Risk assessment method, system and equipment for idiopathic pulmonary hypertension

Abstract

The invention discloses a risk assessment method, system and equipment for idiopathic pulmonary arterial hypertension, and belongs to the field of pulmonary arterial hypertension. According to the method, SNP data containing genotypes and effect values, protein marker expression quantity, metabonomics and clinical data are obtained, the SNP effect values are corrected based on linkage imbalance reference information, and PRS is calculated in combination with the genotypes; constructing a protein expression score by utilizing the site effect value and the expression quantity of the pQTL, and fusing the protein expression score with the PRS to form a target PRS; carrying out dimensionality reduction on metabolome data by adopting sparse coding, extracting sparse coefficients of IPAH related metabolic pathways, and converting the sparse coefficients into metabolic pathway scores; converting the clinical indexes into clinical risk scores; based on the clinical parameter distribution target PRS, the metabolic pathway score and the weight coefficient of the clinical risk score, calculating a risk assessment value; and finally, matching the evaluation value with a preset risk threshold value, and outputting a risk evaluation level. And the IPAH risk assessment accuracy of common people is improved.

Description

Technical Field

[0001] This application belongs to the field of pulmonary hypertension, and in particular relates to a risk assessment method, system and device for idiopathic pulmonary hypertension. Background Technology

[0002] Idiopathic pulmonary arterial hypertension (IPAH) is one of the most common subtypes of pulmonary arterial hypertension (PAH). Because IPAH lacks specific symptoms in its early stages, patients are usually diagnosed only when pulmonary vascular obstruction is severe, at which point the optimal intervention window has been missed. Therefore, early risk assessment of IPAH is crucial for identifying high-risk individuals, enabling prodromal intervention, and improving patient survival.

[0003] Currently, existing technologies primarily focus on risk assessment for patients diagnosed with PAH, mainly based on clinical data, and predict the progression of pulmonary hypertension through dynamic stratification. However, methods for predicting the risk of IPAH in the general population are still lacking, and current technologies lack risk assessment methods for IPAH in the general population.

[0004] Existing PAH risk assessment models rely on clinical indicators after diagnosis, which cannot meet the early screening needs of the general population. Moreover, PAH has significant genetic heterogeneity, and the differences in pathogenic mechanisms among different subtypes limit the predictive efficacy of models using general existing technologies. Therefore, existing risk assessment methods have technical problems such as being unsuitable for risk assessment of IPAH in the general population and having poor accuracy in risk assessment. Summary of the Invention

[0005] This application provides a method, system, device, and computer-readable storage medium for risk assessment of idiopathic pulmonary hypertension, which can be applied to the risk assessment of idiopathic pulmonary hypertension in the general population.

[0006] In a first aspect, this application provides a risk assessment method for idiopathic pulmonary hypertension, the method comprising:

[0007] Acquire single nucleotide polymorphism (SNP) data associated with idiopathic pulmonary hypertension (IPH), as well as protein biomarkers, metabolomics data, and clinical data associated with IPH in the target subjects. Each SNP data includes the corresponding genotype and effect size.

[0008] Based on pre-defined linkage disequilibrium reference information, the effect values ​​of single nucleotide polymorphism (SNP) data are corrected using a Bayesian framework, and a polygenic risk score is determined by combining the genotype and corrected effect value of each SNP.

[0009] Based on the expression levels of protein biomarkers in the protein biomarker data and the site effect values ​​of the protein quantitative trait loci that regulate protein biomarkers, a protein expression score is obtained, and a target polygenic risk score is calculated based on the protein expression score and the polygenic risk score.

[0010] The dimensionality of metabolomics data was reduced by sparse coding, the sparse coefficients of metabolic pathways associated with idiopathic pulmonary hypertension were determined, and the sparse coefficients of metabolic pathways were converted into metabolic pathway scores.

[0011] The clinical indicators in the clinical data are quantified and weighted to obtain the clinical risk score;

[0012] Based on the clinical parameters in the clinical data, determine the weight coefficients corresponding to the target polygenic risk score, metabolic pathway score and clinical risk score, and combine the target polygenic risk score, metabolic pathway score and clinical risk score to determine the risk assessment value.

[0013] The risk assessment value is matched with the preset age-stratified risk threshold range to obtain the risk assessment level information of the target object.

[0014] In one feasible implementation, the method further includes:

[0015] To obtain a set of protein biomarker datasets related to idiopathic pulmonary hypertension in different age groups of the general population;

[0016] Based on the expression levels of protein biomarkers in each age group in the protein biomarker dataset, time series analysis was performed to obtain protein biomarker age parameters that characterize the changing trend of protein biomarker expression levels associated with idiopathic pulmonary hypertension with age.

[0017] The protein expression score was adjusted by weighting the protein expression score using the age parameter of the protein biomarker.

[0018] In one feasible implementation, a protein expression score is obtained based on the expression level of protein biomarkers in protein biomarker data and the site effect value of preset protein quantitative trait loci regulating protein biomarkers. A target polygenic risk score is then calculated based on the protein expression score and the polygenic risk score, including:

[0019] The contribution value of each protein biomarker is obtained by multiplying the expression level of the protein biomarker and the site effect value of each protein quantitative trait locus that regulates the protein biomarker, and then summing them up.

[0020] The contribution values ​​of all protein markers are summed to obtain the protein expression score;

[0021] The target polygenic risk score is obtained by directly adding the polygenic risk score to the protein expression score.

[0022] In one feasible implementation, dimensionality reduction of metabolomics data is performed using sparse coding to determine the sparse coefficients of metabolic pathways associated with idiopathic pulmonary hypertension, and these sparse coefficients are converted into metabolic pathway scores, including:

[0023] Based on a pre-defined database of metabolic pathways related to idiopathic pulmonary hypertension, a metabolic association matrix was constructed, with the rows and columns of the association matrix corresponding to metabolic pathways related to idiopathic pulmonary hypertension and metabolites in the metabolomics data, respectively.

[0024] The concentration values ​​of each metabolite in the metabolomics data are multiplied with the correlation matrix to generate an initial metabolic pathway activity matrix, which is then decomposed into a metabolic pathway weight matrix and a sparse coefficient matrix.

[0025] An alternating direction optimization method is used to iteratively update the metabolic pathway weight matrix and the sparse coefficient matrix. In each iteration, when the metabolic pathway weight matrix is ​​fixed, an L1 norm constraint is applied to the sparse coefficient matrix to determine the key metabolic pathways. When the sparse coefficient matrix is ​​fixed, the update direction of the metabolic pathway weight matrix is ​​limited by the correlation matrix.

[0026] When the update difference of the metabolic pathway weight matrix is ​​less than the preset convergence threshold, the sparse coefficients corresponding to the metabolic pathways included in the metabolic pathway database are selected from the sparse coefficient matrix, and the metabolic pathway score is obtained by summing the sparse coefficients corresponding to all metabolic pathways included in the metabolic pathway database.

[0027] In one feasible implementation, the clinical parameters include age parameters and lung function measurements;

[0028] Based on clinical parameters from clinical data, determine the weighting coefficients corresponding to the target polygenic risk score, metabolic pathway score, and clinical risk score. Then, combine these scores to determine the risk assessment value, including:

[0029] The initial weight set is matched based on age parameters. The initial weight set includes the baseline weights of the target polygenic risk score, metabolic pathway score and clinical risk score.

[0030] Based on the difference between the lung function measurement value and the preset measurement threshold, the baseline weight of the clinical risk score in the initial weight set is adjusted, and the adjustment range of the baseline weight of the clinical risk score is directly proportional to the difference.

[0031] Keep the total baseline weight of the target polygenic risk score and the metabolic pathway score unchanged, and readjust the allocation ratio of the baseline weight of the target polygenic risk score and the metabolic pathway score. The allocation ratio of the baseline weight of the target polygenic risk score is directly proportional to the difference.

