual target focal length switches. It also integrates heart rate, skin conductivity and head posture data collected by the built-in sensors of the head display and the wristband device, and corrects the binocular parallax parameters in real time according to the head movement state to maintain a stable ensemble stimulation.

[0010] The closed-loop adaptive optimization module is used to collect four core training indicators: average focus response time, adjustment amplitude, adjustment accuracy score, and ensemble adjustment coordination score. Based on the current gap in the achievement of each dimension, the weights are dynamically normalized and calibrated to construct a comprehensive adjustment capability assessment score. The assessment results drive the progressive adaptive adjustment of training parameters and dynamically update the user's adjustment capability benchmark file in an incremental manner with time-sensitive weights.

[0011] Preferably, the progressive depth-of-field training module is as follows:

[0012] A multi-layered depth target distribution structure is established in a 3D virtual space using depth-of-field hierarchical building units, and the set of depth-of-field refractive power values ​​for each layer is defined. The refractive power values ​​at each level are as follows: to Uniformly distributed within the range;

[0013] The depth-of-field variation function generation unit generates the depth of field based on the user's initial adjustment range. Constructing a progressive depth-of-field variation amplitude function ,in, Indicates the cumulative training unit number. Represents the individual adaptation coefficient, and simultaneously constructs the depth-of-field change rate modulation function. The two are combined to construct a complete progressive depth-of-field variation function. ;

[0014] The depth-of-field variation function is mapped to the virtual target focal length parameter through the dynamic focal length adjustment unit. ;

[0015] By setting task events on target objects at each depth level through active interaction task units, which require user gaze interaction to trigger, a task trigger condition function is constructed: when... hour, ,on the contrary, ,in, Indicates user The coordinates of the gaze point at any given moment. express The spatial coordinates of the target object at any given time. This represents the gaze hit radius threshold. Indicates the duration of sustained gaze. This indicates the minimum duration of sustained gaze required to trigger the task. express The task trigger status at any given moment.

[0016] Preferably, the adjustment set collaborative training module is specifically as follows:

[0017] During the training preparation phase, a standardized measurement task is set up using the AC / A value estimation unit, presenting distances sequentially. and The virtual gaze target is recorded, and the corresponding binocular convergence angle response value is recorded. and According to the formula Estimate the individual user's adjustment set ratio and store the estimation results in the individual user parameter file;

[0018] Using the focal length parameter via a dual-channel synchronous control unit To adjust the channel drive signal, the target set angle requirement value is calculated synchronously. Mapping the target set angle to binocular disparity adjustment Where IPD is the user's interpupillary distance. This represents the base convergence angle when the user is in a distant gaze state, and outputs the focus adjustment and parallax adjustment to the virtual reality rendering engine simultaneously;

[0019] A virtual prism deflection parameter sequence is constructed using ensemble flexibility training units. Equal-sized, opposite-direction lateral virtual prisms were applied to the left and right eye images, forcing both eyes to complete the binocular fusion maintenance task under increasing convergence load. The fusion recovery time at each prism level was recorded. ;

[0020] By adjusting the lead-lag detection unit, the user's real-time gaze depth estimation value is collected. Calculate the adjustment lead-lag value And construct a set of coordination scores. .

[0021] Preferably, the psychological fatigue monitoring and intervention module is as follows:

[0022] Six fatigue-sensitive indicators were collected using a multi-dimensional indicator acquisition unit: pupil oscillation amplitude. ;

[0023] gaze stability decline rate ;

[0024] Task response latency growth rate ;

[0025] Heart rate variability low-high frequency ratio ;

[0026] Blink rate change rate ;

[0027] Subjective fatigue rating ;

[0028] Constructing a vector of mental fatigue states, we have:

[0029] ;

[0030] A comprehensive psychological fatigue score is constructed by weighting and integrating the six-dimensional indicators through a fatigue grading assessment unit. ,in accordance with With threshold , The relationship is divided into three fatigue levels: mild, moderate, and severe.

[0031] The corresponding intervention is triggered by the differentiated intervention execution unit: when the fatigue is mild, insert a 30-second relaxation scene and reduce the current training difficulty by one level; when the fatigue is moderate, pause the training and switch to the 5-minute active relaxation and recovery module; after the relaxation ends, downgrade and restart the training; when the fatigue is severe, forcibly terminate the current training unit and record the fatigue accumulation rate.

[0032] The psychotherapy content push unit adaptively selects and pushes content from a psychotherapy content library that includes mindfulness meditation guidance, visual art healing scenarios, and breathing rhythm guidance when the user is in a state of severe fatigue. The push duration is no less than 8 minutes. Training can only be restarted after the user has finished the content and the user has recovered to a level of mild or below through a subjective fatigue score.

[0033] Preferably, the active relaxation and recovery module is as follows:

[0034] The fatigue accumulation rate is calculated using the training relaxation cycle management unit. The duration of the next training unit is adaptively adjusted based on the rate of fatigue accumulation. With relaxation time ,in, and These represent the baseline training duration and the baseline relaxation duration, respectively. and Indicates the adjustment factor;

[0035] An open, natural virtual scene with an optical equivalent distance of no less than 6 meters is constructed using a distant view scene generation unit, with scene parameters set to the upper limit of spatial frequency. Weekly, image contrast Color temperature range And overlay progressive visual guidance trajectories on the scene. Inducing the user's gaze point to extend from near to far to promote active relaxation of the ciliary muscle;

[0036] The synchronous output frequency range of the multi-sensory synergistic relaxation unit is [missing information]. to The low-frequency vibration tactile feedback has a vibration intensity that decreases linearly with relaxation time. And simultaneously play content with frequency of to of The band-based binaural beat-based natural environmental soundscape assists the autonomic nervous system in switching to a parasympathetic dominant state.

[0037] Preferably, the multimodal interactive feedback module is as follows:

[0038] The visual-tactile synchronization unit outputs differentiated tactile vibration pulses that match the depth-of-field switching rhythm during focus switching: when the focus switching direction is from far to near... When reduced, the output frequency range to Single pulse duration A high-frequency short-pulse sequence, when the focus switching direction is from near to far, i.e. When increased, the output frequency range to Single pulse duration The low-frequency long vibration pulse, the synchronization delay between the output time of the tactile vibration pulse and the switching time of the focal length parameter does not exceed Establish a direction-specific conditional connection between visual accommodation switching signals and tactile perception signals;

[0039] A multi-source physiological data synchronous acquisition vector is constructed using a physiological data fusion unit. ,in, express Heart rate at all times express skin conductivity at any time express Head pose angle vector at any given time. express Multi-source physiological data vectors at time points, and The output is sent to the mental fatigue monitoring and intervention module for LHR index calculation and auxiliary verification;

[0040] The head posture offset is calculated using a real-time parameter correction unit. Compensate for the parallax adjustment amount based on the offset. ,in, This is the head motion parallax compensation coefficient, ensuring that the ensemble stimulus remains stable while the user's head is moving.

