Myopia early warning method and device, storage medium and electronic equipment

Abstract

The present application relates to the technical field of medical health informatics, in particular to a myopia early warning method and device, a storage medium and an electronic device; the method comprises the following steps: through wearing a device integrating a thermal sensor and a nerve conduction film on the face of a user, static acquisition of eye thermal distribution gradient and nerve response time delay data is used to construct an individualized baseline model; dynamic monitoring of thermal field deviation, nerve conduction change and electromyographic disturbance in visual cognitive tasks; combination of a dynamic deviation mapping algorithm and an adaptive weight distribution model to calculate a multi-dimensional risk index, prediction of a myopia-induced fatigue risk level, and matching of a personalized intervention strategy according to user portrait characteristics, to realize closed-loop early warning and intervention optimization. The present application synchronously analyzes physiological signals and behavior data through real-time acquisition, and executes data processing and early warning decision-making in combination with an electronic device and a storage medium, thereby effectively improving the accuracy of early myopia warning and the intervention adaptation degree.

Description

Technical Field

[0001] This invention relates to the field of medical and health informatics technology, specifically to a myopia early warning method and device, storage medium, and electronic device. Background Technology

[0002] In recent years, the incidence of myopia has shown a trend of occurring at younger ages and at higher rates, becoming an important issue in the global public health field. Traditional myopia prevention and control methods mostly rely on vision tests and refractive error monitoring, but these methods have insufficient timeliness in early warning and are difficult to capture potential risks before changes in visual structure occur. With the advancement of biomedical sensing technology, researchers have begun to explore new paths for myopia early warning through physiological parameter monitoring, such as using physiological indicators such as changes in ocular surface blood flow and ciliary muscle activity to establish predictive models.

[0003] Chinese invention patent application CN112700858A discloses a method and device for early warning of myopia in children and adolescents. The method involves receiving current refractive error data from an eye detection device and determining the current refractive state of the subject; receiving facial images from an image acquisition device and identifying the subject's basic user information based on the facial image; acquiring the subject's eye-use behavior data and current eye physiological data based on the subject's current refractive state and facial image; determining an eye refractive state early warning model from several models based on the subject's current refractive state; and inputting the determined eye refractive state early warning model with the acquired basic user information, eye-use behavior data, and current eye physiological data to obtain corresponding early warning information for the subject.

[0004] However, existing technologies mostly focus on isolated analysis of single-modal data (such as thermal imaging or electromyography signals), lacking in-depth exploration of multi-dimensional physiological-functional correlations, resulting in limited accuracy and adaptability of early warning. At the same time, the mechanisms by which visual cognitive load and abnormal neural regulation function play a role in myopia development are gradually being revealed, providing a theoretical basis for achieving early warning through multimodal data fusion analysis. Against this backdrop, there is an urgent need for a comprehensive technical solution that combines multi-source physiological signal acquisition, dynamic modeling, and personalized intervention to improve the sensitivity and timeliness of myopia early warning and provide precise tool support for clinical prevention and control. Summary of the Invention

[0005] The purpose of this invention is to address the problems existing in the background art by proposing a myopia early warning method, device, storage medium, and electronic device.

[0006] The technical solution of this invention: a myopia early warning method, comprising the following specific implementation steps:

[0007] S1. The face-fitting scanning structure equipped with multi-point thermal sensors and a neural conduction membrane is worn on the user's face;

[0008] S2. Collect data on the periocular thermal distribution gradient and neural conduction delay of users in a resting state through multiple rounds of measurement to construct an individualized ocular baseline model;

[0009] S3. By dynamically presenting blurred interference graphics, letter selection and Stroop task stimuli in visual cognitive tasks, real-time collection of thermal field deviation, neural conduction changes and electromyographic disturbance data, constructing multidimensional vector quantitative indicators and calculating the neurocognitive offset rate, and marking potential myopia-induced fatigue response based on dynamic risk thresholds.

[0010] S4. By constructing a multimodal response matrix and fusing thermal nerve shift, neural conduction changes, electromyographic disturbances and cognitive shift data, risk perception weights are dynamically allocated based on modal variation intensity. A comprehensive risk index is calculated by combining a time-series adaptive integral algorithm, a trend slope predictor is introduced to assess the risk evolution trend, and the risk level is mapped to a four-color early warning method for dynamic monitoring and graded response to myopia risk.

[0011] S5. Construct a multi-dimensional feature profile of users based on thermal neural cognitive data, match the best intervention strategy through vector space mapping, dynamically generate training duration, content and push cycle parameters, collect training physiological response in real time and calculate intervention effect evaluation value to assess risk changes, and combine elasticity index closed-loop optimization strategy parameters to carry out personalized myopia intervention adaptive adjustment.