[0032] Based on the adjusted baseline weights corresponding to the target polygenic risk score, metabolic pathway score, and clinical risk score, the risk assessment value is generated by weighted summation of the corresponding target polygenic risk score, metabolic pathway score, and clinical risk score.

[0033] In one feasible implementation, based on pre-defined linkage disequilibrium reference information, the effect values ​​of single nucleotide polymorphism (SNP) data are corrected using a Bayesian framework. Then, combining the genotype and corrected effect value for each SNP, a polygenic risk score is determined, including:

[0034] Based on the preset linkage disequilibrium reference information, the genotype data region of single nucleotide polymorphism data in the reference panel of the linkage disequilibrium reference information is divided by a sliding window to construct a local linkage disequilibrium matrix.

[0035] Based on the effect values ​​and local linkage disequilibrium matrix of single nucleotide polymorphism data, the posterior mean effect value of each single nucleotide polymorphism data was calculated using a point normal mixture Bayesian model and Markov chain Monte Carlo sampling.

[0036] The posterior mean effect value was used as the corrected effect value for each single nucleotide polymorphism data point.

[0037] The corrected effect value of each single nucleotide polymorphism is multiplied by the corresponding genotype and summed to generate a polygenic risk score.

[0038] In one feasible implementation, clinical data includes physiological indicators, medical history, and demographic information.

[0039] Secondly, embodiments of this application provide a risk assessment system for idiopathic pulmonary hypertension, the system comprising:

[0040] The acquisition module is used to acquire single nucleotide polymorphism data associated with idiopathic pulmonary hypertension, as well as protein biomarker data, metabolomics data and clinical data associated with idiopathic pulmonary hypertension of the target subjects. Each single nucleotide polymorphism data includes the corresponding genotype and effect value.

[0041] The determination module is used to correct the effect values ​​of single nucleotide polymorphism data using a Bayesian framework based on preset linkage disequilibrium reference information, and to determine the polygenic risk score by combining the genotype and corrected effect value of each single nucleotide polymorphism data.

[0042] The calculation module is used to obtain a protein expression score based on the expression level of protein markers in the protein marker data and the site effect value of the protein quantitative trait loci that regulate protein markers, and to calculate the target multigene risk score based on the protein expression score and the multigene risk score.

[0043] The determination module is also used to reduce the dimensionality of metabolomics data through sparse coding, determine the sparse coefficient of metabolic pathways related to idiopathic pulmonary hypertension, and convert the sparse coefficient of metabolic pathways into metabolic pathway scores.

[0044] The quantification module is used to quantify clinical indicators in clinical data and obtain a clinical risk score by weighted summation.

[0045] The determination module is also used to determine the weight coefficients corresponding to the target polygenic risk score, metabolic pathway score and clinical risk score based on the clinical parameters in the clinical data, and to determine the risk assessment value by combining the target polygenic risk score, metabolic pathway score and clinical risk score.

[0046] The matching module is used to match the risk assessment value with the preset age-stratified risk threshold range to obtain the risk assessment level information of the target object.

[0047] Thirdly, embodiments of this application provide an electronic device, the device including: a processor, and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the risk assessment method for idiopathic pulmonary hypertension as described in any embodiment of the first aspect.

[0048] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the risk assessment method for idiopathic pulmonary hypertension as described in any embodiment of the first aspect.

[0049] The risk assessment method, system, device, and computer-readable storage medium for idiopathic pulmonary arterial hypertension (IPAH) in this application effectively solves the technical challenges of early risk assessment for IPAH in the general population by integrating multi-gene risk scores, protein expression scores, metabolic pathway scores, and clinical risk scores, and introducing age-stratified dynamic weighting. Specifically, it corrects single nucleotide polymorphism effect values ​​based on a Bayesian framework and optimizes genetic risk prediction by combining linkage disequilibrium information, improving the accuracy of scoring in the context of genetic heterogeneity. It analyzes the biological contributions of protein biomarkers and metabolic pathways through protein quantitative trait loci and sparse coding, capturing early molecular characteristics of the disease from a multi-omics perspective. Simultaneously, it dynamically weights clinical data and integrates it with multi-omics scores, overcoming the limitations of traditional models that rely on post-diagnosis clinical indicators. Finally, through age-stratified threshold matching, it achieves individualized risk assessment, significantly improving the early screening capability for high-risk IPAH groups in the general population, providing a reliable basis for intervention in the prodromal stage of the disease, and improving the accuracy of risk assessment for asymptomatic individuals.

[0050] Furthermore, by dynamically adjusting the weights of clinical risk scores and optimizing the weight allocation of multi-omics scores in conjunction with age and lung function differences, the individualization and accuracy of risk assessment for idiopathic pulmonary hypertension (IPH) in the general population were significantly improved. An initial weight set was matched based on age parameters to adapt to the physiological differences among different age groups; the weights of clinical risk scores were dynamically adjusted according to the difference between lung function measurements and preset thresholds, strengthening the early warning role of clinical indicators in the early stages of lung function abnormalities; simultaneously, by maintaining the total weights of multi-gene and metabolic pathway scores and adjusting the allocation ratios, the contributions of genetic and metabolic characteristics were ensured to flexibly adapt to lung function status. This approach breaks through the dependence of traditional models on clinical indicators after diagnosis, achieving synergistic risk assessment of subtle lung function abnormalities and multiple omics biomarkers in the general population, thus improving the timeliness and accuracy of risk assessment in the prodromal phase of IPH in the general population. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a schematic flowchart of a risk assessment method for idiopathic pulmonary hypertension provided in one embodiment of this application;

[0053] Figure 2 This is a flowchart illustrating a method for determining metabolic pathway scoring according to an embodiment of this application;

[0054] Figure 3This is a schematic diagram of the structure of a risk assessment system for idiopathic pulmonary hypertension provided in one embodiment of this application;

[0055] Figure 4 This is a schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0056] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0057] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0058] Idiopathic pulmonary arterial hypertension (IPAH) is one of the most common subtypes of pulmonary arterial hypertension (PAH). Because IPAH lacks specific symptoms in its early stages, patients are usually diagnosed only when pulmonary vascular obstruction is severe, at which point the optimal intervention window has been missed. Therefore, early risk assessment of IPAH is crucial for identifying high-risk individuals, enabling prodromal intervention, and improving patient survival.

[0059] Currently, existing technologies primarily focus on risk assessment for patients diagnosed with PAH, mainly based on clinical data and using dynamic stratification to predict the progression of pulmonary hypertension. However, methods for predicting the risk of IPAH in the general population remain lacking, and current technologies lack risk assessment methods for IPAH in the general population. Existing PAH risk assessment models rely on clinical indicators after diagnosis, which cannot meet the early screening needs of the general population. Moreover, PAH exhibits significant genetic heterogeneity, and the differences in pathogenic mechanisms among different subtypes limit the predictive efficacy of general existing models. Therefore, existing risk assessment methods suffer from technical problems such as inapplicability to IPAH risk assessment in the general population and poor accuracy in risk assessment.

[0060] In response to the shortcomings of existing technologies, which rely on clinical indicators after diagnosis and cannot capture early risks in the general population, and whose general models are inaccurate due to genetic heterogeneity, the inventors conceived of biological inspirations based on the synergistic regulation of proteomics and metabolomics data, and combined them with basic clinical indicators of the general population to jointly solve the above-mentioned technical problems, and proposed the technical solution of the embodiments of this application.

[0061] To address the problems of the prior art, embodiments of this application provide a method, system, device, and computer-readable storage medium for risk assessment of idiopathic pulmonary hypertension. The risk assessment method for idiopathic pulmonary hypertension provided in this application embodiment is described below.

[0062] Figure 1 A schematic flowchart of a risk assessment method for idiopathic pulmonary hypertension provided in one embodiment of this application is shown. Figure 1 As shown, the method includes steps S110 to S150.

[0063] S110: Obtain single nucleotide polymorphism (SNP) data associated with idiopathic pulmonary hypertension (IPH), as well as protein biomarker data, metabolomics data, and clinical data associated with IPH in the target subjects. Each SNP data includes the corresponding genotype and effect size.

