Method and system for assessing hyperopic risk in children based on multidimensional data

Abstract

The application discloses a hyperopia risk assessment method and system for children based on multidimensional data, relates to the technical field of intelligent health risk assessment, and comprises the following steps: collecting behavior data of children for preprocessing, constructing a sample set, performing window division on the sample set, and generating risk factors according to a mapping threshold to form a risk sequence; based on the risk sequence, statistical characteristics are calculated, and based on the statistical characteristics, a likelihood ratio value is obtained; whether the risk sequence has a behavior mutation is judged according to the likelihood ratio value; if there is no mutation, the risk factors are weighted and fused by setting a weight, a risk index is obtained, and if a mutation is detected, the risk factors are defined as arms in a multi-arm tiger machine algorithm, and adaptive weight adjustment and feedback are performed to generate the risk index; and the scientificity, intelligence and individualization level of the hyperopia risk assessment for children are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent health risk assessment technology, and in particular to a method and system for assessing the risk of farsightedness in children based on multidimensional data. Background Technology

[0002] In recent years, research on refractive developmental abnormalities in children and adolescents and myopia prevention and control has continued to deepen. Clinical and public health fields have gradually formed an early risk management concept centered on "hyperopia reserve," which involves identifying the depletion trend of hyperopia reserve in children at different ages to predict the probability of myopia and intervene in advance. Against this backdrop, vision risk assessment technologies for everyday life scenarios are gradually evolving from traditional offline optometry screening to "continuous monitoring and dynamic assessment driven by multi-dimensional behavioral data." With the maturity of wearable sensors, edge computing, and the mobile internet, using light sensors to collect ambient illuminance, using nine-axis inertial measurement units to collect head posture angles, and using eye trackers or fixation detection modules to collect eye-use behavior data have become important means to objectively quantify children's eye-use behavior. At the same time, algorithms such as sliding window analysis, threshold mapping risk modeling, statistical distribution parameter estimation, mutation point detection, and online learning are gradually being introduced into health behavior assessment scenarios. This allows risk assessment to not only output single static results but also form time-series risk indices and individualized intervention bases, thus providing more frequent and interpretable decision support for families, schools, and medical institutions.

[0003] Existing technologies often use a single indicator or simple linear rules as the basis for risk assessment, lacking a unified model of the coupling relationship between multiple sources of data such as "lighting, posture, and eye usage time". This results in insufficient adaptability of risk assessment results to complex real-world scenarios. Furthermore, the use of fixed thresholds and fixed weights to weight and fuse risk factors makes it difficult to reflect the non-stationary changes in children's behavior patterns under different time periods, learning tasks, or environmental conditions. In particular, when children experience short-term but significant changes in eye usage behavior (such as sudden prolonged close-range reading, rapid deterioration of posture, or sudden changes in lighting environment), traditional static models often fail to respond in a timely manner, which can easily lead to a lag or misjudgment of the risk index. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method and system for assessing the risk of farsightedness in children based on multidimensional data, which solves the problems of insufficient adaptability of existing risk assessment results to complex real-world scenarios and the tendency to cause lag or misjudgment of risk indices.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for assessing the risk of farsightedness in children based on multidimensional data, comprising, Children's behavioral data is collected and preprocessed to construct a sample set. The sample set is then divided into windows, and risk factors are generated based on mapping thresholds to form a risk sequence. Based on the risk sequence, statistical features are calculated, and based on the statistical features, the likelihood ratio is obtained. The likelihood ratio is used to determine whether there is a behavioral mutation in the risk sequence. If there is no mutation, weights are set to weight and fuse the risk factors to obtain a risk index for risk assessment. If a mutation is detected, the risk factor is defined as an arm in the multi-armed slot machine algorithm, and adaptive weight adjustment and feedback are performed. The risk index is stored in a database.