[0012] The preferred process for constructing an individualized ocular baseline model is as follows:

[0013] A two-dimensional thermal gradient field map is generated by collecting the user's resting periorbital thermal distribution through an integrated multi-point thermal sensor array.

[0014] Calculate the average fluctuation intensity of the thermal field within each sampling period, i.e., the local thermodynamic gradient;

[0015] Using a flexible neural conduction sensing membrane integrated into the adhesive structure, the delay and recovery cycle of the neuromuscular response are recorded under the action of micro-electrical stimulation, and the decay function of the neural conduction response over time is recorded, that is, the dynamic recovery of the nerve to micro-electrical stimulation in the resting state.

[0016] The two-dimensional thermal gradient field map, the local thermodynamic gradient, and the decay function of the neural conduction response over time are combined into an initial physiological feedback baseline vector.

[0017] The initial physiological feedback baseline vector is obtained through repeated measurements in three rounds with a five-minute interval between each round. The individual physiological baseline vector is generated by weighted averaging and difference calculation. The consistency of the data from multiple rounds is evaluated by calculating the average deviation of multiple measurements. If the average deviation is below the threshold, the baseline is considered valid; otherwise, the user is prompted to recalibrate the device.

[0018] Preferably, the process for generating the neurocognitive bias rate is as follows:

[0019] S31. A series of visual semantic stimuli sequences are dynamically presented through an embedded cognitive micro-behavioral testing module. The content includes: blurred interference graphics, letter / pattern selection questions, color-text Stroop cognitive conflict task, and each set of stimuli has a specific number s. i and limit the time window ;

[0020] S32, in each s i After the stimulus is presented, within the time window The following multimodal response characteristics are acquired in real time. :

[0021] ;

[0022] in, This indicates the local periocular thermal field deviation relative to the baseline T0; This indicates changes in neural conduction during the stimulation process; This represents the electromyographic micro-activity perturbation curve;

[0023] S33, For each round of testing s i Construct the following multidimensional vector V i :

[0024] ;

[0025] in, Represents the thermal deviation norm; Indicates the total amount of neural response; Indicates the total amount of electromyographic response; Indicates the time taken to complete the cognitive task;

[0026] S34. Construct a response offset function to measure the neurocognitive offset between the current state and the baseline state:

[0027] ;

[0028] in, Indicator of cognitive bias; , , , Indicates the trainable weight coefficients; This indicates the set electromyographic reference value at rest; This represents the average fluctuation intensity of the thermal field within each sampling period; Indicates the baseline completion time; Indicates the initial conduction response strength; This indicates the rate at which the neural response recovers and decays.

[0029] Preferably, the comprehensive risk index is generated as follows:

[0030] S41. Construct the multimodal response sequence matrix R t : ;

[0031] Where n represents the total number of detection rounds;

[0032] S42. Calculate the variation intensity of each modality within the historical time window [tk,t], as the risk perception weight. : ;

[0033] in, This represents the response sequence of mode m within the sliding window; The dynamic risk perception weight of modality m is represented; M represents the total number of modalities. This represents the sample variance, reflecting the degree of fluctuation; k represents the length of the sliding time window, i.e., the number of historical testing rounds.

[0034] S43. Calculate the comprehensive risk index :

[0035] ;

[0036] in, This represents the raw response value of the m-th mode at the current time, i.e., the multimodal response sequence matrix R. t The element in the m-th column and t-th row; This represents the feature mapping function, used to normalize the original modal values; This represents the overall risk score at time t.

[0037] Preferably, the trend slope predictor is:

[0038] ;

[0039] Where η represents the trend slope predictor, i.e. the average slope value of the risk score; Indicates the length of the trend observation window.

[0040] Preferably, the user multidimensional feature profile is as follows:

[0041] ;

[0042] ;

[0043] in, , , , These represent the user's long-term average thermal neural shift, changes in neural conduction, electromyographic disturbances, and cognitive response shift, respectively. This represents the average slope of an upward risk trend. This represents the historical maximum risk value; This represents the risk recovery resilience index; This represents the time interval required for a state to descend from a high-risk state to a safe threshold. This indicates the total intensity of the intervention stimulus; Indicates the magnitude of the decrease in risk value; U i This represents the user profile of the i-th user, that is, the stable individual characteristics exhibited by the i-th user in long-term testing.

