A dyslipidemia risk prediction apparatus, a terminal device, a method, and a storage medium

Abstract

The application discloses a dyslipidemia risk prediction device and terminal equipment, method and storage medium, and relates to the technical field of medical detection equipment. The device comprises an acquisition module, a first processing module and a second processing module which are connected with each other. The first processing module is used for inputting SNP genotype data of a to-be-predicted object into a pre-constructed scoring model, obtaining a polygenic risk score, and inputting an actual age and a phenotype age into a pre-constructed age linear regression model after obtaining the phenotype age based on clinical biochemical marker data and the actual age of the to-be-predicted object, so as to obtain a phenotype age acceleration value. The second processing module is used for inputting the polygenic risk score, the phenotype age acceleration value and individual characteristic data into a pre-constructed integrated model, obtaining a prediction result set, and taking a mode of the prediction result set as a prediction result of the to-be-predicted object. The device can be used for more comprehensive and accurate prediction and can improve processing efficiency.

Description

Technical Field

[0001] This application relates to the field of medical testing equipment technology, and in particular to a blood lipid dyslipidemia risk prediction device, terminal equipment, method, and storage medium. Background Technology

[0002] Dyslipidemia typically refers to elevated levels of at least one of the following in blood plasma: triglycerides and total cholesterol. In some cases, it may also include elevated low-density lipoprotein cholesterol (LDL-C) and decreased high-density lipoprotein cholesterol (HDL-C). It is a key risk factor for the continued rise in atherosclerotic cardiovascular disease. Therefore, early identification of high-risk individuals is of great public health significance for preventing and reducing the incidence of cardiovascular events.

[0003] Currently, related technologies construct multi-gene risk scores based on four independent blood lipid indicators (triglycerides, total cholesterol, low-density lipoprotein cholesterol, and high-density lipoprotein cholesterol) to capture the impact of genetic variations. However, these technologies only focus on genetic factors, and there is still room for improvement in predictive accuracy. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the related technologies, it is desirable to provide a dyslipidemia risk prediction device, terminal equipment, method, and storage medium that can achieve more comprehensive, accurate and rapid prediction, providing a reliable basis for subsequent intervention and personalized health management.

[0005] In a first aspect, this application provides a dyslipidemia risk prediction device, which includes an acquisition module, a first processing module, and a second processing module connected to each other. The acquisition module is used to acquire the SNP genotype data of the object to be predicted, the clinical biochemical marker data of the object to be predicted, and the individual characteristic data of the object to be predicted, including the actual age of the object to be predicted. The first processing module is used to input the SNP genotype data of the subject to be predicted into a pre-constructed scoring model to obtain the multi-gene risk score of the subject to be predicted, and to obtain the phenotypic age of the subject to be predicted based on the clinical biochemical marker data of the subject to be predicted and the actual age of the subject to be predicted, and then input the actual age and phenotypic age of the subject to be predicted into a pre-constructed age linear regression model to obtain the phenotypic age acceleration value of the subject to be predicted. The second processing module is used to input the multigene risk score of the object to be predicted, the phenotypic age acceleration value of the object to be predicted, and the individual characteristic data of the object to be predicted into a pre-constructed ensemble model to obtain a prediction result set, and use the mode of the prediction result set as the prediction result of the object to be predicted. Each element of the prediction result set is predicted by the corresponding sub-model in the ensemble model.

[0006] Optionally, in some embodiments of this application, in obtaining the phenotypic age of the subject based on the clinical biochemical marker data of the subject to be predicted and the actual age of the subject to be predicted, the clinical biochemical marker data of the subject to be predicted includes albumin, creatinine, glucose, C-reactive protein, lymphocyte percentage, mean cell volume, red blood cell distribution width, alkaline phosphatase, and blood cell count. The first processing module is specifically used to calculate the phenotypic age of the object to be predicted using the following formula: ; In the above formula, Indicates the phenotypic age of the subject to be predicted; ; .