[0007] As a preferred embodiment of the multidimensional data-based method for assessing the risk of farsightedness in children according to the present invention, the step of dividing the sample set into windows and generating risk factors based on mapping thresholds to form a risk sequence includes: The sample set is divided into windows using sliding window technology. Based on ophthalmological recommendations, requirements for healthy reading and writing postures for students, and guidelines for myopia prevention and control in children and adolescents, mapping thresholds are set for each data in the sample set. Each data in the sample set is mapped to a risk factor through the mapping thresholds, and the risk factors of illumination, posture, and eye use duration at each time in each window are obtained respectively. Based on the risk factors of illumination, posture, and eye use duration at each moment within each window, any type of risk factor is combined in chronological order to form three risk sequences for the window, including illumination, posture, and eye use duration risk sequences.

[0008] As a preferred embodiment of the multidimensional data-based method for assessing the risk of farsightedness in children according to the present invention, the step of calculating statistical characteristics based on a risk sequence, obtaining a likelihood ratio based on the statistical characteristics, and determining whether there is a behavioral mutation in the risk sequence based on the likelihood ratio includes: The mean and variance of the risk series within any window are calculated using the maximum likelihood estimation method. Based on the mean and variance, a parameter vector for each window is constructed, and then the probability density function is used to obtain the probability density function value of the risk factor under each window. Each time point in the window is traversed as a candidate mutation point. The current window is then divided into two equally spaced sub-windows, and the mean, variance, and probability density values ​​of the two sub-windows are obtained again. Construct a statistical function and calculate the likelihood ratio for all mutation points; Iterate through the current window and select the largest likelihood ratio for mutation detection; If the likelihood ratio is greater than the mutation detection threshold If the behavior of the current risk factor changes within the window, it is determined that a behavioral mutation has occurred and is marked as 1; otherwise, it is determined that no behavioral mutation has occurred within the window and is marked as 0.

[0009] As a preferred embodiment of the multidimensional data-based method for assessing the risk of farsightedness in children according to the present invention, wherein: if there is no mutation, weights are set to weight and fuse the risk factors to obtain a risk index for risk assessment, including: Regression analysis is performed based on historical data to set the weights of each risk factor. If the result is 0, a weighted fusion is performed based on the weights and each risk factor to obtain the risk index. Assessment thresholds are set based on ophthalmological recommendations and knowledge in related fields. and ,and When the risk index is less than the assessment threshold If the child's current farsightedness reserve is sufficient, then the child is at low risk. When the risk index is greater than or equal to the assessment threshold And less than the evaluation threshold If this occurs, it indicates that the child's farsightedness reserve is being depleted too quickly, requiring attention to their eye-use behavior, and placing them at a medium risk. When the risk index is greater than or equal to the assessment threshold If the reading is negative, it indicates that the child currently has insufficient farsightedness reserve, is at risk of myopia, and is at high risk.

[0010] As a preferred embodiment of the multidimensional data-based method for assessing the risk of farsightedness in children according to the present invention, wherein: if a mutation is detected, the risk factor is defined as an arm in a multi-armed slot machine algorithm, and adaptive weight adjustment and feedback are performed, including: If the labeling result is 1, then each risk factor is defined as an arm in the multi-armed slot machine algorithm, resulting in a set of arms, including the illumination risk factor arm, posture risk factor arm, and eye usage duration risk factor arm. A complementary approach is used to define the instantaneous reward function for each arm and calculate the instantaneous reward function value. Based on the label result 1, the reset rule is executed to generate initial policy state variables for each arm, including the number of updates, the estimated historical average reward, and the reward history set; Based on the number of updates and the historical average reward estimate, a confidence upper bound formula is constructed for each arm, and the confidence upper bound value is obtained by solving the confidence upper bound formula through the bisection method. Based on the confidence upper bound, Softmax normalization is used to calculate the weight of each arm, which serves as the weight of the risk factor corresponding to each arm. Based on the weights, the results are fed back into the risk index weighted fusion formula to solve for the risk index.

[0011] As a preferred embodiment of the multidimensional data-based method for assessing the risk of farsightedness in children according to the present invention, the step of storing the risk index in a database includes: The risk index is stored in a database, and during the storage process, the corresponding child's ID is added to the risk index.