[0041] Preferably, the closed-loop adaptive optimization module is as follows:

[0042] By summarizing the core training metrics after each training unit through a multi-dimensional data acquisition unit, a four-dimensional training unit evaluation dataset is constructed. ,in, Indicates the first Average focus response time per training unit Indicates the first The average adjustment amplitude estimated per training unit Indicates the first The tuning accuracy score for each training unit. Indicates the first The ensemble coordination score of each training unit. Indicates the first A four-dimensional evaluation dataset for each training unit;

[0043] The achievement gap for each dimension is calculated using a dynamic weighting evaluation unit. Perform normalization and dynamic calibration on the weights. The weights of dimensions that have already met the standards are automatically reduced to zero, and the weights are then concentrated on the dimensions with the largest gap in compliance, thus constructing a comprehensive adjustment capability assessment score. ;

[0044] The unit adaptively adjusts training parameters based on comprehensive evaluation results. The depth of field variation range, depth of field variation rate, virtual prism deflection step size, and training-relaxation cycle ratio were progressively adjusted. This indicates the step size coefficient for parameter adjustment, and its value range is... ;

[0045] Define the timeliness weight function through the user baseline profile maintenance unit. The baseline values ​​for each dimension are updated using a weighted average method with time-sensitive weights. And for training sessions that are more than 90 days old and whose score deviation exceeds the baseline value. Perform a clear operation on historical data points;

[0046] After each comprehensive evaluation, the system's global control parameter set is updated based on the comprehensive evaluation score. ,in, Represents the set of global control parameters of the system. This indicates the feedback adjustment step size coefficient. This represents the gradient vector of the overall evaluation score relative to the changes in each control parameter. The updated set of control parameters and the user baseline profile are then re-entered into each functional module for the initialization of the next training unit.

[0047] Preferably, the progressive depth-of-field training module generates a progressive depth-of-field change function through a depth-of-field change function generation unit, as follows:

[0048] Based on the user's initial adjustment range Construct a progressive depth-of-field variation amplitude function based on basic parameters To achieve a gradual increase in difficulty with a logarithmic growth pattern, the individual adaptation coefficient... The closed-loop adaptive optimization module updates the data in real time based on the current adjustment capability assessment results.

[0049] Constructing a depth-of-field rate modulation function ,in, Indicates the first The base rate of change for each training unit Indicates the first The rhythmic cycle of each training unit This represents the local time within a training unit, causing the depth-of-field change rate to exhibit a non-linear sinusoidal modulation rhythm that is slow at first, then fast, and then slows down again within each training unit.

[0050] By combining the depth-of-field variation amplitude function and the rate modulation function, a complete progressive depth-of-field variation function is constructed. , This is the initial depth value for the current training unit;

[0051] The depth-of-field variation function is mapped to the virtual target focal length parameter through the dynamic focal length adjustment unit. It drives the focal length rendering engine of the head-mounted virtual reality device to adjust the virtual focal length of the target object in real time.

[0052] Preferably, the psychological fatigue monitoring and intervention module achieves closed-loop intervention through a psychological fatigue state vector and a three-level fatigue grading assessment, as detailed below:

[0053] Six-dimensional fatigue-sensitive indicators were collected through a multi-dimensional indicator collection unit, and a psychological fatigue state vector was constructed.

[0054] A comprehensive psychological fatigue score is constructed using fatigue grading assessment units, based on... With threshold , If the relationship is divided into three levels, then:

[0055] when At that time, it was determined to be a state of mild fatigue. At that time, it was determined to be a state of moderate fatigue. At that time, it was determined to be a state of severe fatigue;

[0056] The differentiated intervention execution unit triggers corresponding intensity intervention responses based on the three levels of fatigue. The psychotherapy content push unit adaptively selects push content from the psychotherapy content library, which includes three types of content: mindfulness meditation guidance, visual art healing scenarios, and breathing rhythm guidance, based on the user's historical intervention response records when the user is in a state of severe fatigue.

[0057] The comprehensive psychological fatigue score and fatigue accumulation rate are continuously output to the active relaxation and recovery module and the closed-loop adaptive optimization module for adaptive adjustment of the training relaxation cycle and dynamic updating of training parameters.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] 1. This invention achieves the effect of continuously and accurately matching the intensity of training stimuli with the dynamic evolution of the user's adjustment ability, and ensuring that each training unit is always at the boundary of the user's current adjustment ability by constructing a progressive nonlinear depth-of-field change function by combining a logarithmically growing progressive amplitude function and a sinusoidal rhythmic rate modulation function, and updating the individual adaptation coefficient in real time with the focus quality index as a feedback signal.

[0060] 2. This invention employs a dual-channel synchronous control mechanism based on individual AC / A value estimation to construct the accommodation and convergence channels. It coordinates the output of focal length parameter adjustment and binocular disparity adjustment according to individual AC / A values. Furthermore, it uses a virtual prism deflection parameter sequence to apply convergence load step by step while simultaneously calculating accommodation lead-lag values ​​to construct a convergence-accommodation coordination score. This technology achieves synchronous and synergistic enhancement of accommodation and convergence functions in accordance with individual neural coupling patterns, targeted improvement of convergence reserve capacity, and quantitative evaluation of accommodation-accommodation coordination consistency.

[0061] 3. This invention constructs a psychological fatigue state vector comprising six dimensions: pupillary oscillation amplitude, gaze stability decline rate, task response delay growth rate, heart rate variability low-to-high frequency ratio, blink rate change rate, and subjective fatigue score. Through weighted fusion, a comprehensive psychological fatigue score is generated, classifying fatigue states into three levels and triggering differentiated intervention strategies. In severe fatigue states, a psychological therapy content library including guided mindfulness meditation, visual art therapy scenarios, and breathing rhythm guidance is activated. This technology achieves highly sensitive real-time monitoring of both physiological and psychological fatigue states during training, precise graded responses to different fatigue levels, and a systematic intervention effect on emotional stability in severe fatigue states.

[0062] 4. This invention adaptively adjusts the ratio of training to relaxation time based on the fatigue accumulation rate using an exponential and linear decay function. During the relaxation phase, it constructs an open, natural virtual scene with an upper limit to spatial frequency and overlays a progressive visual guidance trajectory to induce the gaze point to extend from near to far. Simultaneously, it outputs low-frequency tactile feedback of 8Hz to 15Hz and 2Hz to 4Hz... The technology of binaural beat-based ambient soundscapes achieves precise matching of relaxation duration and fatigue level, active movement of the ciliary muscle towards maximum relaxation state under low visual cortical load, and multi-sensory synergy assisting the autonomic nervous system in switching to parasympathetic dominance.

[0063] 5. This invention achieves the following effects: multi-dimensional and accurate evaluation of training results, automatic concentration of weight resources towards the weakest adjustment function dimension, and continuous reflection of the current true adjustment capability status in the user's benchmark profile by adopting a four-dimensional evaluation dataset based on average focus response time, adjustment amplitude, adjustment accuracy, and ensemble adjustment coordination; using a normalized dynamic weight calibration mechanism to drive progressive adaptive adjustment of training parameters; incremental updates of user benchmark profiles using a time-sensitive weight function; and periodic removal of expired data older than 90 days and with a deviation exceeding 30%. Attached Figure Description

[0064] Figure 1 This is a schematic diagram of the overall structure of an interactive vision training and rehabilitation system based on virtual reality (VR) according to the present invention. Detailed Implementation

[0065] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.