[0007] Optionally, in some embodiments of this application, in terms of pre-building the ensemble model, the individual characteristic data of the training subjects include the actual age of the training subjects, the body mass index of the training subjects, the gender of the training subjects, the lifestyle of the training subjects, the dietary habits of the training subjects, and the metabolic indicators of the training subjects; The second processing module is also used to preprocess the actual age, body mass index, gender, lifestyle, dietary habits, and metabolic indicators of the training subjects, and to use the preprocessing results, the polygenic risk score, and the phenotypic age acceleration value of the training subjects as independent variables, and the dyslipidemia results of the training subjects as dependent variables, and to perform feature screening on the independent variables using the LASSO regression method.

[0008] Optionally, in some embodiments of this application, the second processing module is further used to calculate the marginal contribution of each feature to the risk score, and generate a feature importance map according to the order of the absolute value of the contribution.

[0009] Optionally, in some embodiments of this application, the ensemble model includes at least three of the following: k-nearest neighbor sub-model, logistic regression sub-model, linear discriminant sub-model, decision tree sub-model, random forest sub-model, support vector machine sub-model, and extreme gradient boosting sub-model.

[0010] Optionally, the scoring model described in some embodiments of this application is: ; In the above formula, Indicates a polygenic risk score; Indicates the SNP locus number. This indicates the total number of SNP sites; Indicates the first The weights of each SNP site, Indicates the first Genotypes of each SNP locus.

[0011] Optionally, the first embodiment described in some embodiments of this application The genotype of each SNP locus is wild-type, heterozygous mutant, or homozygous mutant.

[0012] Secondly, this application provides a terminal device that integrates the dyslipidemia risk prediction device described in any one of the first aspects.

[0013] Thirdly, this application provides a method for predicting the risk of dyslipidemia, wherein the method is used in the terminal device described in the second aspect, and the method includes: Acquire the SNP genotype data of the subject to be predicted, the clinical biochemical marker data of the subject to be predicted, and the individual characteristic data of the subject to be predicted, including the actual age of the subject to be predicted. The SNP genotype data of the subject to be predicted is input into a pre-constructed scoring model to obtain the multi-gene risk score of the subject to be predicted. After obtaining the phenotypic age of the subject to be predicted based on the clinical biochemical biomarker data and the actual age of the subject to be predicted, the actual age and the phenotypic age of the subject to be predicted are input into a pre-constructed age linear regression model to obtain the phenotypic age acceleration value of the subject to be predicted. The multigene risk score, phenotypic age acceleration value, and individual characteristic data of the subject to be predicted are input into a pre-constructed ensemble model to obtain a prediction result set. The mode of the prediction result set is used as the prediction result of the subject to be predicted. Each element of the prediction result set is predicted by the corresponding sub-model in the ensemble model.

[0014] Fourthly, this application provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps of the dyslipidemia risk prediction method described in the third aspect.

[0015] As can be seen from the above technical solutions, the embodiments of this application have the following advantages: This application provides a device, terminal equipment, method, and storage medium for predicting the risk of dyslipidemia. It obtains a prediction result set by fusing multi-dimensional data such as multi-gene risk scores, accelerated phenotypic age values, and individual characteristic data, considering all factors. The mode of the prediction result set is used as the prediction result for the object to be predicted. Each element of the prediction result set is predicted by its corresponding sub-model in the integrated model, effectively mitigating the bias or overfitting problems that may exist with a single model, significantly improving the robustness and accuracy of the overall prediction. Simultaneously, it optimizes the multi-gene risk score, which previously required separate calculations, into a single scoring model, simplifying the calculation process, improving computational efficiency and ease of application. Furthermore, it uses accelerated phenotypic age values ​​to quantify an individual's biological aging degree, effectively overcoming the limitation that relying solely on actual age may not accurately reflect an individual's physiological state. This significantly improves prediction accuracy and population applicability, providing a reliable basis for subsequent intervention and personalized health management. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A structural block diagram of a dyslipidemia risk prediction device provided in an embodiment of this application; Figure 2 A structural block diagram of a terminal device provided in an embodiment of this application; Figure 3 This is a flowchart illustrating a method for predicting the risk of dyslipidemia provided in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The following examples illustrate this. Figures 1 to 3 This application provides a detailed description of the lipid dyslipidemia risk prediction device, terminal equipment, method, and storage medium provided in the embodiments of this application.

