Myopia progress prediction method based on dynamic weight multi-gene risk scoring

Abstract

The invention provides a myopia progress prediction method based on a dynamic weight multi-gene risk score (PRS). The myopia progress prediction method comprises the following steps of: selecting a myopia progress prediction model according to the dynamic weight multi-gene risk score (PRS); the method comprises the following steps of: inputting a whole genome association study (GWAS) effect value of a basic population and individual genotype data of a target population, carrying out dynamic correction on a single nucleotide polymorphism (SNP) effect value in combination with a population differentiation index (Fst) and a linkage disequilibrium (LD) parameter of the target population, and constructing a PRS model with a population generalization ability. The method further collects individual environmental behavior data, including close-range eye load and outdoor exposure, calculates environmental interaction factors, and introduces the environmental interaction factors into a risk scoring model to comprehensively generate an individual dynamic risk score (PRSdynamic). The method has the characteristics of cross-racial adaptability, genetic-environment interaction modeling ability and optimization along with time change, can be used for individualized accurate prediction of myopia progress risks, and provides a scientific basis for myopia prevention and control research and intervention strategy formulation.

Description

Technical Field

[0001] This invention relates to the fields of bioinformatics and medical big data analysis, specifically to a method for predicting myopia progression based on dynamic weighted multigenic risk score (PRS), which is applicable to individualized myopia risk assessment and prediction, thereby improving the accuracy of myopia prevention and control and the value of intervention guidance. Background Technology

[0002] Myopia is one of the major vision health challenges worldwide, especially among adolescents, where its prevalence and incidence of high myopia continue to rise, becoming a key factor affecting public health and educational development. Its causes are mainly determined by the long-term interaction of genetic and environmental factors (such as close-range visual activity and reduced outdoor activities). Among these, genetic susceptibility plays a fundamental role in the occurrence and progression of myopia, possessing predictive and interventional value.

[0003] In recent years, with the advancement of large-scale genome-wide association studies (GWAS), researchers have identified over 200 single nucleotide polymorphisms (SNPs) highly associated with myopia, promoting the application of polygenic risk scores (PRS) in myopia prediction. However, most existing PRS methods employ a fixed-effects weighted summation approach, neglecting the moderating effects of population differences, time variations, and environmental factors on gene effects.

[0004] On the one hand, traditional PRS models are typically built based on GWAS data from European and American populations. When directly applied to East Asian populations, their prediction accuracy drops significantly. Existing research shows that the area under the receiver operating characteristic (AUC) of PRS for myopia prediction in European populations can reach over 0.72, while in East Asian populations it often drops to 0.60–0.65, indicating a significant lack of cross-racial adaptability. This performance degradation mainly stems from the different population groups exhibiting varying degrees of differentiation based on the fixation index between subpopulations and the total population (F...). st Significant differences exist between the original effect size and the linkage disequilibrium (LD) structure, leading to migration bias in the original effect size.

[0005] On the other hand, existing methods generally fail to establish genetic-environment interaction modeling capabilities, and cannot reflect the regulatory effects of behaviors such as outdoor exposure and close-range visual use on PRS. In addition, most PRS models are "static scores" and do not consider the possible biological attenuation or enhancement of gene effects as individuals grow or follow-up time extends, which limits their application value in long-term dynamic prediction.

[0006] Furthermore, existing research indicates that myopia progression during adolescence (especially between 12 and 18 years old) shows a decreasing trend in the genetic contribution with age, which may be closely related to developmental regulatory processes such as scleral remodeling and axial elongation (e.g., the expression of genes in pathways such as BMP2 and TGFB gradually declines after puberty). Therefore, constructing a PRS model with time-sensitive weight adjustment capabilities is particularly crucial. To this end, an effective approach is to use age-stratified F... st The index reflects the differences in genetic structure among populations at different stages, and introduces an exponential decay factor λ to adjust the SNP effect with age, making the scoring results closer to the true biological laws of myopia development.

[0007] Therefore, a novel PRS construction method with the following characteristics is urgently needed: (1) It can dynamically correct the GWAS effect between different ethnic groups and achieve cross-population generalization; (2) It can introduce the interactive regulation effect of environmental behavioral factors; (3) It has the ability to optimize the risk score over time to more accurately assess the risk of myopia progression and improve the scientificity and feasibility of individualized prevention and control. Summary of the Invention

[0008] The purpose of this invention is to overcome the limitations of existing multigene scoring methods for myopia risk and to provide a method for predicting myopia progression based on dynamic weighted PRS. This method can construct a dynamic PRS model by dynamically adjusting the effect value of single nucleotide polymorphism (SNP) in different ethnic backgrounds, combined with individualized environmental and behavioral data and time factors, to achieve accurate prediction of individual myopia progression risk.