[0012] As a preferred embodiment of the multidimensional data-based method for assessing the risk of farsightedness in children according to the present invention, the step of collecting children's behavioral data and preprocessing it to construct a sample set includes: For each child, data was collected using a light sensor, a nine-axis inertial measurement unit, and an eye tracker. The behavioral data of each child is denoised and standardized to construct a sample set for each child; The behavioral data includes light intensity data, attitude angle data, and eye usage time data. The attitude angle data includes pitch angle and yaw angle.

[0013] Secondly, this invention provides a children's farsightedness risk assessment system based on multidimensional data, including, The data acquisition and mapping module is used to collect children's behavioral data, preprocess it, construct a sample set, divide the sample set into windows, and generate risk factors based on the mapping threshold to form a risk sequence. The calculation and judgment module is used to calculate statistical characteristics based on the risk sequence, obtain the likelihood ratio based on the statistical characteristics, and judge whether there is a behavioral mutation in the risk sequence based on the likelihood ratio. The execution generation module is used to set weights to perform weighted fusion of risk factors to obtain a risk index for risk assessment. If a mutation is detected, the risk factor is defined as an arm in the multi-armed slot machine algorithm, and adaptive weight adjustment and feedback are performed. The storage module is used to store the risk index in a database.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the method for assessing the risk of farsightedness in children based on multidimensional data as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for assessing the risk of farsightedness in children based on multidimensional data as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By introducing the generalized likelihood ratio mutation detection method, key turning points in behavioral patterns can be identified. Combined with the multi-armed slot machine algorithm, the weighting coefficients of risk factors in each dimension are dynamically adjusted to achieve intelligent adaptive fusion of the risk index. Secondly, the output hyperopia risk index has the advantages of interpretability and real-time updating capability, which can help parents and doctors track the trend of hyperopia reserve changes in children. Therefore, this invention significantly improves the scientific, intelligent and individualized level of hyperopia risk assessment in children. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the method for assessing the risk of farsightedness in children based on multidimensional data in Example 1.

[0019] Figure 2 This is a structural diagram of the children's farsightedness risk assessment system based on multidimensional data in Example 1. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1, referring to Figures 1-2 This is the first embodiment of the present invention, which provides a method for assessing the risk of farsightedness in children based on multidimensional data, including the following steps: S1. Collect children's behavioral data, preprocess it, construct a sample set, divide the sample set into windows, and generate risk factors based on the mapping threshold to form a risk sequence; Specifically, children's behavioral data is collected, preprocessed, and a sample set is constructed, including: For each child, data was collected using a light sensor, a nine-axis inertial measurement unit, and an eye tracker. Denoising and standardizing behavioral data from 2 or 5 days (e.g., 2 or 5 days) is performed to construct a sample set for each child; The behavioral data includes light intensity data, attitude angle data, and eye usage time data. The attitude angle data includes pitch angle and yaw angle.

[0024] By collecting multidimensional behavioral data on children (including light intensity, posture angle, and screen time) and performing data preprocessing (denoising and standardization), the accuracy and comparability of the data were effectively improved. The preprocessed data provides a reliable foundation for subsequent assessment of children's farsightedness risk. By constructing a behavioral sample set for each child, it is possible to comprehensively reflect their behavioral patterns in daily activities and promptly identify potential vision risk factors.