[0021] Please refer to Figure 1 This is a structural block diagram of a dyslipidemia risk prediction device provided in an embodiment of this application. The dyslipidemia risk prediction device 100 includes an acquisition module 101, a first processing module 102, and a second processing module 103 that are connected to each other.

[0022] The acquisition module 101 acquires the SNP (single nucleotide polymorphism) genotype data, clinical biochemical marker data, and individual characteristic data of the subject to be predicted, including the subject's actual age. The first processing module 102 inputs the SNP genotype data into a pre-constructed scoring model to obtain a polygenic risk score and, based on the clinical biochemical marker data and the subject's actual age, obtains the subject's phenotypic age. Then, it inputs the subject's actual age and phenotypic age into a pre-constructed age linear regression model to obtain an accelerated phenotypic age value. The second processing module 103 inputs the subject's polygenic risk score, accelerated phenotypic age value, and individual characteristic data into a pre-constructed ensemble model to obtain a prediction result set. The mode of the prediction result set is used as the prediction result for the subject, where each element of the prediction result set is predicted by its corresponding sub-model within the ensemble model.

[0023] In some embodiments of this application, regarding the determination of the phenotypic age of the subject based on the clinical biochemical marker data and the actual age of the subject, the clinical biochemical marker data of the subject include, but are not limited to, albumin, creatinine, glucose, C-reactive protein, lymphocyte percentage, mean cell volume, red blood cell distribution width, alkaline phosphatase, and blood cell count, etc. The first processing module 102 specifically calculates the phenotypic age of the subject using formula (1): (1) In equation (1), Indicates the phenotypic age of the subject to be predicted; ; .

[0024] The following section details the construction process of the scoring model, the age-based linear regression model, and the ensemble model.

[0025] ① When constructing the scoring model, firstly, multigene risk loci for dyslipidemia are screened from databases, literature reports, and local case studies. The database can be the GWAS database (https: / / www.ebi.ac.uk / gwas / ), and the gene loci selection criteria are a correlation with hyperlipidemia of less than 5.0 × 10⁻⁶. -8 Literature reports may include, but are not limited to, PubMed, Embase, Scope, CNKI, and Chinese Medical Journal, while local case studies use patients with dyslipidemia or hyperlipidemia from historically published cases as the case group and members of the 1000 Genomes Association as the control group to conduct univariate logistic regression analysis to explore the relationship between variation and hyperlipidemia.

[0026] The selection criteria included: (a) studying the variation frequency of the selected gene loci using the dbSNP database (https: / / www.ncbi.nlm.nih.gov / snp) and the population multi-omics reference database (http: / / bioinformatics.hit.edu.cn / chnpop / ), and excluding loci with a variation frequency of less than 0.01 in East Asian or Asian populations; (b) when multiple studies showed duplicate loci, Hardy-Weinberg equilibrium was performed on the control genotype distribution of each study, and allele frequencies greater than 0.05 were considered unbiased, and the OR value and corresponding 95% CI were recalculated; (c) loci with an OR value of less than or equal to 1 associated with hyperlipidemia were excluded, including those weakly associated with polygenic risk of hyperlipidemia. A total of 353 SNP loci were ultimately selected, and the results are shown in Table 1.

[0027] Table 1. Information on the 353 selected SNP sites

[0028] Secondly, the genotypes of each SNP locus in the two groups (case group and control group) were known, and the distribution frequency of each genotype at each SNP locus in the case group and control group was shown in Table 2. GG indicates carrying 0 risk alleles, GA indicates carrying 1 risk allele, and AA indicates carrying 2 risk alleles. A random function was used to randomly assign each genotype and whether or not the individual had dyslipidemia to obtain a simulated population of 2000 cases and 2000 healthy controls.