[0066] Example 1

[0067] Reference Figure 1 As an embodiment of the present invention, an interactive vision training and rehabilitation system based on virtual reality (VR) is provided, including: a progressive depth-of-field training module, an accommodation set collaborative training module, a psychological fatigue monitoring and intervention module, an active relaxation and recovery module, a multimodal interactive feedback module, and a closed-loop adaptive optimization module.

[0068] Furthermore, the progressive depth-of-field training module is used to construct interactive target objects with multi-layered depth distribution in a virtual environment and generate progressive nonlinear depth-of-field variation functions based on individual accommodation capabilities to drive dynamic adjustment of virtual focal length parameters. It also includes: a depth-of-field hierarchy construction unit for establishing a multi-layered depth target distribution structure covering a refractive power range of 0.25D to 4D in three-dimensional virtual space; a depth-of-field variation function generation unit for generating personalized progressive depth-of-field variation functions based on the user's initial accommodation capability assessment results; a dynamic focal length adjustment unit for updating the focal length parameters of the virtual target in real time according to the depth-of-field variation function; and an active interaction task unit for combining visual accommodation stimuli with user-initiated control tasks to achieve interactive accommodation training.

[0069] The accommodation set coordination training module is used to construct a dual-channel synchronous training mechanism for accommodation and accommodation based on the individual user's accommodation set ratio. It also includes: an AC / A value estimation unit, used to automatically estimate the individual user's accommodation set ratio through a standardized measurement task; a dual-channel synchronous control unit, used to synchronously adjust the focal length and binocular disparity parameters in the virtual scene according to the AC / A value; an accommodation flexibility training unit, used to simulate the application of additional accommodation load through virtual prism deflection to train accommodation reserve capacity; and an accommodation lead-lag detection unit, used to calculate the deviation between the user's actual accommodation response point and the target focal length, and generate an accommodation coordination score.

[0070] The psychological fatigue monitoring and intervention module is used to collect multidimensional fatigue-sensitive indicators to construct a psychological fatigue state vector and trigger differentiated intervention strategies based on fatigue levels. It also includes: a multidimensional indicator collection unit for collecting six-dimensional indicators: pupillary oscillation amplitude, gaze stability decline rate, task response delay growth rate, heart rate variability low-to-high frequency ratio, blink rate change rate, and subjective fatigue score; a fatigue grading assessment unit for weighted fusion of the six-dimensional indicators and classifying fatigue states into mild, moderate, and severe levels; a differentiated intervention execution unit for triggering intervention responses of corresponding intensity based on fatigue levels; and a psychotherapy content push unit for pushing mindfulness meditation guidance, visual art therapy scenarios, and breathing rhythm guidance content to users.

[0071] The active relaxation and recovery module is used to implement alternating cycle management based on fatigue monitoring data between training and relaxation phases. During the relaxation phase, it induces active relaxation of the ciliary muscle through a multi-sensory collaborative environment. It also includes: a distant viewing scene generation unit, used to construct an open natural virtual scene with an optical equivalent distance of no less than 6 meters, and the scene adopts a low spatial frequency, low contrast and soft color temperature design; a multi-sensory collaborative relaxation unit, used to simultaneously output low-frequency vibration tactile feedback and environmental soundscape during the relaxation phase to form a multi-sensory collaborative relaxation environment; and a training-relaxation cycle management unit, used to adaptively adjust the duration ratio of training and relaxation phases according to the user's fatigue accumulation rate.

[0072] The multimodal interactive feedback module is used to construct a tactile feedback mechanism synchronized with the visual training task and to collect physiological data from the wearable device for fatigue monitoring and parameter correction. It also includes: a visual-tactile synchronization unit, which outputs tactile vibration pulses consistent with the depth-of-field switching rhythm when the virtual target focal length switches; a physiological data fusion unit, which integrates heart rate, skin conductivity, and head posture data collected by the built-in sensors of the head-mounted display and the wristband device; and a real-time parameter correction unit, which corrects binocular disparity parameters in real time based on the user's head movement state to maintain a stable ensemble stimulation.

[0073] The closed-loop adaptive optimization module is used to construct a four-dimensional dynamic weight adjustment capability evaluation model and drive the continuous adaptive adjustment of training parameters based on the evaluation results. It also includes: a multi-dimensional data acquisition unit for collecting core training indicators such as gaze time, focus response time, focus stability, and ensemble adjustment coordination; a dynamic weight evaluation unit for dynamically calibrating and evaluating the four dimensions of adjustment flexibility, adjustment amplitude, adjustment accuracy, and ensemble adjustment coordination; a training parameter adaptive adjustment unit for progressively adjusting the depth of field change amplitude, depth of field change rate, ensemble adjustment ratio, and training relaxation cycle ratio based on the evaluation results; and a user baseline profile maintenance unit for dynamically updating the user adjustment capability baseline profile in an incremental manner with time-sensitive weights.

[0074] Furthermore, the progressive depth-of-field training module constructs a multi-layered depth-distribution target structure in a virtual 3D space and generates a progressive nonlinear depth-of-field change function with the user's adjustment capability as the initial parameter and the focus quality index as the feedback signal. This drives the dynamic adjustment of the virtual focal length parameter, thereby realizing interactive ciliary muscle adjustment training. The specific implementation is as follows:

[0075] When a user puts on a head-mounted virtual reality device and enters the training preparation phase, a multi-layered depth target distribution structure covering the complete adjustment range from near to far is established in the three-dimensional virtual space through depth-level building units. The set of refractive power values ​​of each depth target in the virtual scene is set as follows: ,in, Indicates the first The diopter value corresponding to each depth of field level Indicates the total number of depth levels This represents the constructed set of refractive power levels for depth of field. The refractive power values ​​for each level are evenly distributed within the range of 0.25D to 4D, covering the complete accommodation range from close-range fine work to distant natural gazing.

[0076] After the depth-of-field hierarchy is established, the depth-of-field variation function generation unit generates a personalized progressive depth-of-field variation function based on the user's initial adjustment capability assessment results. Specifically:

[0077] Adjustment range based on initial user assessment Using the fundamental parameters of the function, we construct a function for the progressive depth-of-field variation amplitude, then we have:

[0078] ;

[0079] in, Indicates the cumulative training unit number. This represents the individual fitness coefficient, which is updated in real time by the closed-loop adaptive optimization module based on the current adjustment capability assessment results. Indicates the first The depth of field variation corresponding to each training unit;

[0080] Simultaneously, by constructing the depth-of-field change rate modulation function, we have:

[0081] ;

[0082] in, Indicates the first The base rate of change for each training unit Indicates the first The rhythmic cycle of each training unit This represents a local time within a training unit. Indicates the first training units The depth-of-field change rate at any given time, the depth-of-field change rate modulation function makes the depth-of-field change rate exhibit a non-linear sinusoidal modulation rhythm of first slow, then fast and then slow again within each training unit, which is different from the uniform linear change method of existing technologies.

