Multi-modal visual stimulation based on ar glasses and intelligent adaptive combined myopia prevention and control system

Abstract

The multi-modal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses realizes all-weather, personalized and full-scene myopia inhibition through the cooperation of multiple modules such as sensory visual stimulation, visual function training, AI scene dynamic identification and directional intervention, adaptive closed-loop feedback and cloud / offline deployment. Through multi-dimensional visual stimulation and real-time physiological monitoring, the growth of the eye axis is accurately controlled, the precision of myopia prevention and control is improved, the off-focus and contrast setting is automatically optimized based on instant refraction and accommodation feedback to strengthen personalized intervention, light-weight AI and cloud pre-computation are introduced to effectively extend the device endurance, the existing AR platform is compatible, the development process is simplified, and the user compliance is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of myopia prevention and control, and in particular to a multi-modal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses. BACKGROUND

[0002] Peripheral defocus refers to the use of glasses or other devices to make the light rays in the peripheral area of the retina defocus, thereby reducing the growth of the eye axis and thus inhibiting the progression of myopia. Peripheral defocus glasses stimulate the growth mechanism of the eye to change by causing defocus signals in the peripheral area of the retina, thereby delaying the deepening of myopia. Existing studies have shown that peripheral defocus glasses (such as multi-zone optical positive defocus lenses, referred to as DIMS lenses) have a certain effect on slowing down the progression of myopia, but their effect is not effective for all people. For example, Liu et al. found in Liu J, Lu Y, Huang D, et al. Ophthalmology 2023; 130(5): 542-550 that the average increase in myopia degree of DIMS lens wearers within 2 years was only 0.35D lower than that of single-focus glasses, but 12% of the DIMS lens wearers had an increase in diopter of more than 1.5D within 2 years. Similarly, Lam et al. found in Lam CSY, Tang WC, Tse DY, et al. Br J Ophthalmol 2020; 104: 363-368 that 13% of the DIMS lens wearers had an increase in diopter of more than 1D. This shows that the effect of peripheral defocus glasses is affected by individual differences, the peripheral area of the retina, and the structural differences of the cornea and the eye axis. Individual differences may be one of the key factors that cause inconsistent effects of peripheral defocus among different people.

[0003] In addition to peripheral defocus glasses, researchers have also begun to explore the potential of augmented reality (AR) glasses as a tool for myopia prevention in recent years. AR glasses achieve information enhancement by superimposing virtual images in the wearer's field of view. Recent studies have shown that AR technology can provide a kind of "defocus" signal when working at close range, which is expected to slow down the thinning of choroid thickness and blood flow changes caused by long-term close-range eye use. Considering the promising application prospects of AR glasses as a new optical technology in myopia prevention,

[0004] Existing myopia prevention methods include peripheral positive defocus soft lenses, OK lenses, and AR systems based on single defocus, most of which lack coordinated regulation and dynamic adaptive optimization of central and peripheral visual pathways, and cannot reproduce outdoor natural light and spatial characteristics in indoor environments. They also do not fully utilize AI for real-time signal processing and interactive optimization. SUMMARY

[0005] In order to solve the problems in the prior art, the application provides a multi-modal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses.

[0006] The technical solution adopted by the application is that the multi-modal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses comprises:

[0007] The visual stimulation module: relying on the optical projection capability of the AR glasses, personalized stimulation signals are dynamically generated in the peripheral visual field 10-30 degree area, the stimulation parameters are adjusted in real time according to the environmental illumination, eye movement characteristics and use scene, the outdoor visual experience is simulated, and the regulation system load control is strengthened;

[0008] The visual training module: the central visual field area presents dynamic visual tracking tasks, binocular fusion targets and contrast sensitivity tests, the visual aggregation and fixation stability functions are enhanced, the system design of the visual training module is a central-peripheral visual feedback chain: when the central visual training reaches the preset effect, the strength and range of the peripheral defocus stimulation of the visual stimulation module are automatically synchronized and triggered to enhance, and stronger peripheral perception response is excited, so that the functions are coordinated;

[0009] The AI scene dynamic identification and directional intervention module: the system integrates a light semantic recognition model, performs real-time image acquisition and intelligent classification on the environment, and comprehensively evaluates the eye use risk in combination with the illumination intensity, eye use distance and eye movement characteristics, and automatically starts the directional intervention strategy for different risk level scenes;

[0010] The adaptive closed-loop feedback module: by collecting the key physiological parameters of the user's blink frequency, pupil diameter change and fixation stability in real time, the visual fatigue degree and accommodation load state are comprehensively judged, according to the feedback information, the system dynamically fine tunes the stimulation strength, frequency and duration of the three modules of the visual stimulation module, the visual training module and the directional intervention module, and sends a prompt to the user in the form of voice, graphics or vibration, and builds a closed-loop control.