[0009] To achieve the above objectives, the method of the present invention includes the following main technical solutions: a method for predicting myopia progression based on dynamic weighted PRS, the method comprising the following steps:

[0010] Step 1: Dynamic Correction of Cross-Ethnic Effect Values

[0011] Input GWAS summary statistics of the baseline population (e.g., European population) and filter for SNPs associated with myopia (P<5×10⁻⁶). -8), and based on the fixation index between subpopulations and the total population, F st The effect of parameters such as the target population (e.g., East Asian population) on the GWAS effect value (β) base The dynamic effect value (β) is dynamically adjusted. dynamic Calculate using the following formula:

[0012]

[0013] Where, β base The GWAS effect value for myopia in the baseline population, where t represents the individual's current follow-up age (in years), and F st (t) represents the dynamic differentiation index as a function of age. Based on the genotype data of the target population stratified by age, it is compared with the SNP allele frequencies provided by GWAS in the baseline population. The F-value defined by Weir & Cockerham is used. st The calculation formula is used to estimate (the formula is as follows), and the F-value of myopia-related SNPs for each age group is calculated. st Taking the average yields the dynamic degree of differentiation in genetic structure among different age groups. The mean of the overall differentiation index regardless of age group, λ is the decay factor, and LD target LD intensity for the target population.

[0014]

[0015] in:

[0016] The overall expected heterogeneity of the two groups;

[0017] The average heterozygosity of the two groups;

[0018] p1, p2: Allele frequencies of a certain SNP in the two populations;

[0019] Overall average frequency.

[0020] Step 2: Real-time quantification of environmental factors

[0021] Collect individual environmental behavior data, including near-vision load and outdoor exposure, including:

[0022]

[0023] Effective outdoor exposure = ∑ duration (h) of light intensity greater than 1000 lx and movement speed greater than 1 m / s.

[0024] Step 3: Calculation of dynamic polygenic risk score

[0025] First, calculate the baseline polygenic risk score (PRS). base ):

[0026]

[0027] where β base，i The effect size of each SNP in the baseline population GWAS. i The genotype of the target population.

[0028] Then calculate the dynamic polygenic risk score (PRS). dynamic ):

[0029]

[0030] Where, β dynamic,i The cross-racial dynamic adjustment effect value for each SNP, E1 is the daily mean outdoor exposure during the follow-up period, E2 is the daily mean visual load during the follow-up period, γ t δ represents the time-varying weights of the interaction between outdoor exposure and PRS. t The time-varying weight of the interaction term between eye strain and PRS is calculated using the following formula:

[0031] γ t =γ0·(1+α·Δt)

[0032] δ t =δ0·(1-β·Δt)

[0033] Where Δt represents the follow-up time span, and α and β are the interaction effect coefficients.

[0034] Step 4: Risk Classification

[0035] Individual-based PRS dynamic Percentile levels, combined with behavioral load indicators, are used to classify risks and guide myopia prevention and control measures. The classification criteria are set based on statistical distribution characteristics, and the specific classification criteria are as follows:

[0036] Red High Risk Level: PRS dynamic ≥99% and near-vision eye load >4 hours;

[0037] Yellow-level medium-to-high risk: PRS dynamic Between 95% and 98%;

[0038] Green Low Risk Level: PRS dynamic <95%.

[0039] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the aforementioned method for predicting myopia progression based on a dynamic weighted multi-gene risk score.

[0040] A computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the aforementioned method for predicting myopia progression based on dynamic weighted multi-gene risk scores.

[0041] Compared with the prior art, the advantages of the present invention are as follows:

[0042] (1) Cross-ethnic dynamic correction: This invention proposes a population differentiation index F based on age stratification. st (t) and time decay term The dual-factor regulation mechanism effectively addresses the migration bias issue of GWAS effect values ​​from the baseline population (e.g., European populations) to the target population (e.g., East Asian populations). Compared to traditional methods that only use static F... st By making simple proportional adjustments, this invention takes into account the changing trends of genetic structure at different age stages and the biological law that gene expression effects gradually decay with developmental time. It can more accurately adjust the applicability of SNP effects in different ethnic and age groups and improve the generalization performance of cross-ethnic PRS models.