[0083] By combining the depth-of-field variation amplitude function with the rate modulation function to construct a complete progressive depth-of-field variation function, we have:

[0084] ;

[0085] in, This indicates the initial depth value of the current training unit. Indicates the first The target depth of field value at time t for each training unit.

[0086] By mapping the depth-of-field variation function to a real-time update instruction for the virtual target's focal length parameters through the dynamic focal length adjustment unit, we have:

[0087] ;

[0088] in, Indicates the first training units The virtual target focal length parameter at any given time, in meters, is used to drive the focal length rendering engine of the head-mounted virtual reality device to adjust the virtual focal length of the target object in real time.

[0089] While the focal length parameter drives changes in the virtual target, the visual accommodation stimulus is combined with the user's active control task through an active interactive task unit, specifically:

[0090] By setting task events on target objects at each depth level that require user interaction via controller or gaze, and constructing task trigger condition functions, we have:

[0091] when hour, ,on the contrary, ,in, Indicates user The coordinates of the gaze point at any given moment. express The spatial coordinates of the target object at any given time. This represents the gaze hit radius threshold. Indicates the duration of sustained gaze. This indicates the minimum duration of sustained gaze required to trigger the task. express The system's interactive training features are reflected in its real-time task triggering status, which transforms passive visual stimuli into proactive behavioral regulation through active interactive task design.

[0092] It should be noted that the progressive depth-of-field training module uses the focal length parameter sequence and task triggering state sequence The output is sent to the ensemble co-training module to drive the training of the dual-channel synchronous ensemble, and the focusing quality index is fed back to the closed-loop adaptive optimization module to update the individual fitness coefficients. This enables closed-loop adaptive management with progressively increasing difficulty.

[0093] Furthermore, the regulation-ensemble collaborative training module constructs a dual-channel synchronous control mechanism for regulation and ensemble based on individual user AC / A values. It enhances ensemble flexibility training through virtual prism deflection and calculates regulation lead-lag values ​​for ensemble regulation coordination scoring. The specific implementation is as follows:

[0094] First, the individual user's adjustment set ratio is estimated using the AC / A value estimation unit, specifically as follows:

[0095] During the training preparation phase, a standardized AC / A value measurement task is set up, presenting distances sequentially. and For a virtual gaze target, record the corresponding binocular convergence angle response value, then we have:

[0096] ;

[0097] in, and These represent the distances to the target. and The convergence angle of the two eyes measured at the location, in prism diopters / diopters. This represents the ratio of individual user adjustment sets. The estimation result is stored in the individual user parameter file and used as the reference parameter for dual-channel synchronous control.

[0098] Subsequently, the dual-channel synchronization control unit synchronously adjusts the focal length parameter and binocular disparity parameter based on the AC / A value, specifically as follows:

[0099] Focal length parameters output by the progressive depth of field training module To adjust the channel drive signal and synchronously calculate the corresponding target set angle requirement value, we have:

[0100] ;

[0101] in, This represents the base set angle when the user is in a distant gaze state. Indicates the first training units The target set angle requirement value at any given time;

[0102] Mapping the target set angle requirement value to binocular disparity adjustment, we have:

[0103] ;

[0104] in, Indicates the user's interpupillary distance. Indicates the first training units The amount of horizontal parallax adjustment of the binocular images at any given time is output to the virtual reality rendering engine in sync with the focal length adjustment and parallax adjustment, so as to achieve coordinated and consistent changes in adjustment and aggregation requirements.

[0105] Building upon standard depth-of-field alternation training, ensemble flexibility training tasks are periodically inserted into ensemble flexibility training units to enhance training capabilities. Specifically:

[0106] Constructing the virtual prism deflection parameter sequence, we have:

[0107] ;

[0108] in, This represents the basic virtual prism deflection. This indicates the prism deflection step size, which increases progressively. The total level of the set flexibility training is set by the implementers based on the user's set reserve capacity assessment results.

[0109] During the convergence flexibility training task execution phase, equal-sized but opposite-direction lateral virtual prism deflections are applied to the left and right eye images, forcing both eyes to complete the binocular fusion maintenance task under increased convergence load. The fusion recovery time of the user at each prism level is recorded, and then:

[0110] ;

[0111] in, Indicates the first Binocular fusion recovery time under prism deflection and These represent the moments of fusion startup and recovery to stability, respectively. Sequences are used as input metrics for set flexibility scoring.

[0112] Simultaneously, the deviation between the user's actual adjustment response point and the target focal length is calculated by adjusting the lead-lag detection unit, specifically as follows:

[0113] Real-time gaze depth estimates are collected using eye-tracking devices. With the current target focal length By comparing and calculating the adjustment lead-lag value, we have:

[0114] ;

[0115] in, Indicates the first training units The adjustment lead-lag value at time, in diopters. This indicates that the adjustment is ahead of schedule. This indicates a lag in regulation;

[0116] By statistically analyzing the adjustment lead-lag values ​​within a continuous time window and constructing a set adjustment coordination score, we have:

[0117] ;

[0118] in, Indicates the length of the statistical time window. This indicates the reference adjustment lead-lag threshold. Indicates the first The ensemble coordination score for each training unit, with a value range of [value missing]. A higher score indicates better coordination and consistency between regulation and the set.

[0119] It should be noted that the ensemble collaborative training module outputs the dual-channel synchronous control parameters, ensemble flexibility score, and ensemble adjustment coordination score to the closed-loop adaptive optimization module to update the four-dimensional adjustment capability evaluation model. At the same time, it outputs the disparity adjustment amount to the multimodal interactive feedback module for the synchronous generation of tactile feedback signals.

[0120] Furthermore, the psychological fatigue monitoring and intervention module constructs a psychological fatigue state vector by collecting six-dimensional fatigue sensitivity indicators. Based on a weighted fusion comprehensive score, the fatigue state is divided into three levels and corresponding differentiated intervention strategies are triggered. At the same time, structured psychological intervention is implemented for severe fatigue states through a psychotherapy content library. The specific implementation is as follows:

[0121] First, the six fatigue-sensitive dimensions are collected in real time through a multi-dimensional index acquisition unit, specifically:

[0122] By collecting the pupil oscillation amplitude using the eye-tracking component built into the headset, we have:

[0123] ;

[0124] in, Indicates the length of the statistical time window. express Pupil diameter at any given moment This represents the average pupil diameter within the window. express The amplitude of pupillary oscillation at any given moment increases significantly with increasing fatigue.

[0125] If the rate of decrease in gaze stability is collected, then:

[0126] ;

[0127] in, This represents the baseline gaze stability at the beginning of the user's training phase. express The current gaze stability is represented by the inverse of the variance of the gaze point coordinate sequence. Indicates the rate of decrease in gaze stability relative to the baseline;

[0128] The growth rate of the data acquisition task response latency is as follows:

[0129] ;

[0130] in, This indicates the baseline task response latency at the beginning of the user's training phase. express Current task response latency This represents the latency growth rate relative to a benchmark.