[0011] The personalized stimulation signals comprise defocus gratings with variable accommodation strength, low-contrast disturbance patterns and spatial frequency stripes.

[0012] The stimulation parameters in the visual stimulation module comprise defocus strength, contrast, and spatial frequency.

[0013] The automatic directional intervention strategy for different risk level scenes in the AI scene dynamic identification and directional intervention module comprises: improving the display contrast in a low-contrast environment; enhancing the defocus signal and spatial frequency adjustment when continuously fixating at a close distance; and adapting high-frequency stimulation compensation in a task mainly based on low-frequency visual stimulation.

[0014] The self-adaptive closed-loop feedback module sends a prompt to the user, including a rest suggestion, an adjustment of the sitting posture suggestion, and a distant view suggestion.

[0015] The personalized stimulation signal is in the form of a concentric ring or a radial gradient.

[0016] The peripheral visual field in the visual stimulation module further includes a central visual field of 0°-10°, a middle peripheral visual field of 10°-20°, and a far peripheral visual field of 20°-30°.

[0017] The middle peripheral visual field and the far peripheral visual field are respectively set to have a contrast gradient of 5%-20% and 20%-50% to differentially activate the ON / OFF pathway.

[0018] The personalized stimulation signal dynamically generated in the peripheral visual field 10°-30° in the visual stimulation module is a positive defocus pattern of +0.5D to +4.5D or a negative defocus pattern of -0.5D to -4.5D superimposed on the peripheral visual field 10°-30° according to the current refractive power.

[0019] The cloud / offline deployment module is further included, which uploads the multi-modal data to a server in real time, uses a distributed AI model to batch calculate and analyze, periodically generates individual and group reports, pushes the latest parameters to the device, and locally stores the data and synchronizes regularly to avoid data loss.

[0020] The beneficial effects of the present application are as follows: a multi-modal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses, which realizes all-weather, personalized, and full-scene myopia inhibition through the cooperation of multiple modules such as sensory visual stimulation, visual function training, AI scene dynamic identification and targeted intervention, self-adaptive closed-loop feedback, and cloud / offline deployment, accurately regulates the growth of the eye axis through multi-dimensional visual stimulation and real-time physiological monitoring, improves the accuracy of myopia prevention and control, automatically optimizes the defocus and contrast settings based on instant refraction and accommodation feedback to strengthen personalized intervention, introduces lightweight AI and cloud pre-computation to effectively extend the device's endurance, is compatible with existing AR platforms, simplifies the development process, and enhances user compliance. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 The results of the change in choroidal thickness after close-range work under the condition of AR on or off.

[0022] Figure 2 The results of the change in choroidal vascular index after close-range work under the condition of AR on or off.

[0023] Figure 3 The results of the change in axial length after close-range work under the condition of AR on or off. DETAILED DESCRIPTION

[0024] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0025] The present application realizes myopia intervention in all-weather, personalization and full-scene through multi-module cooperation such as sensory visual stimulation, visual function training, AI scene dynamic identification and directional intervention, adaptive closed-loop feedback and cloud / offline deployment.

[0026] Visual stimulation module

[0027] Relying on the optical projection capability of AR glasses, personalized stimulation signals are dynamically generated in the peripheral visual field 10°-30° area, and the stimulation parameters are adjusted in real time according to the environmental illumination, eye movement characteristics and use scene to simulate outdoor visual experience and strengthen the regulation of the accommodation system load.

[0028] Among them, according to the current refractive power, +0.5D to +4.5D positive defocus or -0.5D to -4.5D negative defocus patterns are superimposed in the peripheral visual field (10°-30°) to realize peripheral defocus superposition.