[0025] Furthermore, the sample set is divided into windows, and risk factors are generated based on mapping thresholds to form a risk sequence, including: The sample set is divided into windows using sliding window technology. Based on ophthalmological recommendations, requirements for healthy reading and writing postures for students, and guidelines for myopia prevention and control in children and adolescents, mapping thresholds are set for each data in the sample set. Each data in the sample set is mapped to a risk factor through the mapping thresholds, and the risk factors of illumination, posture, and eye use duration at each time in each window are obtained respectively. The mapping formula for the light risk factor is: In the formula, Indicates at time Light risk factors at that time This indicates the risk value corresponding to the high-risk level. This indicates the risk value corresponding to the medium-risk level. This indicates the risk value corresponding to the low-risk level. Indicates at time Standardized light intensity value at time, , , These represent the low light threshold, medium light threshold, and high light threshold, respectively (set according to ophthalmological recommendations; default values ​​can be set to 150 lux, 300 lux, and 500 lux). , Indicates when...; It should be noted that the risk values ​​for high, medium, and low risk levels can be determined by reviewing clinical literature and fitting the statistical mean of samples labeled by ophthalmologists. The range of each risk value can be set as follows: 0.6 to 1.0 for high risk level, 0.3 to 0.6 for medium risk level, and 0 to 0.3 for low risk level. For the present invention, the risk value for high risk level can be 0.8, the risk value for medium risk level can be 0.5, and the risk value for low risk level can be 0.1. The mapping formula for the attitude risk factor is: In the formula, Indicates at time Posture risk factors at the time Indicates at time pitch angle at time Indicates at time Yaw angle at time This represents the absolute value operation. This indicates the operation of retrieving the maximum value. , , These represent the low-risk, medium-risk, and high-risk thresholds for posture angles, respectively (set according to students' healthy reading and writing posture requirements, with preferred values ​​of 15 degrees, 30 degrees, and 45 degrees). ); The mapping formula for the eye use duration risk factor is: In the formula, Indicates at time Risk factors related to screen time duration Indicates the duration of eye use. , , These represent the low-risk, medium-risk, and high-risk thresholds for screen time, respectively (set according to the Guidelines for Myopia Prevention and Control in Children and Adolescents, with preferred values ​​of 20, 30, and 40 minutes, respectively). ); Based on the risk factors of illumination, posture, and eye use duration at each moment within each window, any type of risk factor is combined in chronological order to form three risk sequences for the window, including illumination, posture, and eye use duration risk sequences.

[0026] The assessment of light risk factors allows for precise analysis of children's eye-use risks under different lighting conditions, especially in environments with varying light intensities, such as classrooms and outdoors. By real-time monitoring of light intensity and combining it with preset thresholds, the system can promptly issue alerts, reminding parents or teachers to adjust lighting conditions and reduce the harmful effects of inappropriate lighting on eyesight. Secondly, real-time assessment of posture risk factors is crucial for correcting children's reading and writing postures. By monitoring children's pitch and yaw angles, the system can identify poor postures, issue timely warnings, and prompt children to adjust their posture, reducing eye strain and effectively preventing eye fatigue caused by improper posture. Furthermore, the system can assess children's visual health risks based on their screen time, preventing myopia caused by prolonged excessive screen time. By setting screen time thresholds, the system can alert parents and teachers when children's screen time approaches a high-risk level, allowing them to adjust study and rest schedules accordingly and avoid excessive screen time. Finally, a comprehensive assessment combining multidimensional data (light intensity, posture, and screen time) can more fully and accurately reflect children's vision health status. The system can dynamically adjust intervention strategies to ensure the flexibility and precision of intervention measures.