[0131] By collecting the low-to-high frequency ratio of heart rate variability using a wearable wristband device, we have:

[0132] ;

[0133] in, express Real-time heart rate variability low-frequency power Indicates high-frequency power. This represents the low-to-high frequency power ratio. An increase in the low-to-high frequency power ratio reflects an increase in the level of tension in the autonomic nervous system.

[0134] If we collect the rate of change in blink rate, then we have:

[0135] ;

[0136] in, Indicates the baseline blink rate, express Current blink rate This indicates the rate of decrease in blink rate relative to the baseline. A continuously decreasing blink rate suggests a state of high concentration fatigue.

[0137] Subjective fatigue scores are collected via voice input or gamepad shortcut keys during training breaks, resulting in:

[0138] ;

[0139] in, express The user inputs a subjective fatigue rating through a visual analog scale at any time, with 0 indicating no fatigue and 10 indicating extreme fatigue.

[0140] If we construct the six-dimensional fatigue index as a psychological fatigue state vector, then we have:

[0141] ;

[0142] in, express The six-dimensional psychological fatigue state vector at any given moment.

[0143] Subsequently, the six-dimensional indicators are weighted and integrated using a fatigue grading assessment unit to construct a comprehensive psychological fatigue score, which is then:

[0144] ;

[0145] in, Indicates the first The weighting coefficients of the dimensional indicators are initially set by the implementers based on prior knowledge of clinical fatigue sensitivity and are adaptively updated by the system during long-term operation. Indicates the first The standardized value of the dimension index after normalization. express A comprehensive psychological fatigue score at any given moment;

[0146] Based on the comprehensive psychological fatigue score, fatigue status is divided into three levels:

[0147] when At that time, it was determined to be a state of mild fatigue;

[0148] when At that time, it was determined to be a state of moderate fatigue;

[0149] when At that time, it was determined to be a state of severe fatigue;

[0150] in, and The thresholds for classifying mild and moderate fatigue states are set by the implementers based on the actual application scenario.

[0151] The differentiated intervention execution unit triggers an intervention response of corresponding intensity based on the fatigue level, specifically as follows:

[0152] When a mild fatigue state is detected, the system inserts a 30-second relaxation scene of looking into the distance, and simultaneously reduces the difficulty of the current training unit by one level, so that the training continues without interruption.

[0153] When the system determines that the patient is in a state of moderate fatigue, it pauses the current adjustment and reinforcement training task and switches to the active relaxation and recovery module for 5 minutes. After the active relaxation ends, the training is restarted with training parameters at a lower difficulty level.

[0154] When a state of severe fatigue is determined, the system forcibly terminates the current training session, records the rate of fatigue accumulation during the training session as a basis for optimizing subsequent training plans, and initiates a structured psychological intervention process through the psychotherapy content push unit.

[0155] The system pushes psychotherapy content to users based on their level of severe fatigue. Specifically:

[0156] The psychotherapy content library includes three categories: mindfulness meditation guidance, visual art healing scenarios, and breathing rhythm guidance. The system adaptively selects the type of content to be pushed based on the user's historical intervention response records. The duration of the pushed content is no less than 8 minutes. After the content ends, the training can only be restarted after the subjective fatigue score confirms whether the fatigue state has recovered to below mild level.

[0157] It should be noted that the psychological fatigue monitoring and intervention module continuously outputs the comprehensive psychological fatigue score Ψ(t) and fatigue accumulation rate to the active relaxation and recovery module and the closed-loop adaptive optimization module for adaptive adjustment of the training relaxation cycle and dynamic updating of training parameters.

[0158] Furthermore, the active relaxation and recovery module adaptively adjusts the duration ratio of the training and relaxation segments based on fatigue monitoring data. During the relaxation segment, it induces active relaxation of the ciliary muscle through a multi-sensory synergistic environment, specifically implemented as follows:

[0159] First, the training relaxation cycle management unit adaptively adjusts the training relaxation cycle parameters based on the user's fatigue accumulation rate, specifically as follows:

[0160] Define the fatigue accumulation rate as the average increase in the overall psychological fatigue score per unit training time, then:

[0161] ;

[0162] in, and They represent the first The overall psychological fatigue score at the beginning and end of each training unit. Indicates the first Training unit duration, Indicates the first The fatigue accumulation rate of each training unit;

[0163] If the duration of the next training unit and the duration of the relaxation period are adaptively adjusted based on the rate of fatigue accumulation, then:

[0164] ;

[0165] ;

[0166] in, and These represent the baseline training duration and baseline relaxation duration, respectively, with initial values ​​set to 15 minutes and 5 minutes, respectively. and Indicates the adjustment factor. and These represent the training and relaxation time for the next training unit, respectively. The higher the rate of fatigue accumulation, the shorter the training segment and the longer the relaxation segment will be.

[0167] During the relaxation phase, a virtual visual environment that induces active relaxation of the ciliary muscle is constructed through a distant viewing scene generation unit. Specifically:

[0168] Generate an open, natural virtual scene with an optical equivalent distance of at least 6 meters. The scene's visual content parameters are set as follows:

[0169] Spatial frequency limit: Week / Degree;

[0170] Image contrast: ;

[0171] Color temperature range: ;

[0172] in, This indicates the upper limit of the spatial frequency of scene content. Represents the Michelson contrast of the scene image. The color temperature of the scene is indicated by the parameters designed to minimize the excitation load on the visual cortex and promote the ciliary muscle to move towards its maximum relaxation state.

[0173] In a distant viewing scene, a progressive visual guide trajectory is overlaid, guiding the user's gaze point to gradually extend along the near-to-far direction. The depth-of-field variation function of the guide trajectory is set as follows:

[0174] ;

[0175] in, Indicates the initial depth of field value for guidance. This indicates the guiding termination depth value. Indicates the duration of the current relaxation period. Indicates relaxation section The depth of field value is guided at any moment to achieve gradual relaxation induction from near to far.

[0176] The multi-sensory synergistic relaxation unit simultaneously outputs tactile and auditory stimuli during the relaxation phase, specifically:

[0177] The headset's built-in vibration unit outputs low-frequency tactile feedback with a frequency range of 8Hz to 15Hz. The vibration intensity decreases linearly with the relaxation time. Therefore:

[0178] ;

[0179] in, Indicates the initial intensity of the vibration. Indicates relaxation section The intensity of vibration at different times simulates the natural relaxation process as the warmth gradually fades;

[0180] Synchronized playback of natural ambient soundscapes containing delta band binaural beats, with the binaural beat frequency set to [value missing]. to It assists the autonomic nervous system in switching to a parasympathetic dominant state through sensory-level neural regulation, thereby synergistically promoting the relaxation of the ciliary muscle.