[0044] Preferably, the process for generating intervention effect evaluation values ​​is as follows:

[0045] A1. Convert the user profile vector U i Projected onto a predefined intervention strategy space, the best-fitting intervention strategy P is obtained through vector similarity matching. j :

[0046] ;

[0047] Among them, P j This represents the j-th intervention training mode; This represents the transpose of the feature weight vector corresponding to the j-th intervention type; This represents the vector dot product operation, i.e., the similarity between user profiles and intervention strategies;

[0048] A2. Based on the matched intervention strategy P j Dynamically generate individual intervention parameter sets for the i-th user. :

[0049] ;

[0050] in, This indicates the duration of a single training session, which is adaptively set based on the user's fatigue recovery ability. Indicates the function of intervening content; Indicates the difficulty coefficient of intervention, based on the slope of the risk trend. Adjustment; This indicates the push interval period, which is adaptively adjusted based on eye usage frequency and risk mutation trends.

[0051] A3. Initiate intervention training and collect micro-behaviors and physiological responses in real time during the training process to form time-series response data. :

[0052] ;

[0053] in, This represents the instantaneous thermal offset value during the training process; This indicates the real-time ocular electromyographic perturbation response during training; This indicates the change in cognitive stimulus-response time; Indicators representing heart rate variability;

[0054] A4. Compare the warning risk levels before and after training to form an intervention effect assessment value. :

[0055] ;

[0056] in, Indicates the duration of the intervention; and These represent the predicted risk values ​​before and after training, respectively. This indicates the feedback adjustment weight factor; This represents the post-training recovery ability index; Indicates the pre-training recovery ability index;

[0057] like If the value exceeds the set threshold, the training effect is good, and the current strategy will be strengthened; otherwise, the intervention mode P will be updated. j Adjusting training parameters .

[0058] The technical solution of the present invention: a myopia early warning device, used to execute the aforementioned myopia early warning method, comprising:

[0059] The facial sensing module is used to collect thermal signals, electromyographic signals and neural response data around the user's eyes through a flexible semi-circular arc-shaped adhesive structure.

[0060] The neural thermal feedback modeling module is used to build individualized baseline models;

[0061] An embedded cognitive micro-behavioral testing module is used to perform visual stimulus tasks and collect dynamic response data;

[0062] The risk fusion prediction engine module is used to generate a fusion risk index and early warning level;

[0063] The user profiling and personalized intervention module is used to match and execute closed-loop intervention strategies, including providing two levels of heat therapy at 36-42℃ via resistance heating pads.

[0064] The technical solution of the present invention: an electronic device, comprising:

[0065] Memory, used to store computer programs;

[0066] A processor for executing the computer program to implement the method.

[0067] The technical solution of the present invention is a computer storage medium storing a computer program thereon, the computer program being executed by a processor to implement the method described thereon.

[0068] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects:

[0069] This invention designs a myopia early warning method and device, storage medium, and electronic device. Through multimodal physiological data fusion and dynamic modeling technology, it significantly improves the accuracy and intervention adaptability of myopia early warning.

[0070] (1) By adopting the combined monitoring of thermal distribution gradient, neural conduction delay and electromyographic disturbance, the traditional single indicator monitoring limitations are overcome. Through the complementary analysis of multidimensional physiological parameters, early visual fatigue and neural regulation abnormalities are accurately captured, and the sensitivity and specificity of myopia risk identification are improved.

[0071] (2) Based on the individualized baseline model construction and dynamic deviation mapping algorithm, combined with the environmental interference compensation mechanism, the influence of external factors such as environmental temperature and humidity and wearing posture on the monitoring results is effectively reduced, and the clinical applicability is enhanced.

[0072] (3) Introduce a time-adaptive weight allocation and risk evolution trend prediction algorithm to optimize the contribution weight of multimodal data in real time, dynamically adjust the early warning threshold, and achieve accurate prediction from short-term risk response to long-term development trend;

[0073] (4) The closed-loop personalized intervention mechanism matches the optimal intervention strategy with user profiles and dynamically adjusts the training intensity and frequency by combining the elastic recovery index, forming a complete closed loop of monitoring-early warning-intervention-feedback, which significantly improves user compliance and intervention effect;

[0074] (5) The flexible sensor array and the host are designed separately to balance wearing comfort and signal acquisition accuracy, providing hardware support for long-term continuous monitoring. Attached Figure Description

[0075] Figure 1 This is a flowchart of a myopia early warning method proposed in this invention;

[0076] Figure 2 This is a schematic diagram of the structure of a myopia early warning device proposed in this invention;

[0077] Figure 3 This is a schematic diagram of the structure of an electronic device proposed in this invention. Detailed Implementation

[0078] Example 1, as Figure 1 As shown, the myopia early warning method proposed in this invention includes the following specific implementation steps:

[0079] S1. The face-fitting scanning structure, equipped with multi-point thermal sensors and a neural conduction membrane, is worn on the user's face.