[0064] Single nucleotide polymorphisms (SNPs) refer to variations in a single nucleotide in the genome (such as C-to-T substitutions) and are an important source of genetic diversity. In idiopathic pulmonary arterial hypertension (IPAH) studies, SNPs specifically refer to genetic variant sites associated with PPAH, such as sites in genes reportedly associated with bone morphogenetic protein receptor type 2 (BMPR2) and T-box transcription factor 4 (TBX4). The effect size corresponding to a single nucleotide polymorphism represents the contribution of that SNP to disease risk and can be obtained and quantified based on data published in the world's largest genome-wide association study (GWAS) for PPAH, such as the odds ratio (OR) or regression coefficient β. Single nucleotide polymorphism (SNP) data refers to information on specific nucleotide variant sites that are significantly associated with genetic susceptibility to idiopathic pulmonary arterial hypertension (IPAH), detected through genotyping techniques. Each SNP consists of two parts: genotype (e.g., AA, AG, GG) indicating the genetic variation status of the target individual at that locus; and effect size (e.g., β coefficient or hazard ratio) quantifying the contribution of that SNP to the risk of developing IPAH.

[0065] Protein biomarkers associated with idiopathic pulmonary hypertension (IPAH) refer to proteins and their expression levels directly related to the pathological mechanism of IPAH. These biomarkers can be quantitatively analyzed using high-throughput proteomics techniques such as mass spectrometry or immunoassay, or screened through causal inference and comparative analysis. These biomarkers may include proteins associated with abnormal pulmonary artery endothelial cell function (such as loss of BMPR2 protein expression and elevated endothelin-1 levels) or inflammatory factors (such as IL-6 and TGF-β). Changes in their expression levels can reflect the process of vascular remodeling or fibrosis. Proteins associated with abnormal pulmonary artery endothelial cell function may be characterized by loss of BMPR2 protein expression and elevated endothelin-1 levels, while inflammatory factors may include IL-6 and TGF-β.

[0066] In one embodiment, proteins associated with idiopathic pulmonary hypertension (IPAH) and the protein quantitative trait loci (pQTLs) regulating these proteins—that is, genetic variation sites associated with protein expression levels—can be identified using the following method. To identify proteins associated with IPAH and the protein quantitative trait loci (pQTLs) they regulate, a nested case-control design was first employed, with 200 IPAH cases in each group matched with 5 asymptomatic controls based on age, sex, and BMI. Protein concentrations in baseline plasma samples were detected using high-precision mass spectrometry, and time-resolved difference analysis was performed using a Cox proportional hazards model. After correcting for confounding factors such as smoking history and comorbidities, candidate proteins significantly associated with IPAH pathogenesis were screened. Further, two-sample Mendelian randomized causal inference was performed on the candidate proteins, using protein-specific pQTLs as instrumental variables (IVs). Strong IVs satisfying F-statistic > 10 were screened, and the causal effect of protein concentration on IPAH risk was verified using an inverse variance weighted method, ultimately obtaining a set of proteins with causal predictive value.

[0067] In the pQTL screening phase, genetic loci regulating candidate protein expression were screened based on genome-wide association study (GWAS) data. The association between pQTLs and protein concentrations was required to be genome-wide significant and without linkage disequilibrium. Co-localization analysis confirmed the genomic regional overlap between pQTLs and IPAH risk sites, and horizontal pleiotropic interference was excluded to ultimately identify key regulatory sites driving protein expression. Candidate proteins and their corresponding pQTLs needed to be validated for biological consistency and clinical predictive efficacy in independent cohorts, forming a protein-genetic regulatory network that can explain the pathogenesis of IPAH.

[0068] Metabolomics data refers to the concentration profiles of metabolites in blood or tissues detected using non-targeted or targeted metabolomics techniques such as liquid chromatography-mass spectrometry. This includes lipid metabolites such as lysophosphatidic acid, amino acids such as arginine, and energy metabolism intermediates such as succinic acid. These metabolites may accumulate abnormally or be depleted in IPAH due to mitochondrial dysfunction or hypoxia-induced metabolic reprogramming.

[0069] Clinical data includes physiological indicators, medical history, and demographic information of the target population, which are used to comprehensively assess disease risk and phenotypic heterogeneity. For example, clinical data may include demographic characteristics: age, sex, education level; lifestyle factors: smoking status, alcohol consumption, insomnia symptoms; metabolic risk factors: body mass index, hypertension, diabetes, hyperlipidemia; pulmonary function indicators: peak expiratory flow (PEF), forced expiratory volume in 1 second (FEV1), forced vital capacity (FVC); routine blood tests: white blood cell count (WBC), red blood cell count (RBC), hemoglobin concentration (Hb), platelet count (PLT); blood biochemical indicators: serum creatinine (SCr), serum albumin (ALB), alkaline phosphatase (ALP), aspartate aminotransferase / alanine aminotransferase ratio (AST / ALT ratio), C-reactive protein (C-Reactive Protein). Protein (CRP), sex hormone-binding globulin (SHBG), etc.

[0070] First, significantly associated SNPs were screened from the GWAS Catalog database or published IPAH genetic studies, and the genotypes of target subjects were obtained using genotyping chips such as the Illumina Global Screening Array or targeted sequencing. SNP effect values ​​were standardized to a uniform dimension of risk ratio per allele based on statistical results from multicenter cohorts. Protein biomarker data were obtained using targeted proteomics techniques, such as quantifying serum BMPR2 protein concentration using enzyme-linked immunosorbent assay (ELISA) or detecting inflammation-related proteins using high-throughput proteomics quantification techniques such as mass spectrometry and the Olink platform. Biomarker selection should be based on IPAH pathological mechanism studies, prioritizing proteins validated to be associated with vasoconstriction or endothelial dysfunction. Metabolomics data acquisition required standardized sample pretreatment procedures, such as plasma centrifugation and metabolite extraction, and metabolite concentrations were obtained using high-resolution mass spectrometry such as Q-Exactive HF-X. Metabolite annotation relied on public databases such as HMDB and KEGG, and metabolites related to IPAH pathways such as fatty acid oxidation and the tricarboxylic acid cycle were screened. When acquiring clinical data, it is necessary to use natural language processing to extract key indicators from unstructured texts such as medical records, and to standardize and normalize continuous variables.

[0071] For example, taking a 55-year-old male as the target subject, firstly, single nucleotide polymorphism (SNP) sites significantly associated with idiopathic pulmonary hypertension (IPH) were screened from the GWAS Catalog database, including the rs12345678 site of the BMPR2 gene and the rs98765432 site of the TBX4 gene. Genotyping data of the target subject was detected using a genotyping chip. The rs12345678 site was identified as the GG genotype, and the rs98765432 site as the CT genotype. The effect sizes were standardized β-coefficients of 0.23 and 0.17, respectively, from a multicenter cohort study. Protein biomarker data were quantified using an electrochemiluminescence immunoassay analyzer, with a serum BMPR2 protein concentration of 12.5 ng / mL. Endothelin-1 levels were measured using enzyme-linked immunosorbent assay (ELISA) at 3.8 pg / mL. Simultaneously, the expression level of the inflammatory factor IL-6 was detected using high-throughput mass spectrometry at 15.2 pg / mL. All protein biomarkers were selected from a set of markers that had been validated by Mendelian randomization as having a causal association with IPD.

[0072] Metabolomics data acquisition used ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry to analyze plasma samples, detecting lysophosphatidic acid concentrations of 1.2 μM, arginine concentrations of 85 μM, and succinic acid concentrations of 4.6 μM. Data were annotated to the arachidonic acid metabolic pathway and the tricarboxylic acid cycle pathway using the KEGG database. Clinical data were extracted from the electronic medical record system, including pulmonary function indicators such as forced expiratory volume in one second (FEV1) of 2.8 L and body mass index (BMI) of 26.3 kg / m², and blood biochemical indicators such as C-reactive protein concentration of 5.2 mg / L and sex hormone-binding globulin concentration of 38 nmol / L. A ten-year smoking history and no history of diabetes were also recorded.