[0029] Reason: Peripheral positive defocus has been proved by a large number of studies to inhibit axial elongation, and by adjusting the defocus degree for different refractive states, personalized intervention can be achieved.

[0030] The defocus pattern is designed as a concentric ring or a radial gradient form to ensure visual comfort; Reason: The concentric ring or the gradient pattern can smoothly transition the focus difference area and reduce visual fatigue and discomfort.

[0031] Segmented contrast adjustment

[0032] The field of view is divided into three segments: central (0°-10°), mid-peripheral (10°-20°), and far-peripheral (20°-30°);

[0033] Reason: Visual sensitivity differs at different angles, and segmentation can achieve more accurate pathway activation.

[0034] 5%-20% and 20%-50% contrast gradients are set in the mid-peripheral and far-peripheral respectively to activate the ON / OFF pathways differently;

[0035] Reason: Differentiated contrast stimulation can enhance the alternating activation of ON / OFF pathways, helping to balance accommodation and eliminate visual fatigue.

[0036] Wavelength and spatial frequency compensation

[0037] Output 460-480 nm short-wave blue light, 600-650 nm red light, and wide-spectrum white light to optimize pupil response; Reason: Short-wavelength blue light can promote retinal dopamine release and inhibit axial growth; Red light and white light are used alternately to maintain comfort and physiological safety.

[0038] Real-time detection of environmental image spatial frequency distribution, superimpose 0.1-5 c / deg virtual texture in high-frequency deficient area, compensate for indoor scene; Reason: Lack of high-frequency information in indoor environment may exacerbate myopia, and compensation of high-frequency texture can simulate natural depth of field and promote accommodation exercise.

[0039] Visual training module

[0040] The central visual field area enhances visual aggregation and gaze stability function by presenting dynamic visual tracking tasks, binocular fusion targets and contrast sensitivity tests. The system design of the visual training module is a "central-peripheral" visual feedback chain: when the central visual training reaches the preset effect, the strength and range of peripheral defocus stimulation of the visual stimulation module is automatically synchronized to trigger the enhancement, thereby realizing functional synergy.

[0041] Accommodation rate and latency training

[0042] Virtual depth of field switching task: by switching between 1D-4D virtual focal planes, record the accommodation response time of ciliary muscle;

[0043] Reason: Evaluate and improve the response speed of the accommodation system to reduce the stretching pressure on the axial length caused by continuous focal length locking.

[0044] Adjust the training difficulty and frequency dynamically according to the measurement results;

[0045] Reason: Personalized training helps to continuously improve accommodation function and avoid one-size-fits-all training solutions.

[0046] Gaze stability and fusion ability

[0047] Central static target recognition: display 0.5-2 arcmin details and record recognition accuracy;

[0048] Reason: Poor gaze stability can increase the burden of accommodation, and target recognition tasks can improve gaze control. Binocular disparity rendering: render ±10 arcsec disparity targets to improve fusion tolerance;

[0049] Reason: Enhance binocular fusion function and reduce abnormal changes in axial length caused by disparity mismatch.

[0050] AI scene dynamic identification and directional intervention module

[0051] System integrates lightweight semantic recognition model, conducts real-time image acquisition and intelligent classification on the environment, and combines light intensity, eye distance and eye movement characteristics to comprehensively evaluate eye use risk. For different risk level scenes, automatically start the targeted intervention strategy.

[0052] Dynamic scene classification

[0053] Real-time recognition of reading, screen, outdoor natural scenes, etc. by using lightweight semantic segmentation model;

[0054] Reason: Accurate distinction of eye use tasks can avoid irrelevant scene interference and optimize myopia prevention and control logic.

[0055] Scene label update frequency ≥ 1 Hz, ensuring immediate response of intervention; high refresh rate can capture scene changes in time and avoid intervention lag.

[0056] Targeted defocus and contrast strategy

[0057] Reading / screen scene: automatically superimposes +1.0D positive defocus and reduces central contrast by 10%;

[0058] Reason: When reading for a long time or using electronic screens, the eye axis is prone to elongation. Positive defocus and contrast reduction can reduce the burden of near accommodation.

[0059] Outdoor natural scene: switch to peripheral positive defocus superposition mode;

[0060] Reason: High outdoor light and mild positive defocus work together to help simulate the protective effect of natural environment on eye axis growth.