[0080] S2. Under the user's resting state, collect the thermal distribution gradient number T0(x,y) around the eyes (including but not limited to the upper eyelid, brow bone, and cheekbone area) and the neural conduction response delay data C0(t). Extract and construct an individual-specific eye thermal response-neural activity baseline model. The specific implementation process is as follows:

[0081] S21. By integrating a multi-point micro-thermal sensor array into the face-fitting structure, the thermal field distribution data around the eyes is collected in the user's resting state (eyes closed for 10 seconds), forming a two-dimensional thermal gradient field map:

[0082] ;

[0083] in, This represents a spatial distribution map formed by temperature sampling in key areas (including but not limited to the eyelids, brow ridges, inner canthus, and outer canthus). Specifically, it represents the initial temperature values ​​measured at a specific two-dimensional location (x, y) in the eye, which together form the temperature distribution map of the entire eye region.

[0084] S22. Calculate the average fluctuation intensity of the thermal field within each sampling period, i.e., the local thermodynamic gradient, which reflects the activity of physiological microcirculation: ;

[0085] in, It is expressed as the rate of change of thermal gradient per second (gradient operator), which is calculated by the Sobel operator through the difference of the heat map. It is the gradient of temperature at spatial position (x,y) in the i-th frame heat map, that is, the spatial partial derivative of the rate of change of temperature; N represents the number of heat map frames collected in the resting state, that is, the total number of frames in the sampling period. This represents the average fluctuation intensity of the thermal field within each sampling period, i.e., the average gradient intensity of the periocular thermal field within the sampling period, reflecting the fluctuation of ocular microcirculation activity.

[0086] S23. Using a flexible neural conduction sensing membrane integrated into the adhesive structure, under the action of micro-electrical stimulation (below the human perception threshold), the delay and recovery period of the neuromuscular response are recorded, and the following time response function is obtained: ;

[0087] in, It represents the decay function of the neural conduction response over time, describing the dynamic recovery of the nerve to micro-electrical stimulation in the resting state; This indicates the initial conduction response strength, representing the initial activity level of the nerve in response to micro-electrical stimulation; a larger value indicates stronger nerve activity. The recovery decay rate of the neural response is represented by t, which is the speed at which the nerve recovers from the activated state to the resting state; t represents the time variable, in seconds, from the start of the stimulus.

[0088] S24. Combine the above thermal field model and conduction function into a set of individual physiological feedback baseline vectors. : ;

[0089] in, Indicates the baseline completion time;

[0090] S25. To eliminate the influence of ambient temperature, wearing posture, and other disturbances, three rounds of measurement are performed by default (with a 5-minute interval between each round), and each measurement generates a raw vector. Then, element-wise weighted averaging and difference calculations are performed on all measurement vectors to form the final individual physiological feedback baseline vector M. base : Its stability is evaluated using the following consistency formula: ;

[0091] like If the baseline is successfully established, then the user is prompted to re-wear the device and adjust the facial fit angle.

[0092] in, This represents the mean steady-state thermal field distribution, which is the average value of the spatial thermal distribution image of the surface temperature of the key area of ​​the human face and eyes during the three rounds of measurement. This represents the average rate of change in eye temperature during the stimulus-response cycle. This represents the average neural conduction-thermal response curve of an individual under standard stimulation; It represents the average deviation of the baseline model and measures the consistency and stability of multiple rounds of measurements; This represents the preset stability threshold, used to determine whether the baseline model is sufficiently stable and reliable; n represents the number of measurements, which is 3 in this embodiment. This indicates the baseline completion time, and the values ​​from multiple rounds of static testing are all output.

[0093] S3. A micro-behavioral response analysis method based on the thermo-neural-cognitive linkage mechanism. Through embedded micro-expression perturbation testing, it collects the individual's thermo-feedback neural response, electromyographic changes, and attentional shifts in visual cognitive tasks, and compares them with the physiological baseline model M established in step S2. base Dynamic bias mapping was performed to reveal signs of neural fatigue induced by early myopia load. The specific implementation process is as follows:

[0094] S31. A series of visual semantic stimulus sequences are dynamically presented through an embedded cognitive micro-behavioral testing module, including:

[0095] Blurred interference pattern (rapid alternation of low-resolution);

[0096] Letter / pattern multiple choice questions;

[0097] Color-text Stroop cognitive conflict task;

[0098] Each stimulus group has a specific number s. i and limit the time window ;