[0029] Table 2 Examples of Genotype Distribution Frequencies

[0030] Next, based on the above, the polygenic risk score for each individual in each group of the simulated population is calculated by multiplying log(OR) by the weighted sum of genotypes. In the 2×2 contingency table, OR = (number of mutated and diseased individuals × number of non-mutated and disease-free individuals) / (number of mutated but disease-free individuals × number of non-mutated and disease-free individuals). Then, the area under the receiver operating characteristic (AUC) curve is calculated using the pROC package in R to evaluate the predictive power of the scoring model. In this analysis, the individual's polygenic risk score is used as the independent variable, and whether or not the individual has the disease is used as a binary dependent variable for AUC calculation. AUC measures the model's ability to distinguish between high and low disease risk levels, ranging from 0.5 to 1.0. The scoring model in this embodiment has an AUC of 0.7, indicating that it has a certain discriminatory ability and is suitable for risk stratification at the population level.

[0031] Furthermore, the scoring model is as follows: (2) In equation (2), Indicates a polygenic risk score; Indicates the SNP locus number. This indicates the total number of SNP sites, for example, 353; Indicates the first The weight of the SNP locus is given by the value of the first SNP locus. ln, the natural logarithm of the number of SNP sites; Indicates the first The genotype of each SNP locus is categorized as wild-type, heterozygous mutant, or homozygous mutant. Wild-type refers to carrying 0 risk alleles (denoted as 0), heterozygous mutant refers to carrying 1 risk allele (denoted as 1), and homozygous mutant refers to carrying 2 risk alleles (denoted as 2). In practice, 353 SNP loci are used. and The sum of the products is obtained by calculation. ,Right now .

[0032] In addition, this application also selected members of the 1000 Genomes Study as the research population, and calculated the polygenic risk score for each individual. The calculation results showed that the polygenic risk scores of the 2504 study participants ranged from 16.53 to 24.76. Based on the distribution of polygenic risk scores in the population, the population could be divided into five levels: very high risk, high risk, average risk, low risk, and very low risk of dyslipidemia. The average risk group was defined as the population with polygenic risk scores at the median of the entire study population. Using the average risk group as a reference population, the polygenic risk threshold was calculated as follows: 5% of the population with a polygenic risk score between 21.49 and 24.76; 20% of the population with a polygenic risk score between 20.48 and 21.49; 20% of the population with a polygenic risk score between 18.32 and 19.22; and 5% of the population with a polygenic risk score between 16.53 and 18.32. Individual genetic variations will be fitted to the corresponding polygenic risk score, as shown in Table 3.

[0033] Table 3 Thresholds and risk stratification of the dyslipidemia risk prediction model based on multigene risk scores

[0034] ② When constructing the age-based linear regression model, the baseline data of the publicly available population from the China Health and Retirement Longitudinal Study (CHARLS) database were first queried, including actual age and clinical biochemical biomarker data. If missing biochemical biomarker data were found, the missing values ​​were filled using mean imputation to obtain complete biochemical biomarker data. Furthermore, the extreme value adjustment method was used to limit the value of each biochemical biomarker to between the 1st and 99th percentiles, obtaining adjusted biochemical biomarker data.

[0035] Similarly, the phenotypic age of each individual in the baseline data is calculated using formula (1), and then an age linear regression model is constructed by combining the actual age of each individual in CHARLS to obtain the intercept. and slope ,Right now: (3) In equation (3), The regression residual represents the difference between an individual's actual phenotypic age and the regression prediction value. It serves as an important basis for judging the aging process. Based on the regression residual, the phenotypic age acceleration value is obtained. A positive value means that the biological aging rate is faster than that of the same age group, while a negative value means that the biological age is slower than that of the same age group.

[0036] ③ When constructing the ensemble model, the individual characteristic data of the training subjects include their actual age, body mass index (BMI), gender, lifestyle, dietary habits, and metabolic indicators. The second processing module 103 can also preprocess the actual age, BMI, gender, lifestyle, dietary habits, and metabolic indicators of the training subjects. For example, the actual age is confirmed with the training subjects and standardized in years; the BMI is normalized; in gender, 1 represents male and 0 represents female; lifestyle and dietary habits are obtained from the diet and lifestyle questionnaire (1-5 points), where 1 represents no (average) and >1 represents yes (good); the metabolic indicators mainly refer to the fasting glucose result (unit: mmol / L), and extreme value adjustment of the biomarker data is performed by setting the lowest 1st percentile and the highest 99th percentile.