[0061] Outdoor mode is silent: turn off the central visual stimulus and only provide defocus signals to the periphery (10°-30°);

[0062] Reason: In real outdoor or simulated outdoor scenes, avoid central content interference, make the peripheral defocus signal more focused, and improve the intervention effect.

[0063] Adaptive closed-loop feedback module

[0064] By collecting the user's blink frequency, pupil diameter changes, and fixation stability key physiological parameters in real time, the system comprehensively judges the degree of visual fatigue and the state of accommodation load. According to the feedback information, the system dynamically adjusts the stimulation intensity, frequency and duration of the three modules of the front visual stimulation module, the visual training module and the targeted intervention module, and sends a prompt to the user in the form of voice, graphics or vibration, building a closed-loop control.

[0065] Multi-modal physiological monitoring

[0066] Eye movement and pupil: sampling frequency ≥ 120Hz, extract saccade, fixation and pupil diameter changes;

[0067] Reason: High-frequency monitoring can capture subtle physiological changes and provide feedback to regulate training effectiveness and fatigue levels.

[0068] Online strategy optimization

[0069] The reinforcement learning algorithm automatically fine-tunes the defocus intensity and distribution based on short-term feedback (adjusting accuracy, fatigue index);

[0070] Reason: Reinforcement learning can explore optimal intervention strategies and achieve personalized dynamic adjustment.

[0071] Long Short-Term Memory (LSTM) networks predict myopia progression trends and trigger long-term parameter adjustments or expert recommendations.

[0072] Reason: Time series forecasting can detect abnormal developments in advance, supporting proactive intervention.

[0073] Based on real-time acquired eye parameters (such as instantaneous accommodation response, pupil diameter changes, and refractive power fluctuations), the defocus mode, contrast distribution, and training task difficulty are dynamically adjusted.

[0074] Reason: Direct feedback of physiological parameters can further optimize the stimulation effect and improve the accuracy and comfort of prevention and control.

[0075] Based on real-time monitoring of refractive status and accommodative hysteresis index, the system automatically selects the optimal defocus strategy and mode (positive defocus / negative defocus or hybrid mode) and switches it instantly when the scene changes.

[0076] Reason: Precise matching of individual refractive characteristics and accommodative ability can maximize the effect of inhibiting axial elongation.

[0077] Abnormal alerts and adaptive eye load mechanism

[0078] Eye strain assessment: The strain level is determined by combining the viewing distance, duration and scene mode (e.g., if the distance is less than 20cm and the continuous use exceeds 20 minutes and is in a gaming scene, it is considered a high strain).

[0079] Reason: Closer distances, longer durations, and more interactive scenarios significantly increase the burden of regulation, requiring stronger intervention.

[0080] Increase intervention intensity under high load: When entering a high load state, automatically expand the defocus range in the peripheral field of view (e.g., from +0.5D to +1.5D) and expand the contrast adjustment area to 0°–30°;

[0081] Reason: Under high load, enhanced visual stimulation can more effectively inhibit axial elongation.

[0082] High load for more than 30 minutes, automatically enter the virtual depth switching task;

[0083] Reason: Prolonged continuous use of eyes can cause visual fatigue and accommodation constant, virtual depth switching task to evaluate and train, relax ciliary muscle, relieve accommodation tension.

[0084] Cloud / Offline Deployment Module

[0085] Real-time upload of multi-modal data to the server, batch calculation and analysis using distributed AI model; Periodic individual and group reports are generated, and the latest parameters are pushed to the device; Local data storage, regular synchronization to avoid data loss.

[0086] Cloud Batch Optimization

[0087] Real-time upload of multi-modal data to the server, batch calculation and analysis using distributed AI model;

[0088] Reason: Cloud powerful computing power can conduct deep mining and model training on large-scale data.

[0089] Periodic individual and group reports are generated, and the latest parameters are pushed to the device;

[0090] Reason: Reports help parents and clinical staff understand the effectiveness of the intervention and make scientific decisions.

[0091] Offline Real-time Inference

[0092] Local embedded lightweight neural network and database, supporting ≥90% functions without network status;

[0093] Reason: To ensure that the system can still provide personalized intervention in offline scenarios.