[0181] It should be noted that the active relaxation recovery module uses the change in fatigue score after each relaxation segment as an evaluation index of relaxation effect, and continuously outputs it to the closed-loop adaptive optimization module to optimize relaxation parameter settings, ensuring that the relaxation module maintains effective ciliary muscle relaxation induction ability during long-term use.

[0182] Furthermore, the multimodal interactive feedback module constructs differentiated tactile vibration feedback synchronized with the rhythm of the visual training task, and integrates multi-source physiological data from wearable devices to enhance fatigue monitoring and real-time correction of training parameters. The specific implementation is as follows:

[0183] The visual-tactile synchronization unit outputs differentiated tactile vibration pulses that are consistent with the depth-of-field switching rhythm when the virtual target's focal length changes. Specifically:

[0184] The tactile vibration pulse coding rules are defined as follows:

[0185] When the focus shifts from far to near, that is... When the frequency decreases, a high-frequency short-pulse sequence is output, with a vibration frequency range of [missing value]. to The duration of a single pulse is ;

[0186] When the focus switching direction is from near to far, that is... When the frequency is increased, a low-frequency long-pulse is output, with a vibration frequency range of [missing information]. to The duration of a single pulse is ;

[0187] The synchronization delay between the tactile vibration pulse output time and the focal length parameter switching time shall not exceed This is to ensure effective time-bound association between visual and tactile stimuli;

[0188] The aforementioned differentiated tactile coding establishes a direction-specific conditional connection between the visual accommodation switching signal and the tactile perception signal, which helps reduce the user's accommodation response time and enhance the training immersion.

[0189] The physiological data fusion unit fuses multi-source physiological data collected from the head-mounted display's built-in sensors and the wearable wristband device, specifically:

[0190] By constructing a multi-source physiological data synchronous acquisition vector, we have:

[0191] ;

[0192] in, express Heart rate at all times express skin conductivity at any time express Head pose angle vector at any given time. express Multi-source physiological data vectors at different times;

[0193] Will and Output to the mental fatigue monitoring and intervention module for auxiliary verification of LHR index calculation and fatigue status assessment;

[0194] Head posture angle The output is sent to the real-time parameter correction unit for real-time compensation and correction of the disparity parameter.

[0195] The binocular disparity parameters are corrected in real time based on head movement by a parameter real-time correction unit. Specifically:

[0196] Calculate the head pose offset of the current frame relative to the reference frame, and we have:

[0197] ;

[0198] in, Indicates the time of the most recent parallax parameter calibration. Indicates the relative offset;

[0199] Compensating for the disparity adjustment output by the dual-channel synchronous control unit based on the head posture offset, we have:

[0200] ;

[0201] in, This represents the head motion parallax compensation coefficient. This indicates the amount of head posture offset in the horizontal direction. This indicates the corrected disparity adjustment amount after head movement compensation, ensuring that the binocular disparity stimulation amount remains stable when the user's head is moving, and avoiding convergence stimulation interference caused by head movement.

[0202] It should be noted that the multimodal interactive feedback module continuously outputs the corrected parallax adjustment amount to the virtual reality rendering engine to achieve dynamic and stable maintenance of the synergistic stimulation of the adjustment set; at the same time, it outputs the fused physiological data vector to the psychological fatigue monitoring and intervention module for real-time calculation of relevant indicators in the six-dimensional fatigue index system.

[0203] Furthermore, the closed-loop adaptive optimization module inputs the multi-dimensional data collected during training into the four-dimensional dynamic weight evaluation model, adaptively calibrates the weights based on the current achievement gap in each dimension, and drives the gradual adjustment of training parameters with the evaluation results. At the same time, it maintains the user's adjustment capability benchmark profile in an incremental manner with time-sensitive weights. The specific implementation is as follows:

[0204] First, by summarizing the core training metrics after each training unit through a multi-dimensional data acquisition unit, a training unit evaluation dataset is constructed, resulting in:

[0205] ;

[0206] in, Indicates the first Average focus response time per training unit Indicates the first The average adjustment amplitude estimated per training unit Indicates the first The tuning accuracy score for each training unit. Indicates the first The ensemble coordination score of each training unit. Indicates the first A four-dimensional evaluation dataset for each training unit.

[0207] Subsequently, dynamic weight calibration is performed on the four evaluation dimensions through a dynamic weight evaluation unit, specifically as follows:

[0208] Based on the current gap in compliance for each dimension, the weight adjustment amount is calculated, and then:

[0209] ;

[0210] in, Indicates the first The target values ​​for each dimension Indicates the first The training unit number The current rating of the dimension, Indicates the first The current gap in meeting the standards for each dimension.

[0211] Based on the achievement gap, the weights of each dimension are normalized and dynamically calibrated, resulting in:

[0212] ;

[0213] in, Indicates the first The training unit number The dynamic calibration weights of dimensions are adjusted so that the weights of dimensions that have met the standards are automatically reduced to zero, and the weights are concentrated on the dimensions with the largest gap in meeting the standards.

[0214] Constructing a comprehensive assessment score for regulatory capacity yields the following results:

[0215] ;

[0216] in, Indicates the first The comprehensive adjustment capability assessment score of each training unit focuses on reflecting the improvement status of the user's weakest adjustment capability dimension.

[0217] The training parameters are progressively adjusted based on the comprehensive evaluation results through an adaptive adjustment unit, specifically as follows:

[0218] The adjustment amount of each training parameter shall not exceed the current parameter value. The adjustment direction is determined based on the sign of the gap between the target and the corresponding dimension, then:

[0219] ;

[0220] in, This indicates the step size coefficient for parameter adjustment, and its value range is... , This indicates the change in the overall score of this assessment relative to the previous one. This indicates the adjusted parameter values ​​for the next training unit. The parameter adjustments cover the depth of field change magnitude, depth of field change rate, virtual prism deflection step size, and training-relaxation cycle ratio.

[0221] The user's adjustment capability baseline profile is dynamically updated in an incremental manner with time-weighted updates through the user baseline profile maintenance unit, specifically as follows:

[0222] Defining the time-sensitivity weight function, we have:

[0223] ;

[0224] Where, n Indicates the current training unit number. Indicates the historical training unit number. Indicates the aging decay coefficient. Indicates the first The timeliness weight of each historical training unit data at the current moment is considered, with more recent training data having a higher weight.

[0225] If the baseline values ​​for each dimension in the baseline file are updated using a weighted average with time-sensitive weights, then:

[0226] ;

[0227] in, Indicates the first Dimension in the th Updated baseline value after training units;

[0228] Historical data points in the baseline archive that are more than 90 days old and whose score deviation exceeds 30% of the baseline value are cleared to prevent expired data from affecting the current evaluation baseline and to ensure that the training parameters are always optimized based on the user's current actual adjustment ability.

[0229] Furthermore, after each comprehensive evaluation, the control parameters of each functional module are updated based on the comprehensive evaluation score, as follows:

[0230] Constructing a feedback update function, we have:

[0231] ;

[0232] in, Represents the set of global control parameters of the system. This indicates the feedback adjustment step size coefficient. This represents the gradient vector of the overall evaluation score relative to the changes in each control parameter.