[0099] S32, in each s i After the stimulus is presented, within the time window The following multimodal response characteristics are acquired in real time. :

[0100] ;

[0101] in, This indicates the local periocular thermal field deviation relative to the baseline T0; This indicates changes in neural conduction during the stimulation process; This represents the perturbation curve of electromyographic microactivity (mainly the orbicularis oculi and levator palpebrae superioris muscles);

[0102] S33, For each round of testing s i Construct the following multidimensional vector V i :

[0103] ;

[0104] in, It represents the thermal deviation norm, which measures the degree of local thermal change; It represents the total amount of neural response and reflects the accumulation of fatigue; It represents the total amount of electromyographic response, revealing involuntary muscle tension; It indicates the time to complete a cognitive task and is used to measure attentional stability;

[0105] S34. Construct a response offset function to measure the neurocognitive offset between the current state and the baseline state:

[0106] ;

[0107] in, Indicator of cognitive bias; , , , Indicates the weight coefficients used in training; This indicates the set electromyographic reference value at rest;

[0108] S35. Set a dynamic risk threshold θ, if If the test result is positive, then the test round is marked as a potential myopia-induced fatigue response. ;

[0109] in, This represents the myopia-induced fatigue response marker in the i-th round of testing.

[0110] S4. Construct a risk fusion method that links oculomotor thermal feedback, cognitive micro-behavioral shift, and dynamic threshold modeling. Through the fusion modeling of previously collected multimodal physiological and behavioral data, a time-adaptive multi-channel risk integration algorithm and trend prediction program are designed. The specific implementation process is as follows:

[0111] S41. Obtain the feature responses from each round of detection from steps S2 and S3, and construct the following multimodal response sequence matrix R. t : ;

[0112] Where n represents the total number of detection rounds;

[0113] It should be noted that the multimodal response sequence matrix R t Each column corresponds to thermal nerve offset, changes in nerve conduction, electromyographic disturbances, and cognitive response offset.

[0114] S42. Construct a time-dependent modality importance assignment function, calculate the variation intensity of each modality in the historical time window [tk,t], and use it as the risk perception weight. :

[0115] ;

[0116] in, This represents the response sequence of mode m within the sliding window; The dynamic risk perception weight of modality m is represented; M represents the total number of modalities, which is set to 4 in this embodiment. The variance represents the sample variance, reflecting the degree of fluctuation; k represents the length of the sliding time window (number of historical testing rounds).

[0117] S43. Calculate the comprehensive risk index by combining modal weights and characteristic responses. :

[0118] ;

[0119] in, This represents the raw response value of the m-th mode at the current time, i.e., the multimodal response sequence matrix R. t The element in the m-th column and t-th row; This represents the feature mapping function, used to normalize the original modal values; This represents the overall risk score at time t, which is the individual's current myopia risk level.

[0120] S44. Introduce the trend slope predictor η to determine whether the risk continues to rise, and construct the following risk evolution trend function:

[0121] ;

[0122] Where η represents the trend slope predictor, i.e. the average slope value of the risk score; Indicates the length of the trend observation window;

[0123] S45, Risk Index Mapped to warning levels, the warning level table is shown in Table 1:

[0124] Table 1 Warning Level Table

[0125] .

[0126] S5. User profiles are constructed using thermal-neural-cognitive trimodal feedback information. Intelligent adaptation of intervention types is achieved through vector space mapping. Intervention parameters are determined by user physiological and behavioral trends. A closed-loop effect evaluation mechanism is introduced to support dynamic adjustment of training content. The specific implementation process is as follows:

[0127] S51. Constructing a user feature profile vector U based on previous steps. i That is, the stable individual characteristics exhibited by the i-th user over long-term detection:

[0128] ;

[0129] ;

[0130] in, , , , These represent the user's long-term average thermal neural shift, changes in neural conduction, electromyographic disturbances, and cognitive response shift, respectively. This represents the average slope of an upward risk trend. This represents the historical maximum risk value; This represents the risk recovery resilience index, which measures the ability to recover. This represents the time interval required for a state to descend from a high-risk state to a safe threshold. This indicates the total intensity of the intervention stimulus (including but not limited to the total energy of heat application and the weighted value of training intensity). Indicates the magnitude of the decrease in the risk value;

[0131] It should be noted that the risk recovery resilience index By introducing two dimensions, "unit intervention effect" and "recovery speed," a longitudinal dynamic feedback mechanism is formed, supporting the system to adaptively adjust the intervention frequency based on individual recovery capabilities during training feedback; if High response indicates a positive user response to training, and the frequency of intervention can be appropriately reduced; if If the level is low, the intervention intensity will be increased or the intervention type will be changed; in the closed-loop assessment, It can serve as a "resilience coefficient" for dynamic health trends;