[0073] S120: Based on pre-defined linkage disequilibrium reference information, the effect values ​​of single nucleotide polymorphism (SNP) data are corrected using a Bayesian framework, and a polygenic risk score is determined by combining the genotype and corrected effect value of each SNP data.

[0074] Linkage disequilibrium reference information refers to pre-established data on linkage disequilibrium (LD) relationships between single nucleotide polymorphism (SNP) sites in the genome. This serves as a reference genome for calculating linkage disequilibrium and is typically derived from large-scale population genetic databases, such as the European genome map from the 1000 Genomes Project. It includes a correlation coefficient matrix between loci, used to quantify the degree of covariance among different loci. A Bayesian framework refers to a probabilistic model based on Bayesian statistics, which uses prior distributions and likelihood functions to probabilistically infer parameters, here used to correct for biases in SNP effect values. A polygenic risk score (PRS) is a comprehensive genetic risk value calculated by integrating genotypes from multiple genetic loci with corrected effect values, characterizing an individual's genetic susceptibility.

[0075] The implementation process of correcting single nucleotide polymorphism (SNP) effect values ​​and generating polygenic risk scores using a Bayesian framework, based on pre-established linkage disequilibrium reference information, is as follows: First, using the pre-established genomic linkage disequilibrium relationship data in the reference panel, local regions are divided along the genome in a sliding window manner, and a linkage disequilibrium matrix between SNPs in each region is constructed. Then, combining the initial effect values ​​of SNPs and the local linkage disequilibrium matrix, a prior distribution is set using a point-normal mixture Bayesian model, assuming that most site effects are zero and a few follow a normal distribution, and the posterior mean effect value of each site is iteratively estimated using the Markov chain Monte Carlo sampling method. Finally, the corrected posterior mean effect value is multiplied site-by-site with the genotype data of the corresponding site in the target sample, and the product results of all sites are summed to obtain a polygenic risk score reflecting the individual's genetic risk level.

[0076] S130: Based on the expression levels of protein biomarkers in the protein biomarker data and the site effect values ​​of the protein quantitative trait loci that regulate protein biomarkers, a protein expression score is obtained, and a target polygenic risk score is calculated based on the protein expression score and the polygenic risk score.

[0077] Protein quantitative trait loci (pQTLs) are genetic variation sites that regulate the expression levels of specific proteins, altering protein abundance by affecting gene transcription or translation efficiency. Locus effect values ​​represent the regulatory strength of each pQTL on target protein expression, typically quantified from large-scale genetic association studies or Mendelian randomization analyses, such as regression coefficients or standardized effect sizes. Protein expression scores are biological risk values ​​calculated by integrating protein biomarker expression levels and their genetic regulatory effects, reflecting the cumulative contribution of protein abnormalities to disease risk.

[0078] By integrating protein expression levels and their contributions to the genetic regulatory network, a protein expression score is constructed to enhance the biological interpretability of polygenic risk scores. Specifically, for each protein biomarker, its detected expression level is multiplied site-by-site by the effect values ​​of all pQTL sites regulating that protein, and the sum is obtained to obtain the protein's contribution value. The contribution values ​​of all protein biomarkers are further summed to generate a total protein expression score. Subsequently, this score is linearly superimposed with the polygenic risk score obtained in step S120 to generate a target polygenic risk score that integrates genetic variation and proteomics information. The pQTL sites and their effect values ​​require two-stage validation: first, Mendelian randomization analysis based on a large-scale cohort is used to screen genetic sites with causal effects on protein expression; then, co-localization analysis is used to exclude linkage disequilibrium interference with disease risk sites.

[0079] S140: Dimensionality reduction of metabolomics data is performed using sparse coding to determine the sparse coefficients of metabolic pathways associated with idiopathic pulmonary hypertension, and the sparse coefficients are converted into metabolic pathway scores.

[0080] Sparse coding is an unsupervised learning method that represents high-dimensional data as a linear combination of a small number of basis vectors by constructing an overcomplete dictionary, thereby revealing the intrinsic structure of the data. In metabolomics, sparse coding is used to identify key metabolic pathways associated with the pathological mechanisms of idiopathic pulmonary arterial hypertension (IPAH). Databases of metabolic pathways associated with IPAH contain sets of metabolic pathways significantly associated with IPAH pathological processes (such as abnormal energy metabolism and fatty acid oxidation disorders), such as the tricarboxylic acid cycle or lipid metabolism pathways in KEGG. The sparse coefficient of a metabolic pathway is the contribution weight of each pathway to overall metabolic abnormality, selected through a sparse coding algorithm, reflecting the correlation between pathway activity and disease. The metabolic pathway score is a comprehensive quantitative value of the sparse coefficients of all relevant pathways, used to characterize the overall metabolic risk.

[0081] First, based on a predefined IPAH-related metabolic pathway database, a metabolite-pathway association matrix is ​​constructed. Matrix elements represent the degree of metabolite involvement in a specific pathway, such as the weight of an enzyme-catalyzed reaction. The metabolite concentration profile of the target metabolite is multiplied by this matrix to generate an initial pathway activity matrix. Subsequently, a sparse coding algorithm is used to decompose the activity matrix into a pathway weight matrix and a sparse coefficient matrix: the weight matrix characterizes the contribution weight of a metabolite to the pathway, and the sparse coefficient matrix reflects the intensity of pathway activity. The two matrices are iteratively updated using an alternating optimization strategy. L1 norm constraints are applied during sparse coefficient updates to screen key pathways, and consistency with the association matrix is ​​restricted during weight updates. When the change in the weight matrix is ​​less than a convergence threshold, the coefficients of IPAH-related pathways in the sparse coefficient matrix are extracted and accumulated to generate a metabolic pathway score.

[0082] For example, using plasma metabolomics data of the target subjects, abnormally elevated concentrations of metabolites such as lysophosphatidic acid and arginine were detected. An association matrix was constructed based on the IPAH metabolic pathway database, where the fatty acid oxidation pathway contained 5 metabolites and the tricarboxylic acid cycle pathway contained 8 metabolites. Multiplying the metabolite concentrations by the association matrix, the initial pathway activity matrix showed high activity in the fatty acid oxidation pathway. Through sparse coding decomposition, the sparse coefficients for the fatty acid oxidation and tricarboxylic acid cycle pathways were identified as 0.7 and 0.3, respectively, while the coefficients for the remaining pathways were close to zero. The sum of the key pathway coefficients yielded a metabolic pathway score of 1.0, reflecting the risk of metabolic disorder.

[0083] S150: Quantify clinical indicators in clinical data and obtain a clinical risk score by weighted summation.

[0084] Clinical indicators refer to structured parameters extracted from the clinical data of the target subjects that are significantly associated with the risk of idiopathic pulmonary arterial hypertension (IPAH). These include physiological indicators such as pulmonary function measurements and blood biochemical indicators, medical history such as cardiovascular disease history, and demographic characteristics such as age and sex. The clinical risk score is a comprehensive score obtained by standardizing and quantifying these indicators, assigning weights according to their association strength with IPAH risk, and then summing the results. It reflects the disease risk level of the target subjects based on their clinical phenotype.

[0085] First, key clinical indicators, such as age and pulmonary function parameters, are screened based on the pathological mechanisms of IPAH and clinical guidelines. Continuous variables such as age are scored according to predefined intervals, categorical variables such as medical history are coded using binary codes, and numerical indicators such as pulmonary function parameters are standardized and mapped to risk contribution using Z-scores. Next, initial weights are obtained through regression analysis in a multi-center cohort, for example, by using a logistic regression model to obtain the regression coefficients for each indicator. Finally, the quantified clinical indicators are multiplied by their corresponding weight coefficients and summed to generate a clinical risk score.