[0233] It should be noted that the system will re-input the updated set of control parameters and user baseline profiles into each functional module for parameter initialization in the next training unit. This forms a continuously iteratively optimized closed-loop system for personalized vision rehabilitation training, which starts with the dynamic baseline of individual adjustment ability, takes four-dimensional dynamic weight assessment as the core, and uses progressive parameter adaptive adjustment as the driving mechanism. This achieves collaborative closed-loop optimization between the training model and the training process.

[0234] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the present invention.

Claims

1. An interactive vision training and rehabilitation system based on virtual reality (VR), characterized in that: include: The progressive depth-of-field training module is used to construct a multi-layer depth target distribution structure covering the refractive power range of 0.25D to 4D in virtual 3D space. Based on the user's initial adjustment ability assessment results, it generates a personalized progressive nonlinear depth-of-field change function with the focus quality index as the feedback signal, which drives the dynamic adjustment of the virtual focal length parameters to induce the ciliary muscle to produce periodic contraction and relaxation. The accommodation-accommodation coordination training module is used to automatically estimate the individual accommodation-accommodation ratio (AC / A) of a user through a standardized measurement task. Based on the AC / A value, a dual-channel synchronous control mechanism for accommodation and convergence channels is constructed. The focal length parameter adjustment and binocular disparity parameter adjustment are output synchronously according to the individual AC / A value. The module also simulates the application of additional convergence load through virtual prism deflection to train convergence reserve capacity. At the same time, the module calculates the accommodation lead-lag value between the user's actual accommodation response point and the target focal length, and generates a convergence accommodation coordination score. The psychological fatigue monitoring and intervention module is used to collect six-dimensional fatigue-sensitive indicators, including pupillary oscillation amplitude, gaze stability decline rate, task response delay growth rate, heart rate variability low-to-high frequency ratio, blink rate change rate, and subjective fatigue score. It constructs a psychological fatigue state vector, generates a comprehensive psychological fatigue score through weighted fusion, classifies fatigue state into three levels: mild, moderate, and severe, and triggers differentiated intervention strategies. In the severe fatigue state, psychological therapy content is pushed out. The active relaxation and recovery module is used to adaptively adjust the duration ratio of the training segment and the relaxation segment according to the fatigue accumulation rate. In the relaxation segment, a low spatial frequency, low contrast open natural virtual scene with an optical equivalent distance of not less than 6 meters is constructed. The ciliary muscle is induced to move to the maximum relaxation state through a progressive visual guidance trajectory from near to far. At the same time, low frequency vibration tactile feedback and environmental sound scene are output simultaneously to form a multi-sensory collaborative relaxation environment. The multimodal interactive feedback module is used to synchronously output differentiated tactile vibration pulses that are consistent with the depth-of-field switching rhythm when the virtual target focal length switches. It also integrates heart rate, skin conductivity and head posture data collected by the built-in sensors of the head display and the wristband device, and corrects the binocular parallax parameters in real time according to the head movement state to maintain a stable ensemble stimulation. The closed-loop adaptive optimization module is used to collect four core training indicators: average focus response time, adjustment amplitude, adjustment accuracy score, and ensemble adjustment coordination score. Based on the current gap in the achievement of each dimension, the weights are dynamically normalized and calibrated to construct a comprehensive adjustment capability assessment score. The assessment results drive the progressive adaptive adjustment of training parameters and dynamically update the user's adjustment capability benchmark file in an incremental manner with time-sensitive weights.

2. The interactive vision training and rehabilitation system based on virtual reality (VR) as described in claim 1, characterized in that: The progressive depth-of-field training module is as follows: A multi-layered depth target distribution structure is established in a 3D virtual space using depth-of-field hierarchical building units, and the set of depth-of-field refractive power values ​​for each layer is defined. The refractive power values ​​at each level are as follows: to Uniformly distributed within the range; The depth-of-field variation function generation unit generates the depth of field based on the user's initial adjustment range. Constructing a progressive depth-of-field variation amplitude function ,in, Indicates the cumulative training unit number. Represents the individual adaptation coefficient, and simultaneously constructs the depth-of-field change rate modulation function. The two are combined to construct a complete progressive depth-of-field variation function. ; The depth-of-field variation function is mapped to the virtual target focal length parameter through the dynamic focal length adjustment unit. ; By setting task events on target objects at each depth level through active interaction task units, which require user gaze interaction to trigger, a task trigger condition function is constructed: when... hour, ,on the contrary, ,in, Indicates user The coordinates of the gaze point at any given moment. express The spatial coordinates of the target object at any given time. This represents the gaze hit radius threshold. Indicates the duration of sustained gaze. This indicates the minimum duration of sustained gaze required to trigger the task. express The task trigger status at any given moment.

3. The interactive vision training and rehabilitation system based on virtual reality (VR) as described in claim 2, characterized in that: The adjustment set collaborative training module is specifically as follows: During the training preparation phase, a standardized measurement task is set up using the AC / A value estimation unit, presenting distances sequentially. and The virtual gaze target is recorded, and the corresponding binocular convergence angle response value is recorded. and According to the formula Estimate the individual user's adjustment set ratio and store the estimation results in the individual user parameter file; Using the focal length parameter via a dual-channel synchronous control unit To adjust the channel drive signal, the target set angle requirement value is calculated synchronously. Mapping the target set angle to binocular disparity adjustment Where IPD is the user's interpupillary distance. This represents the base convergence angle when the user is in a distant gaze state, and outputs the focus adjustment and parallax adjustment to the virtual reality rendering engine simultaneously; A virtual prism deflection parameter sequence is constructed using ensemble flexibility training units. Equal-sized, opposite-direction lateral virtual prisms were applied to the left and right eye images, forcing both eyes to complete the binocular fusion maintenance task under increasing convergence load. The fusion recovery time at each prism level was recorded. ; By adjusting the lead-lag detection unit, the user's real-time gaze depth estimation value is collected. Calculate the adjustment lead-lag value And construct a set of coordination scores. .

4. The interactive vision training and rehabilitation system based on virtual reality (VR) as described in claim 3, characterized in that: The psychological fatigue monitoring and intervention module is as follows: Six fatigue-sensitive indicators were collected using a multi-dimensional indicator acquisition unit: pupil oscillation amplitude. ; gaze stability decline rate ; Task response latency growth rate ; Heart rate variability low-high frequency ratio ; Blink rate change rate ; Subjective fatigue rating ; Constructing a vector of mental fatigue states, we have: ； A comprehensive psychological fatigue score is constructed by weighting and integrating the six-dimensional indicators through a fatigue grading assessment unit. ,in accordance with With threshold , The relationship is divided into three fatigue levels: mild, moderate, and severe. The corresponding intervention is triggered by the differentiated intervention execution unit: when the fatigue is mild, insert a 30-second relaxation scene and reduce the current training difficulty by one level; when the fatigue is moderate, pause the training and switch to the 5-minute active relaxation and recovery module; after the relaxation ends, downgrade and restart the training; when the fatigue is severe, forcibly terminate the current training unit and record the fatigue accumulation rate. The psychotherapy content push unit adaptively selects and pushes content from a psychotherapy content library that includes mindfulness meditation guidance, visual art healing scenarios, and breathing rhythm guidance when the user is in a state of severe fatigue. The push duration is no less than 8 minutes. Training can only be restarted after the user has finished the content and the user has recovered to a level of mild or below through a subjective fatigue score.