[0027] S2. Based on the risk sequence, calculate the statistical characteristics, and obtain the likelihood ratio based on the statistical characteristics. Determine whether there is a behavioral mutation in the risk sequence based on the likelihood ratio. If there is no mutation, set weights to weight and fuse the risk factors to obtain a risk index for risk assessment. If a mutation is detected, define the risk factor as an arm in the multi-armed slot machine algorithm and perform adaptive weight adjustment and feedback. Specifically, based on the risk sequence, statistical characteristics are calculated, and based on these statistical characteristics, the likelihood ratio is obtained. The likelihood ratio is then used to determine whether the risk sequence exhibits behavioral mutations, including: The mean and variance of the risk series within any window are calculated using the maximum likelihood estimation method. The formula for calculating the mean of the risk sequence within any window is as follows: In the formula, This represents the mean of the risk series within any window. Indicates the start time of the window. Indicates the end time of the window. Indicates time Risk factors; The formula for calculating the variance of the risk sequence within any window is as follows: In the formula, This represents the variance of the risk sequence within any window; Based on the mean and variance, a parameter vector is constructed for each window. Then, using the probability density function, the probability density function value of the risk factor under each window is obtained. The formula is as follows: In the formula, Indicates risk factors In window parameter vector The probability density value below, Represents the standard constant term ( ), Represents an exponential function; Each time point in the window is traversed as a candidate mutation point. The current window is then divided into two equally spaced sub-windows, and the mean, variance, and probability density values ​​of the two sub-windows are obtained again. Construct a statistical function and calculate the likelihood ratio for all mutation points. The formula is as follows: In the formula, This represents the statistical degree of mutation. Indicates candidate mutation points, Represents the natural logarithm function. Indicates risk factors In the parameter vector of the sub-window The probability density value below, Indicates risk factors In the parameter vector of the sub-window The probability density value under; Iterate through the current window and select the largest likelihood ratio for mutation detection; If the likelihood ratio is greater than the mutation detection threshold (This value can be set experimentally, and can be set from 5 to 20. For this invention, 16 can be used as the default value.) If the current risk factor undergoes a behavioral mutation within the window, it is determined that the current risk factor has not undergone a behavioral mutation within the window and is marked as 1. Otherwise, it is determined that the current risk factor has not undergone a behavioral mutation within the window and is marked as 0.

[0028] By calculating the mean and variance of risk factors within any time window, this invention can accurately quantify the central trend and volatility of risk factors, providing a reliable foundation for subsequent mutation detection. This method not only ensures data stability but also effectively identifies significant changes in any behavioral pattern. Secondly, a parameter vector for each window is constructed using the mean and variance, and the probability density function of the risk factor is calculated based on this. This step provides a comprehensive description of the distribution of risk factors within the window, enabling timely capture of trends in risk factor changes. The application of the probability density function allows the system to more sensitively identify subtle behavioral deviations, improving the system's early warning capabilities. Moreover, the GLR mutation detection method, by comparing the likelihood ratios of different sub-windows, can accurately assess the mutation degree of risk factors, exhibiting higher sensitivity and accuracy compared to traditional methods. This dynamic detection based on likelihood ratios allows the system to capture irregular changes in children's behavior in real time, such as sudden changes in light intensity and head posture. Finally, by setting appropriate thresholds, the system's performance can be optimized in different scenarios, avoiding false detections while ensuring that abnormal behaviors are detected promptly.

[0029] Furthermore, if no mutation occurs, weights are assigned to weighted and fused risk factors to obtain a risk index for risk assessment, including: Regression analysis is performed based on historical data to assign weights to each risk factor. If the result is 0, a weighted fusion is performed based on the weights and each risk factor to obtain the risk index, as shown in the formula: In the formula, Indicates at time Risk index at that time , , These represent the weights of risk factors such as illumination, posture, and duration of eye use; Assessment thresholds are set based on ophthalmological recommendations and knowledge in related fields. and ,and When the risk index is less than the assessment threshold If the child's current farsightedness reserve is sufficient, then the child is at low risk. When the risk index is greater than or equal to the assessment threshold And less than the evaluation threshold If this occurs, it indicates that the child's farsightedness reserve is being depleted too quickly, requiring attention to their eye-use behavior, and placing them at a medium risk. When the risk index is greater than or equal to the assessment threshold If the reading is negative, it indicates that the child currently has insufficient farsightedness reserve, is at risk of myopia, and is at high risk.

[0030] Employing a multi-dimensional risk assessment, this method comprehensively reflects children's vision risks under different environments and behaviors, avoiding misjudgments caused by single factors. For example, prolonged head-down posture or behavior in low-light environments can be promptly identified and alerted. Furthermore, the real-time monitoring and warning system can promptly detect and provide feedback on children's risk index; once it exceeds a safe threshold, the system will proactively remind parents or teachers to make adjustments to prevent vision problems. Finally, based on personalized assessment and a dynamic weighting mechanism, this method can better adapt to individual differences, provide precise health guidance, and effectively improve the management of children's eye health.