[0086] S160: Based on the clinical parameters in the clinical data, determine the weight coefficients corresponding to the target polygenic risk score, metabolic pathway score, and clinical risk score, and combine the target polygenic risk score, metabolic pathway score, and clinical risk score to determine the risk assessment value.

[0087] The risk assessment value refers to the comprehensive quantitative result after integrating the target multigene risk score, metabolic pathway score and clinical risk score through a dynamic weighting model, reflecting the overall risk level of the target subject to idiopathic pulmonary arterial hypertension (IPAH).

[0088] First, a pre-defined initial weight set is matched based on clinical parameters such as the age of the target subjects. The baseline weights for the polygenic risk score, metabolic pathway score, and clinical risk score are determined through regression analysis of a multi-center cohort. Then, the weights of each score are dynamically adjusted according to specific clinical parameters such as the degree of lung function abnormality: if the correlation between a certain risk characteristic, i.e., abnormal clinical indicators, and disease increases, its weight ratio is increased, while the total weight of other scores remains unchanged to maintain model stability. Finally, the adjusted weight coefficients are multiplied by the corresponding scores and summed to generate a comprehensive risk assessment value.

[0089] S170: Match the risk assessment value with the preset age-stratified risk threshold range to obtain the risk assessment level information of the target object.

[0090] Age-stratified risk threshold ranges refer to the pre-defined scoring cutoff values ​​corresponding to different risk levels based on the target population's age range. These threshold ranges are determined through multi-center cohort studies analyzing the distribution characteristics of risk assessment values ​​in healthy individuals and IPAH patients of different age groups, combined with survival analysis to identify the critical points distinguishing low, medium, and high risk states within each age group. Risk assessment level information is the disease risk stratification based on the threshold range within which the target population's assessment value falls, typically including low, medium, and high risk levels, used to guide clinical intervention decisions.

[0091] First, based on the age information of the target population, pre-defined age stratification rules are matched, such as dividing the population into three age groups: youth, middle-aged, and elderly. Each age group corresponds to an independent risk threshold range. By analyzing the distribution of risk assessment values ​​among people of the same age group in historical cohorts, the cutoff values ​​for risk levels are determined. Specifically, based on the 50th and 75th percentiles of the predicted probability of IPAH occurrence over 10 years, individuals are divided into low-risk, medium-risk, and high-risk groups. The comprehensive risk assessment value of the target population is compared with the threshold range of the corresponding age group. If the assessment value is below the low-risk threshold, it is classified as Level 1 risk; if it is between the low and high-risk thresholds, it is classified as Level 2 risk; and if it exceeds the high-risk threshold, it is classified as Level 3 risk. The threshold range is dynamically updated by periodically incorporating new case data and recalibrating the cutoff points using a sliding window method.

[0092] For example, according to age stratification rules, the preset risk threshold range for the middle-aged group (45-59 years old) is below 8.0 for low risk, 8.0-10.0 for medium risk, and above 10.0 for high risk. This threshold is based on the probability distribution of 10-year IPAH incidence risk in multicenter cohort studies, combined with survival analysis to determine the 50th percentile for low-risk individuals and the 75th percentile for high-risk individuals. The target subject's comprehensive risk assessment value is 9.38, falling within the medium-risk threshold range for the middle-aged group, and is therefore classified as a level 2 medium-risk individual, indicating the need for enhanced regular pulmonary function monitoring and lifestyle interventions. The threshold range is updated annually using a sliding window method incorporating the latest clinical data. For example, when new case data indicates a shift in the risk distribution of the middle-aged group, the cutoff values ​​are dynamically adjusted to 8.2 and 10.3 to ensure the timeliness and accuracy of risk stratification.

[0093] This application presents a risk assessment method, system, device, and computer-readable storage medium for idiopathic pulmonary arterial hypertension (IPAH). By integrating multi-gene risk scores, protein expression scores, metabolic pathway scores, and clinical risk scores, and introducing age-stratified dynamic weighting, it effectively addresses the technical challenges of early risk assessment for IPAH in the general population. Specifically, it corrects single nucleotide polymorphism effect values ​​based on a Bayesian framework and optimizes genetic risk prediction by combining linkage disequilibrium information, improving scoring accuracy in the context of genetic heterogeneity. It analyzes the biological contributions of protein biomarkers and metabolic pathways through protein quantitative trait loci and sparse coding, capturing early molecular characteristics of the disease from a multi-omics perspective. Simultaneously, it dynamically weights clinical data and integrates it with multi-omics scores, overcoming the limitations of traditional models that rely on post-diagnosis clinical indicators. Finally, through age-stratified threshold matching, it achieves individualized risk assessment, significantly improving the early screening capability for high-risk IPAH populations in the general population, providing a reliable basis for prodromal intervention, and improving the accuracy of risk assessment for asymptomatic individuals.

[0094] In one feasible implementation, the method further includes: acquiring a set of protein biomarkers associated with idiopathic pulmonary hypertension (IPH) from different age groups in the general population; performing time series analysis based on the expression levels of protein biomarkers in each age group in the protein biomarker set to obtain protein biomarker age parameters characterizing the trend of changes in the expression levels of protein biomarkers associated with IPH with age; and weighting the protein expression score using the protein biomarker age parameters to obtain an adjusted protein expression score.

[0095] The protein biomarker dataset refers to a dataset of protein expression levels related to IPAH in the general population across different age groups, collected through a longitudinal cohort. It includes age-stratified biological sample testing results from healthy individuals and IPAH-susceptible populations. The protein biomarker age parameter refers to the parameters of a quantitative model established through time-series analysis that measures the changes in protein expression levels with age, characterizing the baseline expression trend of a specific protein during normal aging, such as a linear growth coefficient or an exponential decay coefficient.

[0096] Age-specific calibration can optimize protein risk contribution. First, a protein expression database covering the entire age range of 20-80 years is established, grouped into intervals of 10 years, with each group containing protein biomarker detection data from at least 100 samples. A mixed-effects model is used for time series modeling, with age as a fixed effect and individual differences as a random effect, to fit the curves of protein expression levels versus age, extracting age-related slope or curvature parameters as age parameters. Then, the protein expression score of the target individual is compared with the parameter corresponding to their age: if the actual expression level significantly deviates from the expected level for age (e.g., exceeding the expected value by two standard deviations), the original score is weighted and amplified by the parameter ratio to generate an age-corrected protein expression score. In this embodiment, the adjusted protein expression score is used to calculate the target polygenic risk score based on the adjusted protein expression score and the polygenic risk score.

[0097] For example, an expression dataset of protein biomarkers related to idiopathic pulmonary hypertension was first obtained from a general population cohort stratified between 20 and 80 years of age. For the BMPR2 protein, mixed-effects model analysis revealed a linear decrease in serum concentration with age, with a time-series model fitting yielding an age parameter of 1 pg / ml decrease per year. Based on this parameter, the expected expression level for the 55-year-old age group was calculated to be 95 pg / ml, while the actual measured value in the target subjects was 70 pg / ml, a deviation of 26% from the expected value. An age-weighted adjustment factor of 0.74 was applied to the original contribution value of BMPR2 based on this deviation, correcting the original contribution value of 5.0 to 5.0 × 0.74 = 3.7. Similarly, the age parameter for endothelin-1 protein showed an exponential increase in expression after age 50, with an expected value of 3.2 pg / ml at age 55, while the measured value of 3.8 pg / ml exceeded the expectation by 18%. An adjustment factor of 1.18 was used to increase the original contribution value from 1.18 to 1.39. The age-related model of IL-6 protein showed no significant association between its expression level and age, so the original contribution value of 4.71 was kept unchanged. The adjusted contribution values ​​of all protein markers after being weighted by the age parameter were summed to obtain the corrected total protein expression score of 3.7 + 1.39 + 4.71 = 9.8.