[0037] Secondly, the preprocessing results, the multigenic risk scores of the training subjects, and the phenotypic age acceleration values ​​of the training subjects were used as independent variables, and the dyslipidemia results of the training subjects were used as dependent variables. LASSO regression was used to screen the independent variables to remove insignificantly relevant ones. It should be noted that LASSO is a regularized regression method, extremely important in feature selection for high-dimensional data. It achieves feature screening by adding an L1 regularization term to the linear model, which can reduce some unimportant feature coefficients, even reducing them to zero. For example, the linear model is... , For an n×1 observation vector, Design a matrix for n×p. Let p×1 be an unknown parameter vector. For random error, Expressing expectations, Represents the covariance matrix. The variance of the random error. It is the identity matrix. Estimated parameter vector. The basic method is the least squares method, the idea of ​​which is the error vector. To make Q as small as possible, that is, to make Q( )=‖ ||=( )'( ) to reach the minimum.

[0038] The generalization function has changed from the original Become , It is usually the L1 or L2 norm. It is an adjustable parameter that controls the strength of regularization. When used on linear models, L1 regularization and L2 regularization are also called Lasso and Ridge, respectively. The 1-norm of a vector is the sum of the absolute values ​​of its elements; L1 regularization sets the coefficient to 1 / 2. The L1 norm is added as a penalty term to the loss function. Since the regularization term is non-zero, this forces the coefficients corresponding to weak features to become 0. Therefore, L1 regularization often makes the learned model very sparse (coefficients...). (often equal to 0), this characteristic makes L1 regularization a good feature selection method. If we continue to increase... The value of will result in a sparser model, meaning more and more feature coefficients. It will become 0. The embodiments of this application add additional constraints or penalty terms to the existing model (loss function) through regularization, and remove independent variables that are not significantly related to the occurrence of dyslipidemia to prevent overfitting of multiple indicators, thereby effectively improving the generalization ability of the final model.

[0039] Finally, based on the independent and dependent variables selected through feature selection, at least three of the following sub-models—k-nearest neighbor, logistic regression, linear discriminant, decision tree, random forest, support vector machine, and extreme gradient boosting—are chosen as the ensemble model architecture for training. The final hyperparameters used for model optimization are determined through a combination of automated tuning and manual fine-tuning. This application embodiment employs a voting method as the model ensemble approach, integrating the prediction results of multiple base learners through majority voting or averaging mechanisms. It also uses SHAP values ​​to interpret the model output. Specifically, the second processing module 103 can calculate the marginal contribution of each feature to the risk score and generate a feature importance map based on the absolute value of the contribution. For example, in binary classification data, the prediction results of the k-nearest neighbor sub-model, linear discriminant sub-model, decision tree sub-model, random forest sub-model, support vector machine sub-model, and extreme gradient boosting sub-model are either 0 or 1, where 1 represents a high risk of dyslipidemia and 0 represents a low risk. The prediction result of the logistic regression sub-model is the probability of a high risk of dyslipidemia, defined as >0.5 as a high risk (represented by 1), and vice versa (represented by 0). Furthermore, this application embodiment can utilize monthly updated usage data to retrain and optimize the model, achieving continuous model iteration and improving performance evolution and generalization capabilities. For first-time users, their multi-gene risk locus genotypes and related clinical biochemical indicators will be collected simultaneously. After genetic risk stratification based on predetermined thresholds, the data will be input into the model to complete the initial risk assessment. For users who are not first-time users, only their updated clinical biochemical indicators need to be detected, and the genotype data stored during the first use will be automatically linked.

[0040] In another aspect, embodiments of this application provide a terminal device. Please refer to... Figure 2 This is a structural block diagram of a terminal device provided in an embodiment of this application. The terminal device 200 integrates... Figure 1 The corresponding embodiment is the dyslipidemia risk prediction device 100.