[0094] Local storage of the last 30 days of data, regular synchronization to avoid data loss.

[0095] Reason: Local storage and delayed synchronization mechanism can prevent data loss and protect user privacy.

[0096] System Hardware Architecture

[0097] Contains high-brightness micro display, diffraction grating waveguide, wide-spectrum LED light source, binocular camera and eye pupil tracking sensor, high-performance AI processor, power management and cooling system, to support the efficient operation of the above functional modules.

[0098] Experimental Data

[0099] This invention will explore the impact of AR glasses on choroid and blood flow, and study their potential for controlling myopia progression.

[0100] Method:

[0101] Participants

[0102] This study recruited adults with myopia at the Eye Hospital of Wenzhou Medical University for a prospective comparative study. The recruitment criteria for participants were as follows: (1) age 18 to 30 years; (2) binocular equivalent spherical refractive error (SE) -1.00D to -5.75D, astigmatism ≤1.00D, and spherical asymmetry ≤1.00D; (3) best corrected visual acuity of at least 0.0 logMAR; (4) intraocular pressure (IOP) ≤21 mmHg and interocular pressure difference ≤5 mmHg; (5) no history of eye surgery, organic eye disease, or systemic disease; (6) exclusion of patients who cannot adapt to contact lens wear, those with cognitive impairment, patients with neurological and mental diseases, and those who cannot cooperate.

[0103] Intervention and experimental procedures

[0104] In this study, the "on" state of AR glasses served as the intervention, and the "off" state served as the control (INMO, China; 640×400 resolution per eye, binocular display). The order of "on" or "off" of the devices was randomized. The experiment employed a self-controlled pre- and post-operative design. All participants underwent two follow-up visits, with their vision fully corrected via contact lenses. Within one week, participants were required to participate in two reading sessions at least 24 hours apart. During these two reading sessions, the AR glasses were randomly switched on or off. All tests were completed between 13:00 and 17:00. Participants were required to read different chapters of the same novel on a tablet computer (Huawei, China, 2560x1600 resolution), which was placed 33cm away from their eyes in the actual space. To avoid the image interfering with the participants' normal reading, a white screen image was displayed 2 meters above their field of vision for 30 minutes when the AR glasses were on.

[0105] Measurement

[0106] Before and after near work, choroidal thickness (ChT) and choroidal vascular index (CVI) were acquired using swept-source optical coherence tomography (SS-OCT, VG200I, Intalight, Henan, China). Axial length (AL) was measured using a Haag-Streit Lenstar LS900 (Haag Streit AG, Könitz, Switzerland). All measurements were performed in the right eye. Each data collection was performed three times, and the average was taken.

[0107] Choroidal thickness (ChT) was defined as the distance between Bruch's membrane (BM) and the inner choroidal interface. The 6 mm macular scan area was automatically divided into 9 regions. The 9 regions were 1 mm central region (C), 1 mm to 3 mm nasal region (N3), 1 mm to 3 mm superior region (S3), 1 mm to 3 mm temporal region (T3), 1 mm to 3 mm inferior region (I3), 3 mm to 6 mm nasal region (N6), 3 mm to 6 mm superior region (S6), 3 mm to 6 mm temporal region (T6), and 3 mm to 6 mm inferior region (I6). The same OCT device (VG200I) was also used to obtain optical coherence tomography angiography (OCTA) images. The images were automatically divided into 9 regions with the same region division as ChT.

[0108] Statistical analysis

[0109] Statistical analysis was performed using SPSS 26.0. Normality was tested by Shapiro-Wilk test. Repeated measures ANOVA (RM-ANOVA) was used to analyze the differences in choroidal thickness (ChT), choroidal vessel index (CVI), and axial length (AL) after near work between the “on” and “off” states of AR glasses. P values less than 0.05 were considered statistically significant. Results were expressed as mean ± standard error (SE).

[0110] Results

[0111] A total of 21 participants were included, and 1 participant was excluded due to poor cooperation. There were 8 males and 13 females. The average age was 24 ± 0.51 years, and the spherical equivalent refraction (SER) was -3.74 ± 0.32 D. There were no significant differences in choroidal thickness (ChT), choroidal vessel index (CVI), and axial length (AL) before the two follow-ups (p ≥ 0.189).