[0098] This embodiment establishes an age-specific expression baseline by acquiring protein biomarker datasets from different age groups in the general population, effectively overcoming the assessment bias caused by neglecting the natural changes in protein expression with age in traditional methods. By quantifying the age-related trends of protein biomarkers based on time-series analysis, the protein expression scores of target subjects are dynamically calibrated, significantly reducing the risk of false positives due to physiological aging. For example, considering the linear decline in BMPR2 protein expression with age, the deviation of actual detected values ​​from expected values ​​is corrected using age parameters, accurately distinguishing between pathological abnormalities and natural attenuation. This improves the specificity of protein biomarkers in IPAH risk assessment, making it particularly suitable for early screening in the general population and providing a reliable basis for risk stratification of asymptomatic individuals.

[0099] In one feasible implementation, step S130: Based on the expression level of protein biomarkers in the protein biomarker data and the site effect value of the protein quantitative trait loci of the preset regulatory protein biomarkers, a protein expression score is obtained, and a target polygenic risk score is calculated based on the protein expression score and the polygenic risk score, including: multiplying the expression level of the protein biomarker and the site effect value of each protein quantitative trait locus of the regulatory protein biomarker and summing them to obtain the contribution value of each protein biomarker; summing the contribution values ​​of all protein biomarkers to obtain the protein expression score; and directly adding the polygenic risk score to the protein expression score to obtain the target polygenic risk score.

[0100] The contribution value of a protein biomarker is a quantitative indicator obtained by multiplying and summing the expression level of a specific protein in the target object with the effect value of its regulatory site (pQTL).

[0101] First, for each protein biomarker, the corresponding regulatory pQTL site is identified. The correspondence between a protein and the pQTL site that regulates its expression is determined by pQTL analysis, which shows a significant correlation between protein expression levels and genetic variations. Combined with Mendelian randomization verification, the causal genetic sites driving changes in the expression of specific proteins are identified.

[0102] The algorithm iterates through the regulatory pQTL sites corresponding to each protein biomarker. It multiplies the actual expression level of the protein corresponding to the genotype of the target protein in the protein biomarker data with the site effect value of each pQTL site. The products of all sites are then summed to generate the protein's contribution value. For example, if a protein is regulated by two pQTLs with an expression level of X and effect values ​​of β1 and β2 at the two sites, the contribution value is X × (β1 + β2). Subsequently, the contribution values ​​of all protein biomarkers are summed to obtain the total protein expression score. Finally, the polygenic risk score calculated after correcting for SNP effect values ​​using a Bayesian framework is standardized in dimensions and then added to the protein expression score to form the target polygenic risk score. The formula for calculating the target polygenic risk score is as follows:

[0103] (1)

[0104] in, Indicates the target polygenic risk score; This indicates that the baseline polygenic risk score was calculated after correcting for SNP effect values ​​using a Bayesian framework. This represents the total regulatory effect value of the k-th protein biomarker, which is the sum of the effect values ​​of all pQTL sites regulating that protein. , where n is the number of pQTLs that regulate this protein; This represents the standardized expression level of the k-th protein biomarker.

[0105] For example, the process of calculating a target polygenic risk score based on protein biomarker data is as follows: First, protein biomarkers associated with idiopathic pulmonary hypertension and their corresponding pQTL sites are extracted from a pre-defined protein-pQTL regulatory relationship database. For instance, BMPR2 protein is regulated by two pQTLs, rs12345678 and rs98765432, with site effect values ​​of 0.23 and 0.17, respectively. The serum BMPR2 protein expression level in the target subject is 12.5 ng / mL. When calculating the BMPR2 contribution value, the expression level is multiplied by the sum of the effect values ​​of each pQTL to obtain 12.5 × (0.23 + 0.17) = 5.0. Similarly, endothelin-1 protein is regulated by rs11223344 with an effect value of 0.31. The detected expression level is 3.8 pg / mL, and the contribution value is 3.8 × 0.31 = 1.18. IL-6 protein is jointly regulated by rs55667788 and rs66778899, with effect values ​​of 0.19 and 0.12, respectively. The expression level detected by mass spectrometry is 15.2 pg / mL, and the contribution value is 15.2 × (0.19 + 0.12) = 4.71. The sum of the contribution values ​​of all protein markers gives a total protein expression score of 5.0 + 1.18 + 4.71 = 10.89.

[0106] The protein expression score was then integrated with the Bayesian-corrected polygenic risk score. Assuming that the polygenic risk score was calculated to be 6.5 after correcting for the effect sizes of rs12345678 and rs98765432, the two scores were standardized to eliminate unit differences and then directly added together. The final target polygenic risk score was 6.5 + 10.89 = 17.39.

[0107] Figure 2 A schematic flowchart of a method for determining metabolic pathway scoring according to an embodiment of this application is shown. Figure 2 As shown, the method includes steps S210 to S240.

[0108] In one feasible implementation, step S140: dimensionality reduction of the metabolomics data is performed using sparse coding to determine the sparse coefficients of metabolic pathways associated with idiopathic pulmonary hypertension, and the sparse coefficients are converted into metabolic pathway scores, including:

[0109] S210: Based on a pre-defined database of metabolic pathways related to idiopathic pulmonary hypertension, construct a metabolic association matrix. The rows and columns of the association matrix correspond to metabolic pathways related to idiopathic pulmonary hypertension and metabolites in metabolomics data, respectively.

[0110] The Idiopathic Pulmonary Hypertension (IPH)-Related Metabolic Pathways Database is a predefined collection of metabolic pathways significantly associated with disease pathogenesis mechanisms such as abnormal energy metabolism and fatty acid oxidation disorders, for example, the tricarboxylic acid cycle pathway in KEGG. The metabolic association matrix is ​​a two-dimensional matrix where rows represent predefined metabolic pathways, columns represent metabolites from metabolomics data, and matrix elements represent the participation weight of metabolites in specific pathways, such as the intensity of enzymatic reactions based on reaction pathways.

[0111] The metabolic association matrix is ​​a two-dimensional matrix. Its rows correspond to predefined metabolic pathways associated with the pathological mechanisms of IPAH, such as the tricarboxylic acid cycle and fatty acid oxidation. The columns correspond to metabolites detected in metabolomics data, such as lysophosphatidic acid and succinic acid. Matrix elements are quantified using the participation relationships of metabolites in biological pathway databases such as KEGG. For example, if a metabolite is a substrate or product of an enzymatic reaction in a pathway, the corresponding element is assigned a value of 1; if a metabolite plays a key regulatory role in the pathway, it is assigned a higher value. The matrix is ​​constructed based on literature evidence and pathway annotation tools, such as MetaboAnalyst, to ensure that the association strength between each metabolite and the pathway reflects its biological importance.

[0112] S220: Multiply the concentration values ​​of each metabolite in the metabolomics data with the correlation matrix to generate an initial metabolic pathway activity matrix, and decompose the initial metabolic pathway activity matrix into a metabolic pathway weight matrix and a sparse coefficient matrix.

[0113] The initial metabolic pathway activity matrix is ​​an intermediate matrix generated by multiplying the metabolite concentration spectrum with the metabolic correlation matrix, characterizing the initial activity level of each pathway. The metabolic pathway weight matrix and the sparse coefficient matrix are two sub-matrices obtained through matrix factorization; the former describes the contribution weight of metabolites to the pathway, and the latter reflects the sparse distribution of pathway activity intensity.

[0114] The metabolite concentrations of the target sample are standardized to Z-scores and used as row vectors. These vectors are then multiplied with the metabolic association matrix to generate an initial metabolic pathway activity matrix. Each row of this matrix represents a sample, and each column represents the initial activity value of a pathway. For example, if a sample has a lysophosphatidylcholine concentration of 1.2 μM and a succinic acid concentration of 1.2 μM, belonging to the fatty acid oxidation and tricarboxylic acid cycle pathways respectively, then the activity value of the corresponding pathway is the sum of the products of the two metabolite concentrations and their association weights.