[0041] In another aspect, embodiments of this application provide a method for predicting the risk of dyslipidemia, which is used for... Figure 2 The terminal device 200 in the corresponding embodiment. Please refer to... Figure 3 This is a flowchart illustrating a method for predicting the risk of dyslipidemia provided in an embodiment of this application. The method specifically includes the following steps: S101, Obtain the SNP genotype data, clinical biochemical marker data, and individual characteristic data of the subject to be predicted. The individual characteristic data of the subject to be predicted includes the actual age of the subject.

[0042] S102, input the SNP genotype data of the subject to be predicted into the pre-constructed scoring model to obtain the multi-gene risk score of the subject to be predicted, and obtain the phenotypic age of the subject to be predicted based on the clinical biochemical marker data and the actual age of the subject to be predicted. Then input the actual age and phenotypic age of the subject to be predicted into the pre-constructed age linear regression model to obtain the phenotypic age acceleration value of the subject to be predicted.

[0043] S103, input the multigene risk score of the subject to be predicted, the phenotypic age acceleration value of the subject to be predicted, and the individual characteristic data of the subject to be predicted into the pre-constructed ensemble model to obtain the prediction result set, and use the mode of the prediction result set as the prediction result of the subject to be predicted. Each element of the prediction result set is predicted by the corresponding sub-model in the ensemble model.

[0044] For example, if the prediction result set is {1,1,0,0,0,1,1}, and the voting integration determines that this subject is assessed as having a high risk of dyslipidemia, it is recommended that they subsequently take targeted health management measures, including adjusting their diet, optimizing their lifestyle, and regularly monitoring blood lipids and related metabolic indicators, so that timely clinical intervention can be carried out if abnormalities occur. As another example, if the prediction result set is {1,0,0,0,0,1,1}, this subject is assessed as having a low risk of dyslipidemia, and it is recommended that they continue to maintain a healthy diet and a good lifestyle, and regularly monitor blood lipids and related metabolic indicators to achieve long-term dynamic risk management.

[0045] It should be noted that the descriptions of the same steps and contents as in other embodiments in this embodiment can be found in the descriptions in other embodiments, and will not be repeated here.

[0046] The lipid dyslipidemia risk prediction device, terminal equipment, and method provided in this application obtain a prediction result set by integrating multi-dimensional data such as multi-gene risk scores, phenotypic age acceleration values, and individual characteristic data. This comprehensive approach considers all factors and uses the mode of the prediction result set as the prediction result for the object to be predicted. Each element of the prediction result set is predicted by the corresponding sub-model in the integrated model, thereby effectively mitigating the bias or overfitting problems that may exist in a single model and significantly improving the robustness and accuracy of the overall prediction. At the same time, the multi-gene risk scores that previously needed to be calculated separately are optimized into a single scoring model, simplifying the calculation process, improving computational efficiency and ease of application. Furthermore, the use of phenotypic age acceleration values ​​to quantify the degree of biological aging of an individual effectively compensates for the limitation that relying solely on actual age may not accurately reflect the individual's physiological state, significantly improving prediction accuracy and population applicability, and providing a reliable basis for subsequent intervention and personalized health management.

[0047] In another aspect, embodiments of this application provide a computer-readable storage medium for storing program code for executing the aforementioned... Figure 3 Any implementation of the dyslipidemia risk prediction method in the corresponding embodiment.

[0048] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0049] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed between each other can be through some interfaces, indirect coupling or communication connection between devices or modules, and can be electrical, mechanical, or other forms. Modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0050] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each module can exist physically separately, or two or more units can be integrated into one module. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0051] Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the lipid dyslipidemia risk prediction method of various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0052] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0053] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A device for predicting the risk of dyslipidemia, characterized in that, The dyslipidemia risk prediction device includes an acquisition module, a first processing module, and a second processing module that are interconnected. The acquisition module is used to acquire the SNP genotype data of the object to be predicted, the clinical biochemical marker data of the object to be predicted, and the individual characteristic data of the object to be predicted, including the actual age of the object to be predicted. The first processing module is used to input the SNP genotype data of the subject to be predicted into a pre-constructed scoring model to obtain the multi-gene risk score of the subject to be predicted, and to obtain the phenotypic age of the subject to be predicted based on the clinical biochemical marker data of the subject to be predicted and the actual age of the subject to be predicted, and then input the actual age and phenotypic age of the subject to be predicted into a pre-constructed age linear regression model to obtain the phenotypic age acceleration value of the subject to be predicted. The second processing module is used to input the multigene risk score of the object to be predicted, the phenotypic age acceleration value of the object to be predicted, and the individual characteristic data of the object to be predicted into a pre-constructed ensemble model to obtain a prediction result set, and use the mode of the prediction result set as the prediction result of the object to be predicted. Each element of the prediction result set is predicted by the corresponding sub-model in the ensemble model.