[0043] (2) Dynamic interaction modeling of genetic and environmental factors: This invention not only constructs a PRS (Prognostics and Relationships System), but also introduces individual behavioral factors (such as near-field eye strain and outdoor light exposure) as real-time variable environmental variables. These are incorporated into a dynamic weight adjustment mechanism through mathematical modeling, achieving interactive modeling of genetic and environmental factors. Compared to traditional PRS, which only considers the linear accumulation of gene main effects, this invention can adjust risk weights according to the individual's actual behavioral state, better reflecting the risk evolution process in real-life situations and enhancing the interpretability and dynamic adaptability of the prediction.

[0044] (3) Model weights can be dynamically optimized over time: This invention introduces the follow-up time span Δt as a dynamic adjustment factor, allowing the interaction weights between behavioral factors and genetic risk scores to change over time, thereby enhancing the model's longitudinal tracking ability and risk evolution expression ability. This design can more realistically reflect the timeliness and cumulative effect of individual behavioral interventions, which is helpful for developing long-term, individualized myopia prevention and control strategies, and is compatible with multi-frequency data collection platforms such as mobile devices, improving the model's flexibility and accuracy in practical application scenarios.

[0045] (4) Clearer Risk Grading Logic: After constructing the scoring model, this invention further designs a clear risk grading mechanism, combining percentile distribution and behavioral threshold conditions for multi-dimensional grading, so that the output results not only have predictive ability but also behavioral guidance value. Compared with existing methods that only provide a single risk score and do not have intervention guidance significance, this invention has higher practical application value and operability, and is convenient for carrying out personalized intervention measures in clinical, family or public health scenarios.

[0046] In summary, this invention, through dynamic cross-ethnic effect correction, real-time interactive environmental modeling, time-varying weight optimization, and clear risk grading standards, can achieve individualized and accurate prediction of myopia progression in multi-ethnic and multi-factor contexts, thereby improving the scientific nature and application value of risk assessment. Attached Figure Description

[0047] Figure 1 This is a flowchart of a myopia progression prediction method based on dynamic weighted PRS according to the present invention, illustrating the main steps from data acquisition to risk classification.

[0048] Figure 2 This is a diagram illustrating the calculation of dynamic effect values, showing the calculation based on F... st The dynamic effect value correction process for LD and follow-up age t.

[0049] Figure 3 This is a schematic diagram of real-time quantification of environmental interaction factors, showing the calculation of near-field eye load and effective outdoor exposure and its interaction with PRS.

[0050] Figure 4 It is PRS dynamic The structural diagram of the construction and risk classification illustrates the composition of the dynamic PRS formula and its application in risk level classification. Detailed Implementation

[0051] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading the present invention, any modifications of the present invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0052] Example: This example uses data from a cohort of Chinese adolescents with myopia progression (follow-up period of 2 years, sample size of 1000) to predict the risk of myopia progression using the dynamic weighted PRS method of this invention. The specific steps are as follows:

[0053] Step 1: Data Preparation

[0054] The myopia GWAS meta-statistics of 460,536 samples from Europe (GWAS ID: ukb-b-6353) were downloaded from OpenGWAS, and SNPs were screened and effect sizes were extracted (P < 5 × 10⁻⁶). -8 A total of 150 SNPs were screened. Genotype data of each Chinese adolescent was obtained, and environmental behavior data (near-field eye use time, outdoor exposure time, pupil diameter, and light intensity) of the subjects were collected.

[0055] Step 2: Calculation of dynamic effect value

[0056] LD was calculated based on publicly available 1000 Genomes Phase 3 East Asian haplotype data. EAS Group differentiation index F st Data were obtained from publicly available literature, grouped into sets of 2 years. λ was set to 0.15 (±0.03), β... dynamic It can be obtained using the following formula:

[0057]

[0058] Among them F st (t) takes the corresponding value for the age group.

[0059] Step 3: Real-time quantification of environmental factors

[0060] Mobile devices collect environmental data every 30 minutes and calculate:

[0061]

[0062] (For example, 3h × 200lx / 4mm = 150 / day)

[0063] Effective outdoor exposure = ∑ duration (h) of light intensity greater than 1000 lx and movement speed greater than 1 m / s.