[0115] Subsequently, nonnegative matrix factorization (NMF) was used to decompose the initial activity matrix into two sub-matrices: a metabolic pathway weight matrix, which describes the contribution weight of each metabolite to the pathway, with the matrix dimension being the number of metabolites multiplied by the number of pathways; and a sparse coefficient matrix, which characterizes the activity intensity of each pathway in each sample, with the matrix dimension being the number of samples multiplied by the number of pathways. Key pathways were screened through sparsity constraints.

[0116] S230: The metabolic pathway weight matrix and sparse coefficient matrix are iteratively updated using an alternating direction optimization method. In each iteration, when the metabolic pathway weight matrix is ​​fixed, the L1 norm constraint is applied to the sparse coefficient matrix to determine the key metabolic pathways. When the sparse coefficient matrix is ​​fixed, the update direction of the metabolic pathway weight matrix is ​​limited by the correlation matrix.

[0117] Alternating direction optimization is an iterative optimization algorithm that alternately fixes one matrix and updates another to approximate the optimal solution. The L1 norm constraint is a regularization term imposed on the sparse coefficient matrix, forcing sparsity to filter critical pathways.

[0118] The Alternating Directional Multiplier (ADMM) method is used to iteratively optimize the weight matrix and sparse coefficient matrix. First, the weight matrix is ​​fixed, and the sparse coefficient matrix is ​​updated: In each iteration, the weight matrix is ​​fixed, and the sparse coefficient matrix is ​​updated by minimizing the reconstruction error (the difference between the initial matrix and the product of the weight matrix and the sparse coefficient matrix) while applying L1 norm constraints. L1 regularization forces most pathway coefficients to approach zero, retaining only the significant coefficients of a few key pathways. For example, the optimization objective function is:

[0119] (2)

[0120] Where X is the initial activity matrix, W is the weight matrix, S is the sparse coefficient matrix, λ is the regularization strength parameter, and ∥⋅∥F represents the Frobenius norm of the matrix.

[0121] Simultaneously, the sparse coefficient matrix is ​​fixed, and the weight matrix is ​​updated using the projection gradient method, ensuring that it is non-negative and consistent with the biological relationships in the metabolic association matrix. For example, the update direction of the weight matrix is ​​restricted to assigning weights to metabolites only in their known associated pathways. Finally, convergence is checked: the Frobenius norm difference between two iterations of the weight matrix is ​​calculated; if the difference is less than a preset threshold, the iteration is terminated.

[0122] S240: When the update difference of the metabolic pathway weight matrix is ​​less than the preset convergence threshold, the sparse coefficients corresponding to the metabolic pathways included in the metabolic pathway database are selected from the sparse coefficient matrix, and the metabolic pathway score is obtained by summing the sparse coefficients corresponding to all metabolic pathways included in the metabolic pathway database.

[0123] A preset convergence threshold is the condition for the algorithm to stop iterating. It is typically set to a value where the difference between the weight matrix before and after the update is less than a preset specific value, such as 1e-5. Sparse coefficients corresponding to preset metabolic pathways are extracted from the converged sparse coefficient matrix. For example, the coefficient for the fatty acid oxidation pathway is 0.7, and the coefficient for the tricarboxylic acid cycle pathway is 0.3. The sparse coefficients of all pathways are summed to obtain the metabolic pathway score. For example, in the sparse coefficient matrix of a sample, the coefficients for key pathways are 0.7 and 0.3, and the coefficients for the remaining pathways are 0; then the metabolic pathway score is 0.7 + 0.3 = 1.0. This score reflects the overall degree of abnormality in the metabolic pathway; a higher value indicates a greater risk of metabolic disorder.

[0124] For example, a metabolic correlation matrix is ​​constructed based on a pre-defined IPAH-related metabolic pathway database, where rows correspond to the tricarboxylic acid cycle and fatty acid oxidation pathways, and columns correspond to detected metabolites such as lysophosphatidic acid, succinic acid, and arginine. Matrix elements are assigned weights according to KEGG pathway annotations; for example, lysophosphatidic acid has a weight of 1.2 in the fatty acid oxidation pathway, and succinic acid has a weight of 0.8 in the tricarboxylic acid cycle pathway. The metabolite concentrations of the target objects are standardized and multiplied by the correlation matrix to generate an initial activity matrix. For example, the product of a lysophosphatidic acid concentration of 1.2 μM and a weight of 1.2 contributes 1.44 to the fatty acid oxidation pathway activity, and the product of a succinic acid concentration of 4.6 μM and a weight of 0.8 contributes 3.68 to the tricarboxylic acid cycle activity. The initial activity matrix is ​​decomposed using an alternating direction optimization method. When fixing the metabolic pathway weight matrix, an L1 regularization constraint is applied to the sparse coefficient matrix, selecting a fatty acid oxidation pathway coefficient of 0.7 and a tricarboxylic acid cycle coefficient of 0.3, and setting the coefficients of the remaining pathways to zero. After iteratively updating the weight matrix until the difference is less than a preset threshold, the sparse coefficients of key pathways are extracted and summed to obtain a metabolic pathway score of 0.7 + 0.3 = 1.0. This score reflects the degree of abnormality in the target's metabolic pathways and will serve as an important input parameter for subsequent risk assessment models.

[0125] In one feasible implementation, clinical parameters include age and pulmonary function measurements. Based on the clinical parameters in the clinical data, weight coefficients are determined for the target polygenic risk score, metabolic pathway score, and clinical risk score. A risk assessment value is then determined by combining these three scores, including: matching a preset initial weight set based on the age parameter, where the initial weight set includes baseline weights for the target polygenic risk score, metabolic pathway score, and clinical risk score; adjusting the baseline weights of the clinical risk score in the initial weight set based on the difference between the pulmonary function measurements and preset measurement thresholds, with the adjustment magnitude proportional to the difference; keeping the sum of the baseline weights of the target polygenic risk score and metabolic pathway score constant, readjusting the allocation ratio of the baseline weights of the target polygenic risk score and metabolic pathway score, with the allocation ratio of the baseline weights of the target polygenic risk score proportional to the difference; and finally, based on the adjusted baseline weights corresponding to the target polygenic risk score, metabolic pathway score, and clinical risk score, weighted summing is performed on each score to generate the risk assessment value.

[0126] The initial weight set refers to the pre-defined baseline weight allocation ratios for polygenic risk scores, metabolic pathway scores, and clinical risk scores based on different age groups. This set is determined based on the independent predictive power of each score for IPAH risk and age-related interaction effects analysis from multicenter cohort studies. For example, genetic scores may be given higher weights in younger populations, while clinical scores may be given increased weights in older populations. Pulmonary function measurements are quantitative indicators obtained through pulmonary function testing, such as forced expiratory volume in one second (FEV1), forced vital capacity (FVC), or the FEV1 / FVC ratio, reflecting the degree of airway obstruction or restrictive ventilatory dysfunction. Preset measurement thresholds refer to normal reference ranges for pulmonary function set based on clinical guidelines or large-scale healthy population statistical data. For example, an FEV1 / FVC ≥ 0.7 is the lower limit of normal; values ​​below this indicate abnormal pulmonary function.

[0127] First, an initial weight set is matched, and a preset age stratification weight rule is applied based on the age parameter of the target subjects. For example, the age stratification is: youth (≤44 years), middle-aged (45-59 years), and elderly (≥60 years). The initial weight set for each stratum is determined by analyzing the risk contribution patterns of different age groups through historical data analysis, ensuring that the weight allocation conforms to age-related pathological mechanisms. The youth stratum may be assigned a weight of 0.5 for the polygenic risk score, 0.3 for the metabolic score, and 0.2 for the clinical score; the middle-aged stratum is adjusted to a weight of 0.4 for the polygenic score, 0.3 for the metabolic score, and 0.3 for the clinical score; and the elderly stratum is assigned weights of 0.3, 0.2, and 0.5.