[0132] S52, Convert the user profile vector U i Projected onto a predefined intervention strategy space, the best-fitting intervention strategy P is obtained through vector similarity matching. j :

[0133] ;

[0134] Among them, P j This represents the j-th type of intervention training mode (including but not limited to visual relaxation training and optic nerve heat therapy coordination training). The transpose of the feature weight vector corresponding to the j-th intervention type is derived from a large training dataset; This represents the vector dot product operation, i.e., the similarity between user profiles and intervention strategies;

[0135] S53. Based on the matched intervention strategy P j Dynamically generate individual intervention parameter sets for the i-th user. :

[0136] ;

[0137] in, This indicates the duration of a single training session, which is adaptively set based on the user's fatigue recovery ability. The intervention content function (including but not limited to visual-motor tracking training and dynamic thermal stimulus response training) is used. Indicates the difficulty coefficient of intervention, based on the slope of the risk trend. Adjustment; This indicates the push interval period, which is adaptively adjusted based on eye usage frequency and risk mutation trends.

[0138] S54. Initiate intervention training and collect micro-behaviors and physiological responses in real time during the training process to form time-series response data. : ;

[0139] in, This represents the instantaneous thermal offset value during the training process; This indicates the real-time ocular electromyographic perturbation response during training; This indicates the change in cognitive stimulus-response time; It represents an index of heart rate variability, reflecting the state of autonomic nervous system regulation;

[0140] S55. Compare the warning risk levels before and after training to form an intervention effect assessment value. :

[0141] ;

[0142] in, Indicates the duration of the intervention; and These represent the predicted risk values ​​before and after training, respectively. This indicates the feedback adjustment weight factor; This represents the post-training recovery ability index; Indicates the pre-training recovery ability index;

[0143] like If the value exceeds the set threshold, the training effect is good, and the current strategy will be strengthened; otherwise, the intervention mode P will be updated. j Or adjust training parameters This enables closed-loop optimization of individual intervention pathways.

[0144] Example 2, as Figure 2 As shown, the present invention proposes a myopia early warning device, which is used to execute a myopia early warning method proposed in Embodiment 1, comprising:

[0145] Facial sensing module: used for non-contact acquisition of thermal signals, facial electromyography signals and surface nerve responses in the user's eye area; adopts a flexible semi-circular arc-shaped attachment structure that conforms to the structure of the human face, with the main unit and scanning head separated to improve wearing comfort and physiological signal fit; internally integrated including but not limited to: high-frequency thermal array sensor, sEMG orbicularis oculi muscle array electrode, and high-frequency micro nerve potential sensing.

[0146] The neural thermal feedback modeling module: embeds an individual modeling chip and an edge modeling algorithm unit to construct an individual ocular neural thermal feedback baseline model;

[0147] Embedded cognitive micro-behavioral testing module: includes a miniature OLED visual stimulation display component; an embedded fast response detection sensor; combines thermal-electromyographic-neural response data to perform micro-stimulation-response delay calculation; supports testing tasks such as low-frequency flicker, dot matrix offset, and complex trajectory recognition.

[0148] Risk fusion prediction engine module: Built-in integrated data fusion prediction processor; performs the following operations: multimodal data joint modeling; risk trend fitting and risk index generation; peak risk identification and rebound risk assessment;

[0149] User profiling and personalized intervention module: Constructs multi-dimensional feature vectors for users and performs intervention pattern matching; integrates graph neural network (GNN) models for training content generation; and supports multiple types of intervention methods (including but not limited to simultaneous hot compress training, eye tracking, and cognitive stimulation modulation); automatically obtains feedback and updates the intervention path after intervention is executed;

[0150] It should be noted that the heat therapy training in this embodiment uses a flexible resistive heating pad. Closed-loop temperature control is achieved through a temperature sensor and an MCU (microcontroller unit) control circuit, and is synchronized in real time with neural conduction response data. When resistive heating is initiated, the system increases the temperature in stages based on the user's baseline model completion time. In other words, the heat therapy function is implemented by an independent resistive heating pad, physically isolated from the thermistor. The heat therapy temperature is precisely controlled between 36-42℃ using a PID algorithm, synergistically enhancing the intervention effect with cognitive stimulation.

[0151] Normal mode: constant temperature 36℃ (±0.5℃ fluctuation);

[0152] Enhanced mode: Heat up to 42℃ (lasts ≤5 minutes);

[0153] When the temperature in the user's temporal artery region is detected to be >37.5℃, the power supply to the heating pad is immediately cut off and cognitive stimulation is stopped.