[0064] (e.g., 1.5 hours / day)

[0065] Step 4 Risk Scoring and Classification

[0066] calculate:

[0067]

[0068] in:

[0069] γ t =0.05·(1+0.01Δt)

[0070] δ t =0.07·(1-0.02Δt)

[0071] Δt=2

[0072] According to PRS dynamic Percentile:

[0073] • Red: PRS dynamic ≥99% and eye strain >4 hours (actually 82 cases)

[0074] Yellow: PRS dynamic 95%–98% (actual 122 cases)

[0075] • Green: The rest (actual 796 cases)

[0076] Step 5: Verification and Application

[0077] The study validated the occurrence of myopia progression (refractive error increase ≥1D) after a 2-year follow-up period against the risk level calculated using this method. Results showed that the proportion of myopia progression in the red high-risk group was significantly higher than that in the green low-risk group (P<0.01), indicating that the method of this invention can effectively distinguish individuals at different risk levels.

[0078] Furthermore, using the occurrence of myopia progression as the outcome variable, Nagelkerke R... based on logistic regression... 2 It is 5.69%, compared to the traditional PRS model's Nagelkerke R. 2 =2.49%, indicating that the method of this invention has a higher explanatory power in explaining the variation in risk of myopia progression.

[0079] The odds ratio (OR) of the dynamic weighted PRS was calculated using a logistic regression model. The results showed that the OR was 1.61 (95% CI: 1.06–2.44), indicating that individuals with high dynamic risk scores had a significantly higher risk of myopia progression than those with low risk scores.

[0080] Furthermore, the area under the receiver operating characteristic curve (AUC) constructed by the method of the present invention reaches 0.72, which is higher than the AUC of 0.64 of the traditional fixed-effects PRS, demonstrating the superior performance of the method of the present invention in binary classification prediction of myopia progression.

[0081] Optional data compliance processing solutions

[0082] In practical applications, this invention involves the processing of individual genotype information and the calculation of risk scores. To protect user privacy and data compliance, existing data security technologies can be used to anonymize the genetic data.

[0083] Federated learning mechanisms can be used for multi-center modeling or model updates without transferring raw genetic data, only sharing local model parameters, enabling collaborative training of risk models across institutions;

[0084] Differential privacy methods can inject perturbation noise when publishing PRS model parameters or calculation results to prevent sensitive individual information from being inferred from the model.

[0085] These alternative solutions do not affect the main technical concept of the present invention and can be integrated according to the use scenario to further improve the compliance and feasibility of the system in large-scale clinical deployment or multi-institutional collaboration environments.

[0086] Effect evaluation

[0087] This invention discloses a method for predicting myopia progression based on dynamic weighted PRS. In implementation, it involves inputting GWAS effect values, individual genotype data, and F... st β is calculated using the method of this invention, taking into account parameters such as LD, follow-up age t, and follow-up time span Δt, combined with environmental behavioral data (such as outdoor exposure and near-field visual load). dynamic and PRS dynamic And classify the risk of myopia progression in individuals.

[0088] Test results demonstrate that this method effectively combines cross-ethnic genetic effect adjustment and environmental-behavioral interaction modeling to achieve accurate prediction of myopia progression risk in individuals from different ethnic groups and under different environmental backgrounds. Compared with the traditional fixed-weight PRS method, the method of this invention significantly improves the accuracy and interpretability of risk stratification, providing a scientific and feasible algorithmic tool for myopia prevention and control research and individualized intervention.

[0089] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any transformations or substitutions that can be conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A method for predicting myopia progression based on dynamic weighted multi-gene risk scores, characterized in that, Includes the following steps: Step 1: Dynamic Adjustment of Cross-Ethnic Effect Values: Obtain the effect values ​​of the genome-wide association study (GWAS) in the baseline population and the individual genotype data of the target population, based on individual age and the fixation index between subpopulations and the total population (F...). st The GWAS effect value was dynamically corrected using the linkage disequilibrium (LD) parameter to obtain the cross-ethnic dynamically adjusted effect value. Step 2: Real-time quantification of environmental interaction factors: Collect individual environmental behavior data, including information on near-field visual load and outdoor exposure, and calculate environmental interaction factors; Step 3: Dynamic weight PRS calculation: Construct a PRS model that includes environmental interaction items and calculate the dynamic weight PRS of individuals; Step 4: Risk classification based on scoring criteria: According to the preset scoring criteria, the dynamic risk score is divided into different risk levels for myopia progression risk classification assessment and intervention guidance.