5. The interactive vision training and rehabilitation system based on virtual reality (VR) as described in claim 4, characterized in that: The active relaxation and recovery module is as follows: The fatigue accumulation rate is calculated using the training relaxation cycle management unit. The duration of the next training unit is adaptively adjusted based on the rate of fatigue accumulation. With relaxation time ,in, and These represent the baseline training duration and the baseline relaxation duration, respectively. and Indicates the adjustment factor; An open, natural virtual scene with an optical equivalent distance of no less than 6 meters is constructed using a distant view scene generation unit, with scene parameters set to the upper limit of spatial frequency. Weekly, image contrast Color temperature range And overlay progressive visual guidance trajectories on the scene. Inducing the user's gaze point to extend from near to far to promote active relaxation of the ciliary muscle; The synchronous output frequency range of the multi-sensory synergistic relaxation unit is [missing information]. to The low-frequency vibration tactile feedback has a vibration intensity that decreases linearly with relaxation time. And simultaneously play content with frequency of to of The band-based binaural beat-based natural environmental soundscape assists the autonomic nervous system in switching to a parasympathetic dominant state.

6. The interactive vision training and rehabilitation system based on virtual reality (VR) as described in claim 5, characterized in that: The multimodal interactive feedback module is as follows: The visual-tactile synchronization unit outputs differentiated tactile vibration pulses that match the depth-of-field switching rhythm during focus switching: when the focus switching direction is from far to near... When reduced, the output frequency range to Single pulse duration A high-frequency short-pulse sequence, when the focus switching direction is from near to far, i.e. When increased, the output frequency range to Single pulse duration The low-frequency long vibration pulse, the synchronization delay between the output time of the tactile vibration pulse and the switching time of the focal length parameter does not exceed Establish a direction-specific conditional connection between visual accommodation switching signals and tactile perception signals; A multi-source physiological data synchronous acquisition vector is constructed using a physiological data fusion unit. ,in, express Heart rate at all times express skin conductivity at any time express Head pose angle vector at any given time. express Multi-source physiological data vectors at time points, and The output is sent to the mental fatigue monitoring and intervention module for LHR index calculation and auxiliary verification; The head posture offset is calculated using a real-time parameter correction unit. Compensate for the parallax adjustment amount based on the offset. ,in, This is the head motion parallax compensation coefficient, ensuring that the ensemble stimulus remains stable while the user's head is moving.

7. The interactive vision training and rehabilitation system based on virtual reality (VR) as described in claim 6, characterized in that: The closed-loop adaptive optimization module is as follows: By summarizing the core training metrics after each training unit through a multi-dimensional data acquisition unit, a four-dimensional training unit evaluation dataset is constructed. ,in, Indicates the first Average focus response time per training unit Indicates the first The average adjustment amplitude estimated per training unit Indicates the first The tuning accuracy score for each training unit. Indicates the first The ensemble coordination score of each training unit. Indicates the first A four-dimensional evaluation dataset for each training unit; The achievement gap for each dimension is calculated using a dynamic weighting evaluation unit. Perform normalization and dynamic calibration on the weights. The weights of dimensions that have already met the standards are automatically reduced to zero, and the weights are then concentrated on the dimensions with the largest gap in compliance, thus constructing a comprehensive adjustment capability assessment score. ; The unit adaptively adjusts training parameters based on comprehensive evaluation results. The depth of field variation range, depth of field variation rate, virtual prism deflection step size, and training-relaxation cycle ratio were progressively adjusted. This indicates the step size coefficient for parameter adjustment, and its value range is... ; Define the timeliness weight function through the user baseline profile maintenance unit. The baseline values ​​for each dimension are updated using a weighted average method with time-sensitive weights. And for training sessions that are more than 90 days old and whose score deviation exceeds the baseline value. Perform a clear operation on historical data points; After each comprehensive evaluation, the system's global control parameter set is updated based on the comprehensive evaluation score. ,in, Represents the set of global control parameters of the system. This indicates the feedback adjustment step size coefficient. This represents the gradient vector of the overall evaluation score relative to the changes in each control parameter. The updated set of control parameters and the user baseline profile are then re-entered into each functional module for the initialization of the next training unit.

8. The interactive vision training and rehabilitation system based on virtual reality (VR) as described in claim 7, characterized in that: The progressive depth-of-field training module generates progressive depth-of-field variation functions through the depth-of-field variation function generation unit, as follows: Based on the user's initial adjustment range Construct a progressive depth-of-field variation amplitude function based on basic parameters To achieve a gradual increase in difficulty with a logarithmic growth pattern, the individual adaptation coefficient... The closed-loop adaptive optimization module updates the data in real time based on the current adjustment capability assessment results. Constructing a depth-of-field rate modulation function ,in, Indicates the first The base rate of change for each training unit Indicates the first The rhythmic cycle of each training unit This represents the local time within a training unit, causing the depth-of-field change rate to exhibit a non-linear sinusoidal modulation rhythm that is slow at first, then fast, and then slows down again within each training unit. By combining the depth-of-field variation amplitude function and the rate modulation function, a complete progressive depth-of-field variation function is constructed. , This is the initial depth value for the current training unit; The depth-of-field variation function is mapped to the virtual target focal length parameter through the dynamic focal length adjustment unit. It drives the focal length rendering engine of the head-mounted virtual reality device to adjust the virtual focal length of the target object in real time.

9. The interactive vision training and rehabilitation system based on virtual reality (VR) as described in claim 8, characterized in that: The psychological fatigue monitoring and intervention module achieves closed-loop intervention through a psychological fatigue state vector and a three-level fatigue grading assessment, as detailed below: Six-dimensional fatigue-sensitive indicators were collected through a multi-dimensional indicator collection unit, and a psychological fatigue state vector was constructed. A comprehensive psychological fatigue score is constructed using fatigue grading assessment units, based on... With threshold , If the relationship is divided into three levels, then: when At that time, it was determined to be a state of mild fatigue. At that time, it was determined to be a state of moderate fatigue. At that time, it was determined to be a state of severe fatigue; The differentiated intervention execution unit triggers corresponding intensity intervention responses based on the three levels of fatigue. The psychotherapy content push unit adaptively selects push content from the psychotherapy content library, which includes three types of content: mindfulness meditation guidance, visual art healing scenarios, and breathing rhythm guidance, based on the user's historical intervention response records when the user is in a state of severe fatigue. The comprehensive psychological fatigue score and fatigue accumulation rate are continuously output to the active relaxation and recovery module and the closed-loop adaptive optimization module for adaptive adjustment of the training relaxation cycle and dynamic updating of training parameters.

Images