[0031] Furthermore, if a mutation is detected, the risk factor is defined as an arm in the multi-armed slot machine algorithm, and adaptive weight adjustment and feedback are performed, including: If the labeling result is 1, then each risk factor is defined as an arm in the multi-armed slot machine algorithm, resulting in a set of arms, including the illumination risk factor arm, posture risk factor arm, and eye usage duration risk factor arm. Using a complementary approach, an immediate reward function is defined for each arm, and the value of the immediate reward function is calculated using the following formula: In the formula, Indicates the first Each risk factor arm at time The instant reward function value at that time, Indicates at time The first time Risk factor values ​​for each risk factor arm; Based on the label result 1, the reset rule is executed to generate initial policy state variables for each arm, including the number of updates, the estimated historical average reward, and the reward history set; The execution reset rule is defined by the following formula: In the formula, Indicates the first Each risk factor arm at time The number of updates in the policy state variables at that time. Indicates the first Each risk factor arm at time Historical average reward estimates in the policy state variables. Indicates the first Each risk factor arm at time The reward history set in the policy state variables at the time; Based on the number of updates and the historical average reward estimate, a confidence upper bound formula is constructed for each arm, and the confidence upper bound value is obtained by solving the confidence upper bound formula through the bisection method. The formula for constructing the upper bound of confidence is as follows: In the formula, Indicates the first Each risk factor arm at time The upper bound of confidence at that time This indicates finding the maximum value in the set that satisfies the given conditions. Indicates candidate values, Denotes the KL divergence function. Indicates at time The confidence interval adjustment function value at that time; The specific formula for obtaining the value of the confidence interval adjustment function is as follows: In the formula, , These represent the main adjustment coefficient and the initial correction coefficient, respectively. Represents the natural logarithm function; It should be noted that: and The coefficients are positive real numbers, and their range of values ​​can be set as follows: , The default values ​​can be set as follows: , ; The KL divergence function adopts the Bernoulli form, and the formula is: It should be noted that: when or fall into At the boundary, for and The formula is modified as follows: In the formula, This indicates the operation of finding the minimum value. This indicates the operation of retrieving the maximum value. This represents the stability coefficient, which can be set to... ; Based on the confidence upper bound, Softmax normalization is used to calculate the weight of each arm, which serves as the weight of the risk factor corresponding to each arm. The formula is as follows: In the formula, Indicates the first Each risk factor arm at time Weight of time, This represents the natural exponential function. Indicates temperature parameter, This represents the total dimensions of the arms. Indicates the first Each risk factor arm at time The upper bound of confidence at that time; It should be noted that: The range of values ​​can be set to And 5 can be taken as the default value; Based on the weights, the results are fed back into the risk index weighted fusion formula to solve for the risk index.

[0032] By combining the Multi-Armed Slot Machine (MAB) algorithm with risk assessment, a dynamic and accurate assessment of farsightedness risk in children is achieved. First, based on an immediate reward function, the system can quantify changes in each risk factor in real time and adjust strategies accordingly, ensuring that children's behavioral patterns are consistently monitored and intervened in a scientific manner. When changes occur in the environment or behavior, the system utilizes a GLR mutation point detection mechanism to quickly identify abrupt changes in behavioral patterns, promptly adjusting assessment strategies and enhancing early warning capabilities. By introducing techniques such as upper confidence bound (UCB) and KL divergence, the system can dynamically adjust the weight of each risk factor based on historical data, allowing the weight of each factor to adapt adaptively under different conditions. For example, when there is insufficient light, the weight of the light risk factor increases, thus reminding children to improve their lighting environment. This weighting mechanism not only ensures the flexibility of the assessment results but also allows the system to better adapt to individual differences and provide personalized health recommendations. Furthermore, the Softmax normalization method is used for the normalization calculation of weights, ensuring that the comprehensive influence of each risk factor is reasonably distributed and avoiding excessive influence of any single factor on the assessment results. The system also sets a stability coefficient to prevent interference from extreme data, thereby enhancing the robustness and stability of the model.