[0154] Example 3: An electronic device, such as Figure 3 As shown, it includes:

[0155] processor;

[0156] Memory, used to store computer programs;

[0157] Specifically, a computer program includes program code, which includes computer operation instructions; memory includes high-speed RAM and non-volatile memory, such as at least one disk storage device.

[0158] The processor is used to execute computer execution instructions stored in the memory to implement the technical solution of the myopia early warning method embodiment proposed in Embodiment 1;

[0159] The processor is a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0160] Optionally, the electronic device also includes a communication interface to communicate and interact with external devices, such as user terminals (e.g., mobile phones, tablets). In specific implementations, if the communication interface, memory, and processor are implemented independently, the communication interface, memory, and processor can be interconnected through a bus to complete communication between them.

[0161] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc.; buses are divided into address buses, data buses, control buses, etc.

[0162] Optionally, in a specific implementation, if the communication interface, memory, and processor are integrated on a single chip, then the communication interface, memory, and processor can communicate through an internal interface.

[0163] Example 4: This example also provides a computer storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the above-mentioned technical solution of a myopia early warning method.

[0164] In one possible implementation, the computer-readable medium may include random access memory (RAM), read-only memory (ROM), compact discread-only memory (CD-ROM) or other optical disc storage, disk storage or other magnetic storage device, or any other medium on which the desired program code is intended to be carried or stored in the form of instructions or data structures and is accessible by a computer; and any connection thereof is properly referred to as a computer-readable medium.

[0165] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A myopia early warning device, characterized in that, The method comprises the following steps: A face perception module is used to collect eye thermal signal, electromyographic signal and neural response data through a flexible semicircular arc attachment structure; A neural thermal feedback modeling module is used to build an individualized baseline model; An embedded cognitive micro-behavior test module is used to perform visual stimulation tasks and collect dynamic response data; A risk fusion prediction engine module is used to generate a fusion risk index and a warning level; A user portrait and individualized intervention module is used to match and execute a closed-loop intervention strategy, including providing 36-42℃ two-grade hot compress treatment through a resistance heating sheet; The working method of the myopia early warning device comprises the following specific implementation steps: S1. A face fitting scanning structure equipped with a multi-point thermal sensor and a nerve conduction film is worn on the face of a user; S2. The eye thermal distribution gradient and nerve conduction time delay data of the user in a resting state are collected through multiple measurements to build an individualized eye baseline model; S3. Fuzzy interference patterns, letter selection and Stroop task stimuli are dynamically presented in visual cognitive tasks to collect thermal field deviation, nerve conduction change and electromyographic disturbance data in real time, build a multi-dimensional vector quantization index and calculate a neural cognitive deviation rate, and mark potential myopia-induced fatigue response based on a dynamic risk threshold; S4. A multi-modal response matrix is built to fuse thermal field deviation, nerve conduction change, electromyographic disturbance and cognitive deviation data, a risk perception weight is dynamically assigned based on modal variation intensity, a comprehensive risk index is calculated by combining a time series adaptive integral algorithm, a trend slope prediction factor is introduced to evaluate the risk evolution trend, and the risk level is mapped to a four-color warning method for dynamic monitoring and grading response of myopia risk; S5. A user multi-dimensional feature portrait is built based on thermal and neural cognitive data, the best intervention strategy is matched through vector space mapping, training duration, content and push cycle parameters are dynamically generated, training physiological responses are collected in real time and intervention effect evaluation values are calculated to evaluate risk changes, and strategy parameters are optimized in a closed loop combined with an elastic index to adaptively adjust individualized myopia intervention.

2. The myopia warning device according to claim 1, wherein The individualized eye baseline model is built as follows: A two-dimensional thermal gradient field map is generated by integrating a multi-point thermal sensor array to collect the resting eye thermal distribution of a user; The average fluctuation intensity of the thermal field in each sampling period, i.e. the local thermal dynamic gradient, is calculated; Under the action of micro-electric stimulation, the delay and recovery period of the neuromuscular response are recorded using the flexible nerve conduction sensing film integrated in the fitting structure, and the attenuation function of the nerve conduction response over time, i.e. the recovery dynamics of the nerve to micro-electric stimulation in a resting state, is recorded; The two-dimensional thermal gradient field map, the local thermal dynamic gradient and the attenuation function of the nerve conduction response over time are combined into an initialized physiological feedback baseline vector; The initialized physiological feedback baseline vector is obtained through 3 rounds of repeated measurements with an interval of 5 minutes, an individual physiological baseline vector is generated through weighted averaging and difference calculation, the consistency of multiple measurements is evaluated by calculating the average deviation, and if the average deviation is below a threshold, the baseline is determined to be valid, otherwise the user is prompted to recalibrate the device.