[0112] Choroidal thickness

[0113] The “on” and “off” states of AR glasses had a significant main effect on the change in choroidal thickness after near work (F = 15.599, p = 0.001). Different regions had no effect on the change in choroidal thickness (F = 0.527, p = 0.584), and the interaction between the “on” and “off” states of AR glasses and different regions was also not significant (F = 1.407, p = 0.257).

[0114] There were significant differences in C (0-1 mm, 4.967 ± 1.389 μm, p = 0.010), T3 and T6 (1-3 mm and 3-6 mm, 4.302 ± 1.032 μm and 4.429 ± 1.829 μm, p < 0.001 and p = 0.025) and I6 (3-6 mm, 2.731 ± 0.982 μm, p = 0.012) between the AR glasses “on” and “off” states. The magnitude of the significant decrease in choroidal thickness was significantly reduced when the AR glasses were turned on.

[0115] Choroidal vascular index and axial length

[0116] There was no significant main effect of AR glasses “on” and “off” states on the choroidal vascular index (CVI) (F = 11.474, p = 0.239). There was no effect of different locations on the change in choroidal vascular index (F = 1.979, p = 0.160), and the interaction between AR glasses “on” and “off” states and different locations was also not significant (F = 1.499, p = 0.236). No significant differences were observed in the axial length (F = 0.352, p = 0.559).

[0117] Discussion

[0118] The results of the experimental study indicate that the simulated defocus effect significantly reduces the thinning of the choroidal thickness (ChT) after near reading when the AR glasses are turned on. Although the choroidal vascular index (CVI) and the axial length (AL) do not show statistically significant differences, they exhibit similar trends to the changes in choroidal thickness, suggesting that AR glasses may have some potential in controlling changes in ocular structures.

[0119] In terms of the selection of the defocus region, this study considered the comfort and safety of the visual field and chose to simulate the defocus in the upper region of the retina. This choice was based on existing research, particularly the method proposed by Woodman-Pieterse et al. By simulating the defocus signal in the upper part of the retina, the potential risk of obscuring the lower part of the visual field was avoided, especially in dynamic visual tasks. Studies have shown that the choroidal changes in the temporal and lower regions are more significant when facing accommodation stimuli, as these regions are more susceptible to visual stimuli. The study by Barbara et al. indicates that the peripheral retina (6 to 10 degrees) of myopic individuals is the “best location” for positive defocus signals, which is significant in controlling axial length growth and accommodation development. Therefore, by simulating the peripheral defocus signal, especially in the 1 to 3 mm and 3 to 6 mm regions, this study obtained results consistent with previous studies, i.e., the significant changes in the choroid mainly occur in the temporal and lower regions. These research results further support the feasibility and effectiveness of AR glasses as an intervention tool.

[0120] However, unlike traditional research methods such as watching movies from a distance, the task setting of this study is close reading, which increases the burden on the eyes. Therefore, the results of this study are more significant in the effect of choroidal thickening, and the defocus effect of AR glasses may be more prominent. Compared with the studies of Sander et al. and Zhu et al. using drugs (such as atropine) or corneal orthotic lenses, the defocus effect in this study is not only in the central area, but also in a smaller defocus area (e.g. 1mm central area), the change of choroidal thickness reaches 3.967μm, which means that AR glasses may be superior to other conventional methods in simulating the intensity and effect of defocus signals.

[0121] This study also has certain limitations, especially in the selection of the participant group. The participants in this study are all young adults, while in actual clinical applications, especially in the high myopia population of children and adolescents, the effect of AR glasses may be different. Although young adults respond similarly to defocus signals as children, future studies still need to focus on the differences in response between children and adolescents. At the same time, although this study has used AR glasses equipment that has been safety certified, due to the differences in usage and purpose from traditional glasses or contact lenses, this may affect the general applicability of the experimental results. In order to further verify the long-term effect of AR glasses, future studies need to conduct long-term follow-up investigations to observe the sustained impact of AR glasses on eye structure (such as choroidal thickness and axial length).