[0033] S3. Store the risk index in a database; Specifically, risk indices are stored in a database, including: The risk index is stored in a database, and during the storage process, the corresponding child's ID is added to the risk index.

[0034] By precisely linking a child's farsightedness risk index with their ID and storing it in a database, a personalized and accurate vision health assessment is provided. Furthermore, it allows parents and doctors to quickly access information about a child's vision health status.

[0035] This embodiment also provides a method for assessing the risk of farsightedness in children based on multidimensional data, including: The data acquisition and mapping module is used to collect children's behavioral data, preprocess it, construct a sample set, divide the sample set into windows, and generate risk factors based on the mapping threshold to form a risk sequence. The calculation and judgment module is used to calculate statistical characteristics based on the risk sequence, obtain the likelihood ratio based on the statistical characteristics, and judge whether there is a behavioral mutation in the risk sequence based on the likelihood ratio. The execution generation module is used to set weights to perform weighted fusion of risk factors to obtain a risk index for risk assessment. If a mutation is detected, the risk factor is defined as an arm in the multi-armed slot machine algorithm, and adaptive weight adjustment and feedback are performed. The storage module is used to store the risk index in a database.

[0036] This embodiment also provides a computer device applicable to the case of a children's hyperopia risk assessment method based on multidimensional data, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the children's hyperopia risk assessment method based on multidimensional data as proposed in the above embodiment.

[0037] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0038] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the method for assessing the risk of farsightedness in children based on multidimensional data as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0039] 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 claims of the present invention.

Claims

1. A method for assessing the risk of farsightedness in children based on multidimensional data, characterized by: include, Children's behavioral data is collected and preprocessed to construct a sample set. The sample set is then divided into windows, and risk factors are generated based on mapping thresholds to form a risk sequence. Based on the risk sequence, statistical features are calculated, and based on the statistical features, the likelihood ratio is obtained. The likelihood ratio is used to determine whether there is a behavioral mutation in the risk sequence. If there is no mutation, weights are set to weight and fuse the risk factors to obtain a risk index for risk assessment. If a mutation is detected, the risk factor is defined as an arm in the multi-armed slot machine algorithm, and adaptive weight adjustment and feedback are performed. The risk index is stored in a database.

2. The method for assessing the risk of farsightedness in children based on multidimensional data as described in claim 1, characterized in that: The step of dividing the sample set into windows and generating risk factors based on mapping thresholds to form a risk sequence includes: The sample set is divided into windows using sliding window technology. Based on ophthalmological recommendations, requirements for healthy reading and writing postures for students, and guidelines for myopia prevention and control in children and adolescents, mapping thresholds are set for each data in the sample set. Each data in the sample set is mapped to a risk factor through the mapping thresholds, and the risk factors of illumination, posture, and eye use duration at each time in each window are obtained respectively. Based on the risk factors of illumination, posture, and eye use duration at each moment within each window, any type of risk factor is combined in chronological order to form three risk sequences for the window, including illumination, posture, and eye use duration risk sequences.

3. The method for assessing the risk of farsightedness in children based on multidimensional data as described in claim 2, characterized in that: The process of calculating statistical characteristics based on risk sequences, obtaining likelihood ratios based on these statistical characteristics, and determining whether behavioral mutations exist in the risk sequences based on the likelihood ratios includes: The mean and variance of the risk series within any window are calculated using the maximum likelihood estimation method. Based on the mean and variance, a parameter vector for each window is constructed, and then the probability density function is used to obtain the probability density function value of the risk factor under each window. Each time point in the window is traversed as a candidate mutation point. The current window is then divided into two equally spaced sub-windows, and the mean, variance, and probability density values ​​of the two sub-windows are obtained again. Construct a statistical function and calculate the likelihood ratio for all mutation points; Iterate through the current window and select the largest likelihood ratio for mutation detection; If the likelihood ratio is greater than the mutation detection threshold If the behavior of the current risk factor changes within the window, it is determined that a behavioral mutation has occurred and is marked as 1; otherwise, it is determined that no behavioral mutation has occurred within the window and is marked as 0.