3. The myopia warning device according to claim 2, wherein The neural cognitive deviation rate is generated as follows: S31, a series of visual semantic stimulus sequences are dynamically presented through the embedded cognitive micro-behavior test module, including: fuzzy interference graphics, letter and pattern selection questions, color-text Stroop cognitive conflict tasks; Each set of stimuli has a clear number s i and a defined length of time ; S32, in each set of stimuli s i Presentations are followed by a length of time The following multi-modal response features are collected in real-time within a length of time : ； wherein, represents the local periorbital thermal field deviation relative to the baseline TO; represents the neurostimulation change during the stimulation; represents the electromyographic microactivity perturbation curve; S33, for each round of testing s i , is constructed as follows a multidimensional vector V i : ； wherein, denotes the thermal deviation norm; denotes the total amount of neural responses, i.e. ; denotes the total amount of myoelectric responses, i.e. ; denotes the cognitive task completion time; S34, a response offset rate function is constructed to measure the neuro-cognitive offset rate of the current state and the baseline state: ； wherein, represents a cognitive bias degree index; , , , represents a trainable weight coefficient; represents a set reference value of myoelectricity in a resting state; represents an average fluctuation intensity of a thermal field in each sampling period; represents a baseline completion duration; represents an initial conduction response intensity; represents a recovery decay rate of a neural response.

4. The myopia warning device according to claim 3, wherein The comprehensive risk index generation process is as follows: S41, construct a multi-modal response sequence matrix R: ; Wherein, n represents the total number of detection rounds, i=1, 2, …, n; S42, calculate the variability strength of each modality in the historical time window [i-k, i] as the risk perception weight : ; wherein, represents the response sequence of the modality m within the sliding window; represents the dynamic risk perception weight of the modality m; M represents the total number of modalities; represents the sample variance; k represents the length of the sliding time window, i.e., the number of historical detection rounds; S43, calculate the comprehensive risk index: ； wherein, represents the original response value of the mth modality in the ith round of detection; represents a feature mapping function for normalizing the modality original value; represents the comprehensive risk score of the ith round of detection.

5. The myopia warning device according to claim 4, wherein The trend slope prediction factor is: ； wherein η denotes a trend slope predictor, i.e. the average slope value of the risk score; denotes the length of the trend observation window.

6. The myopia warning device according to claim 5, wherein The user multi-dimensional feature portrait is: ； ； wherein, , , , respectively represent the user long-term average thermal field deviation, the nerve conduction change, the electromyographic disturbance, and the cognitive response shift; represents the average slope of the risk trend; represents the historical maximum risk value; represents the risk recovery elasticity index; represents the time interval required to drop from a high-risk state to a safe threshold; represents the total amount of intervention stimulation intensity; represents the risk value drop amplitude; represents the user portrait of the th user, i.e., the stable individual characteristics exhibited by the th user in the long-term detection.

7. The myopia warning device according to claim 6, wherein The intervention effect evaluation value generation process is: A1, projecting the user profile vector into a predefined intervention strategy space, obtaining the best adapted intervention strategy by vector similarity matching : ； wherein, represents the first intervention training mode; represents the transpose of the feature weight vector corresponding to the first intervention type; represents a vector dot product operation, i.e., the similarity of the user portrait and the intervention strategy; A2. According to the matched intervention strategy , the first user is dynamically generated an individual intervention parameter set : ; wherein, represents the intervention duration, which is set adaptively according to the user's fatigue recovery ability; represents the intervention content function; represents the intervention difficulty coefficient, which is based on the average slope of the risk trend adjustment; represents the push interval period, which is adaptively adjusted in combination with the eye use frequency and the risk mutation trend; A3, start intervention training, and collect micro-behaviors and physiological responses in the training process in real time to form time series response data; A4, compare the early warning risk level before and after training to form an intervention effect evaluation value : ； wherein, represents the duration of the intervention; and represent the integrated risk score before and after training, respectively; represents the feedback regulation weight factor; represents the risk recovery elasticity index after training; represents the risk recovery elasticity index before training; If If the training effect is good, the current strategy will be strengthened; otherwise the intervention mode will be updated , adjust the whole body intervention parameter set .

8. An electronic device, comprising: Comprise: A memory for storing a computer program; A processor for executing the computer program to control the device of any one of claims 1-7 to run.

9. A computer storage medium, characterized in that A computer program is stored thereon, which is executed by a processor to configure the device of any one of claims 1-7.

Images