2. The dyslipidemia risk prediction device according to claim 1, characterized in that, In determining the phenotypic age of the subject based on the clinical biochemical biomarker data and the subject's actual age, the clinical biochemical biomarker data of the subject includes albumin, creatinine, glucose, C-reactive protein, lymphocyte percentage, mean cell volume, red blood cell distribution width, alkaline phosphatase, and blood cell count. The first processing module is specifically used to calculate the phenotypic age of the object to be predicted using the following formula: ； In the above formula, Indicates the phenotypic age of the subject to be predicted; ； 。 3. The dyslipidemia risk prediction device according to claim 1, characterized in that, Regarding the pre-built ensemble model, the individual characteristic data of the training subjects include the actual age of the training subjects, the body mass index of the training subjects, the gender of the training subjects, the lifestyle of the training subjects, the dietary habits of the training subjects, and the metabolic indicators of the training subjects; The second processing module is also used to preprocess the actual age, body mass index, gender, lifestyle, dietary habits, and metabolic indicators of the training subjects, and to use the preprocessing results, the polygenic risk score, and the phenotypic age acceleration value of the training subjects as independent variables, and the dyslipidemia results of the training subjects as dependent variables, and to perform feature screening on the independent variables using the LASSO regression method.

4. The dyslipidemia risk prediction device according to claim 3, characterized in that, The second processing module is also used to calculate the marginal contribution of each feature to the risk score and generate a feature importance map according to the order of the absolute value of the contribution.

5. The dyslipidemia risk prediction device according to claim 3, characterized in that, The ensemble model includes at least three of the following: k-nearest neighbor sub-model, logistic regression sub-model, linear discriminant sub-model, decision tree sub-model, random forest sub-model, support vector machine sub-model, and extreme gradient boosting sub-model.

6. The dyslipidemia risk prediction device according to any one of claims 1 to 5, characterized in that, The scoring model is as follows: ； In the above formula, Indicates a polygenic risk score; Indicates the SNP locus number. This indicates the total number of SNP sites; Indicates the first The weights of each SNP site, Indicates the first Genotypes of each SNP locus.

7. The dyslipidemia risk prediction device according to claim 6, characterized in that, The first The genotype of each SNP locus is wild-type, heterozygous mutant, or homozygous mutant.

8. A terminal device, characterized in that, The terminal device integrates the dyslipidemia risk prediction device according to any one of claims 1 to 7.

9. A method for predicting the risk of dyslipidemia, characterized in that, The method for predicting the risk of dyslipidemia is used in the terminal device of claim 8, and the method for predicting the risk of dyslipidemia includes: Acquire the SNP genotype data of the subject to be predicted, the clinical biochemical marker data of the subject to be predicted, and the individual characteristic data of the subject to be predicted, including the actual age of the subject to be predicted. The SNP genotype data of the subject to be predicted is input into a pre-constructed scoring model to obtain the multi-gene risk score of the subject to be predicted. After obtaining the phenotypic age of the subject to be predicted based on the clinical biochemical biomarker data and the actual age of the subject to be predicted, the actual age and the phenotypic age of the subject to be predicted are input into a pre-constructed age linear regression model to obtain the phenotypic age acceleration value of the subject to be predicted. The multigene risk score, phenotypic age acceleration value, and individual characteristic data of the subject to be predicted are input into a pre-constructed ensemble model to obtain a prediction result set. The mode of the prediction result set is used as the prediction result of the subject to be predicted. Each element of the prediction result set is predicted by the corresponding sub-model in the ensemble model.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps of the lipid dyslipidemia risk prediction method of claim 9.

Images