4. The method for assessing the risk of farsightedness in children based on multidimensional data as described in claim 3, characterized in that: If no mutation occurs, weights are assigned to weighted and fused risk factors to obtain a risk index for risk assessment, including: Regression analysis is performed based on historical data to set the weights of each risk factor. If the result is 0, a weighted fusion is performed based on the weights and each risk factor to obtain the risk index. Assessment thresholds are set based on ophthalmological recommendations and knowledge in related fields. and ,and When the risk index is less than the assessment threshold If the child's current farsightedness reserve is sufficient, then the child is at low risk. When the risk index is greater than or equal to the assessment threshold And less than the evaluation threshold If this occurs, it indicates that the child's farsightedness reserve is being depleted too quickly, requiring attention to their eye-use behavior, and placing them at a medium risk. When the risk index is greater than or equal to the assessment threshold If the reading is negative, it indicates that the child currently has insufficient farsightedness reserve, is at risk of myopia, and is at high risk.

5. The method for assessing the risk of farsightedness in children based on multidimensional data as described in claim 4, characterized in that: If a mutation is detected, the risk factor is defined as an arm in the multi-armed slot machine algorithm, and adaptive weight adjustment and feedback are performed, including: If the labeling result is 1, then each risk factor is defined as an arm in the multi-armed slot machine algorithm, resulting in a set of arms, including the illumination risk factor arm, posture risk factor arm, and eye usage duration risk factor arm. A complementary approach is used to define the instantaneous reward function for each arm and calculate the instantaneous reward function value. Based on the label result 1, the reset rule is executed to generate initial policy state variables for each arm, including the number of updates, the estimated historical average reward, and the reward history set; Based on the number of updates and the historical average reward estimate, a confidence upper bound formula is constructed for each arm, and the confidence upper bound value is obtained by solving the confidence upper bound formula through the bisection method. Based on the confidence upper bound, Softmax normalization is used to calculate the weight of each arm, which serves as the weight of the risk factor corresponding to each arm. Based on the weights, the results are fed back into the risk index weighted fusion formula to solve for the risk index and the assessment threshold. and A new comparison was performed, and the comparison results were obtained.

6. The method for assessing the risk of farsightedness in children based on multidimensional data as described in claim 5, characterized in that: The storage of risk indices in a database includes: The risk index is stored in a database, and during the storage process, the corresponding child's ID is added to the risk index.

7. The method for assessing the risk of farsightedness in children based on multidimensional data as described in claim 6, characterized in that: The collected children's behavioral data is preprocessed to construct a sample set, including: For each child, data was collected using a light sensor, a nine-axis inertial measurement unit, and an eye tracker. The behavioral data of each child is denoised and standardized to construct a sample set for each child; The behavioral data includes light intensity data, attitude angle data, and eye usage time data. The attitude angle data includes pitch angle and yaw angle.

8. A children's hyperopia risk assessment system based on multidimensional data, based on the children's hyperopia risk assessment method based on multidimensional data as described in any one of claims 1 to 7, characterized in that: include, The data acquisition and mapping module is used to collect children's behavioral data, preprocess it, construct a sample set, divide the sample set into windows, and generate risk factors based on the mapping threshold to form a risk sequence. The calculation and judgment module is used to calculate statistical characteristics based on the risk sequence, obtain the likelihood ratio based on the statistical characteristics, and judge whether there is a behavioral mutation in the risk sequence based on the likelihood ratio. The execution generation module is used to set weights to perform weighted fusion of risk factors to obtain a risk index for risk assessment. If a mutation is detected, the risk factor is defined as an arm in the multi-armed slot machine algorithm, and adaptive weight adjustment and feedback are performed to generate the risk index. The storage module is used to store the risk index in a database.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for assessing the risk of farsightedness in children based on multidimensional data as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for assessing the risk of farsightedness in children based on multidimensional data as described in any one of claims 1 to 7.

Images