2. The method for predicting myopia progression based on dynamic weighted multi-gene risk scores according to claim 1, characterized in that, Step 1: Dynamic correction of cross-race effect values, as detailed below. Input baseline GWAS abstract statistics and screen for myopia-related single nucleotide polymorphisms (SNPs) (P < 5 × 10⁻⁶). -8 ) and its effect size (β) base Based on individual age, F st The dynamic effect value β is dynamically adjusted with the target population LD weight. dynamic The calculation formula is: Where, β base The GWAS effect value for myopia in the baseline population, where t represents the individual's current follow-up age (in years), and F st (t) represents the dynamic differentiation index as a function of age. Based on the genotype data of the target population stratified by age, it is compared with the SNP allele frequencies provided by GWAS in the baseline population. The F-value defined by Weir & Cockerham is used. st The calculation formula is as follows, and the F-values ​​of myopia-related SNPs for each age group are estimated. st Taking the average yields the dynamic degree of differentiation in genetic structure among different age groups. The mean of the overall differentiation index regardless of age group, λ is the decay factor, and LD target For the target population's LD intensity, in: The overall expected heterogeneity of the two groups; The average heterozygosity of the two groups; p1, p2: Allele frequencies of a certain SNP in the two populations; Overall average frequency.

3. The method for predicting myopia progression based on dynamic weighted multi-gene risk scores according to claim 2, characterized in that, Step 2: Real-time quantification of environmental interaction factors, as detailed below. Real-time data collection of near-field eye strain and outdoor exposure using mobile devices, including: Effective outdoor exposure = ∑ duration (h) of light intensity greater than 1000 lx and movement speed greater than 1 m / s.

4. The method for predicting myopia progression based on dynamic weighted multi-gene risk scores according to claim 3, characterized in that, Step 3: Calculation of dynamic weighted multi-gene risk score, as detailed below. An environmentally responsive multigene risk scoring model is constructed, and the calculation formula is as follows: Where β base,i The effect size of each SNP in the baseline population GWAS. i For the genotype of the target population, β dynamic,i The cross-racial dynamic adjustment effect value for each SNP, E1 is the daily mean outdoor exposure during the follow-up period, E2 is the daily mean visual load during the follow-up period, γ t δ represents the time-varying weights of the interaction between outdoor exposure and PRS. t The time-varying weights of the interaction term between eye load and PRS.

5. The method for predicting myopia progression based on dynamic weighted multi-gene risk scores according to claim 4, characterized in that, Step 4: Individual-based PRS dynamic Percentile levels, combined with behavioral load indicators, are used to classify risks and guide myopia prevention and control measures. The classification criteria are set based on statistical distribution characteristics, and the specific classification criteria are as follows: Red High Risk Level: PRS dynamic ≥99% and near-vision eye load >4 hours; Yellow-level medium-to-high risk: PRS dynamic Between 95% and 98%; Green Low Risk Level: PRS dynamic <95%.

6. The method for predicting myopia progression based on dynamic weighted multi-gene risk scores according to claim 5, characterized in that, The dynamic differentiation index F st (t) Calculated by age stratification, with each group consisting of 2 years, covering ages 6 to 25; The decay factor λ represents the exponential decay rate of the genetic effect over time, and its value ranges from 0.15 ± 0.

03. Based on the longitudinal myopia progression data of the target population, a nonlinear regression model is used to analyze PRS. dynamic The fitting results with age were determined; The LD target Generated from haplotype frequency data of the 1000 Genomes Phase 3 target population.

7. The method for predicting myopia progression based on dynamic weighted multi-gene risk scores according to claim 4, characterized in that, The interaction weights are dynamically optimized over time, and the calculation formula is as follows: c t =γ0·(1+α·Δt) d t =δ0·(1-β·Δt) in: γ0, δ0: Outdoor exposure behavior and near-field visual behavior and PRS of the target population in the initial stage (Δt=0). base The interaction weights obtained from the regression model fitting results typically range from 0.01 to 0.5 and are updated based on the myopia behavior intervention data of the target cohort. Δt: Follow-up time span; α, β: Adjustment coefficients for effects that vary over time; γ t : Time-varying weights of outdoor exposure and PRS interaction items; δ t : Time-varying weights of the interaction term between eye load and PRS.

8. The method for predicting myopia progression based on dynamic weighted multi-gene risk scores according to claim 4, characterized in that, This method outputs individualized dynamic PRS and risk grading results, which can be used for myopia progression risk research, risk warning, or intervention strategy development.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the myopia progression prediction method based on dynamic weighted multi-gene risk score as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by the processor, the computer instructions implement the method for predicting myopia progression based on dynamic weighted multi-gene risk scores as described in any one of claims 1-8.

Images