[0122] In summary, AR glasses have broad application prospects in myopia prevention and control, especially in slowing down the deepening of myopia. By simulating peripheral defocus signals, AR glasses can reduce the change in choroidal thickness after near work to some extent, which may play an important role in myopia management. With the advancement of technology and further verification of its use effect, AR glasses are expected to become an innovative and effective intervention tool, providing new ideas and solutions for myopia prevention and control.

[0123] The above only describes the preferred embodiments of the present application, and the protection scope of the present application is not limited to the above examples and experimental examples. Any technical solution that follows the concept of the present application is included in the protection scope of the present application. It should be emphasized that for ordinary skilled persons in the art, any modification or equivalent replacement without deviating from the purpose and scope of the present application should be considered as part of the protection scope of the present application.

Claims

1. A multimodal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses, characterized in that, include: Visual stimulation module: Relying on the optical projection capability of AR glasses, personalized stimulation signals are dynamically generated in the 10°–30° peripheral field of vision. The stimulation parameters are adjusted in real time according to the ambient illumination, eye movement characteristics and usage scenario to simulate the outdoor visual experience and enhance the load control of the adjustment system. The field of vision includes three segments: 0°–10° mid-peripheral field of vision, 10°–20° mid-peripheral field of vision, and 20°–30° far-peripheral field of vision. The mid-peripheral field of vision and far-peripheral field of vision are respectively set with contrast gradients of 5%–20% and 20%–50% to differentiate the activation of the ON / OFF pathway. The personalized stimulation signal dynamically generated in the 10°–30° peripheral field of vision is to superimpose a positive defocus pattern of +0.5D to +4.5D or a negative defocus pattern of -0.5D to -4.5D in the 10°–30° peripheral field of vision according to the current refractive power. Visual training module: The central visual field enhances visual convergence and fixation stability by presenting dynamic visual tracking tasks, binocular fusion targets, and contrast sensitivity tests. The system design of the visual training module is a "central-peripheral" visual feedback chain: when the central visual training reaches the preset effect, the intensity and range of peripheral defocus stimulation in the visual stimulation module are automatically and synchronously enhanced, stimulating a stronger peripheral perception response, thereby achieving functional synergy. AI Scene Dynamic Recognition and Targeted Intervention Module: The system integrates a lightweight semantic recognition model to perform real-time image acquisition and intelligent classification of the environment, and comprehensively assesses eye risk by combining light intensity, eye distance and eye movement characteristics. For scenarios with different risk levels, targeted intervention strategies are automatically activated. Adaptive closed-loop feedback module: By collecting key physiological parameters such as blink frequency, pupil diameter change, and fixation stability of the user in real time, the system comprehensively judges the degree of visual fatigue and accommodation load. Based on the feedback information, the system dynamically adjusts the stimulation intensity, frequency, and duration of the three modules: visual stimulation module, visual training module, and directional intervention module, and issues prompts to the user in the form of voice, graphics, or vibration to build a closed-loop control.

2. The multimodal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses according to claim 1, characterized in that, The personalized stimulation signals include a defocus grating with adjustable intensity, a low-contrast perturbation pattern, and spatial frequency stripes.

3. The multimodal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses according to claim 1, characterized in that, The stimulation parameters in the visual stimulation module include defocus intensity, contrast, and spatial frequency.

4. The multimodal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses according to claim 1, characterized in that, The AI ​​scene dynamic recognition and targeted intervention module automatically activates targeted intervention strategies for different risk levels of scenarios, including: increasing display contrast in low-contrast environments; enhancing defocus signals and spatial frequency adjustment during close-range continuous fixation; and adapting high-frequency stimulus compensation in tasks dominated by low-frequency visual stimuli.

5. The multimodal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses according to claim 1, characterized in that, The adaptive closed-loop feedback module provides prompts to the user, including suggestions for resting, adjusting posture, and looking into the distance.

6. The multimodal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses according to claim 2, characterized in that, The personalized stimulation signal is in the form of concentric rings or radial gradients.

7. The multimodal visual stimulation and intelligent adaptive combined myopia prevention and control system based on AR glasses according to claim 1, characterized in that, It also includes a cloud / offline deployment module, which uploads multimodal data to the server in real time, uses distributed AI models for batch calculation and analysis, periodically generates individual and group reports, pushes the latest parameters to devices, and stores data locally and synchronizes it regularly to avoid data loss.

Images