[0128] Next, the baseline weights of the clinical risk score are adjusted by calculating the difference between the pulmonary function measurement value FEV1=75% and the preset measurement threshold of 80% and -5%. The weight of the clinical risk score is then increased proportionally to the absolute value of the difference (5%). For example, for every 1% deviation from the threshold, the weight increases by 0.02, adjusting the clinical score weight from 0.3 to 0.3 + 5 × 0.02 = 0.4, resulting in a total weight of 0.6 for the polygenic and metabolic scores. Simultaneously, the weight allocation for the polygenic and metabolic scores is adjusted, maintaining the total weight of 0.6, and the weights are redistributed according to the proportion of the pulmonary function difference. For example, a 5% difference corresponds to an increase of 0.05 in the polygenic weight (adjusted to 0.45) and a decrease of 0.05 in the metabolic weight (adjusted to 0.15), ensuring the total remains 0.6.

[0129] Finally, a weighted summation is performed to generate a risk assessment value. The adjusted weights (0.45 for multigene, 0.15 for metabolism, and 0.4 for clinical) are then weighted and summed with the corresponding multigene risk score, metabolic pathway score, and clinical risk score to obtain a comprehensive risk assessment value.

[0130] For example, taking a 55-year-old male as the target subject, firstly, an initial weight set for the middle-aged group is matched according to the age parameter: a baseline weight of 0.4 for the polygenic risk score, 0.3 for the metabolic pathway score, and 0.3 for the clinical risk score. The target subject's forced expiratory volume in one second (FEV1) is 75%, which is 5% different from the preset threshold of 80%. According to the rule that the clinical weight increases by 0.02 for every 1% deviation from the threshold, the baseline weight of the clinical risk score is increased to 0.4. Keeping the total weight of the polygenic and metabolic scores unchanged at 0.6, the allocation is adjusted proportionally according to the difference: the polygenic weight increases by 0.05 to 0.45, and the metabolic weight decreases by 0.05 to 0.15. Assuming the target polygenic risk score is 17.39, the metabolic pathway score is 1.0, and the clinical risk score is 3.5, then the weighted risk assessment value is 17.39 × 0.45 + 1.0 × 0.15 + 3.5 × 0.4 = 9.38.

[0131] In one feasible implementation, step S120: Based on preset linkage disequilibrium reference information, the effect value of single nucleotide polymorphisms is corrected using a Bayesian framework, and a polygenic risk score is determined by combining the genotype of each single nucleotide polymorphism with the corrected effect value, including:

[0132] Based on the preset linkage disequilibrium reference information, the genotype data region of single nucleotide polymorphisms in the reference panel of the linkage disequilibrium reference information is divided by windowing, and a local linkage disequilibrium matrix is ​​constructed.

[0133] Sliding window partitioning of genotype data refers to dividing the genomic data in the reference panel into continuous and partially overlapping intervals based on physical location. For example, each window covers approximately 2 million base pairs, and each window contains a set of neighboring single nucleotide polymorphism (SNP) sites. The local linkage disequilibrium matrix is ​​a symmetric matrix that describes the genotype correlations among all SNP sites within the same window. The matrix elements are the correlation coefficients between pairs of SNPs, reflecting their degree of covariation.

[0134] Based on a reference panel, such as the European population data from the 1000 Genomes Project, regions are divided along the genome using a sliding window approach to ensure the physical location of SNPs within each window is continuous. For the SNP genotype data within each window, Pearson correlation coefficients are calculated between every two genotypes to generate a local linkage disequilibrium (LD) matrix. The window size is adaptively adjusted by default based on genome density to ensure coverage of local linkage disequilibrium (LD) structures. For example, in idiopathic pulmonary arterial hypertension (IPAH) risk assessment, the European population reference panel from the 1000 Genomes Project is used to divide the genome into multiple 2 Mb windows. The SNP correlation coefficient matrix for IPAH-related gene regions such as BMPR2 and TBX4 within each window is calculated as the basis for subsequent LD correction.

[0135] Based on the effect values ​​of single nucleotide polymorphisms and the local linkage disequilibrium matrix, the posterior mean effect value of each single nucleotide polymorphism is calculated using a point normal mixture Bayesian model and Markov chain Monte Carlo sampling.

[0136] A point-normal mixture Bayesian model is a statistical model that assumes most SNP effects are zero and a minority follow a normal distribution, used to describe the sparsity of genetic effects. Markov chain Monte Carlo sampling (MCMC) is an iterative algorithm that approximates the posterior distribution through random sampling to estimate the Bayesian posterior mean of SNP effects.

[0137] First, prior settings are established. Heritability h² is estimated using LD score regression based on GWAS summary statistics, characterizing the overall disease heritability explained by the SNP. The causal marker proportion p refers to the proportion of SNPs with hypothetical non-zero effects, optimized in validation data through isogrid search. Then, model iterations are performed, including MCMC initialization, randomly assigning SNP effect values ​​to zero or non-zero states, and setting initial parameters (h², p). Gibbs sampling is used to update the effect value state (zero or non-zero) for each SNP, adjusting the posterior distribution based on the local LD ​​matrix and marginal effect values. Convergence is confirmed by parameter stability after multiple iterations. For example, for IPAH-related SNPs, h² is estimated to be 0.15 using GWAS statistics, and the causal marker proportion p is set to 0.01. Through 100 MCMC iterations, updating the effect state of SNPs in the BMPR2 gene region each time, a stable posterior distribution is finally obtained through convergence.

[0138] The posterior mean effect value is used as the corrected effect value for each single nucleotide polymorphism; the corrected effect value of each single nucleotide polymorphism is multiplied by the corresponding genotype and summed to generate a polygenic risk score.

[0139] The posterior mean effect size (PRS) is the expected value of the SNP effect calculated using a Bayesian model. It integrates information from prior assumptions and observational data, eliminating bias caused by the LD. The mean of the effect size for each SNP is calculated from the MCMC sampling results and used as the corrected effect size to obtain the corrected effect size for each single nucleotide polymorphism (SNP). Then, the SNP genotypes of the target population are converted to allele doses, such as AA=0, AG=1, GG=2. The genotype and corrected effect size are multiplied for each SNP, and the sum is obtained to obtain the PRS. The PRS is then Z-score standardized according to the population distribution for easy risk stratification. For example, the genotype of the BMPR2 gene rs234567 locus associated with IPAH is AG, i.e., coded as 1, with a corrected effect size of 0.12 and a risk contribution of 0.12. The PRS is calculated after accumulating the contributions of all SNPs.

[0140] For example, the linkage disequilibrium structure of the target genotyping region was first extracted from the European population reference panel of the 1000 Genomes Project. A 2 Mb sliding window was used to divide the SNP sites in the BMPR2 gene region, constructing a local linkage disequilibrium matrix. For instance, the window contained two SNPs, rs12345678 and rs98765432, with a genotypic correlation coefficient of 0.65, forming a matrix describing local genetic covariance. Based on the heritability parameters estimated by GWAS abstract statistics and the preset causal marker ratio, Markov chain Monte Carlo sampling was performed using a point-normal mixture Bayesian model. After 100 iterations of sampling at the rs12345678 site in the BMPR2 gene region, its posterior mean effect size was corrected from the initial 0.23 to 3.0; the effect size at the rs98765432 site of the TBX4 gene was adjusted from 0.44 to 0.5. The corrected effect values ​​are multiplied by the genotype of the target subject and summed. For example, the rs12345678 genotype GG is encoded as 2, and its contribution value is 2 × 3.0 = 6.0; the rs98765432 genotype CT is encoded as 1, and its contribution value is 1 × 0.5 = 0.5. The sum of the contribution values ​​of all SNPs is 6.5, which directly generates a polygenic risk score, serving as the basis for integrating subsequent protein expression scores and metabolic pathway scores.

[0141] In one embodiment, the polygenic risk score for determining idiopathic pulmonary hypertension described above can be calculated using the LDpred algorithm of the prior art.

[0142] In one feasible implementation, clinical data includes physiological indicators, medical history, and demographic information. Clinical data, including the target subject's physiological indicators, medical history, and demographic information, is used to comprehensively assess disease risk and phenotypic heterogeneity. Clinical data may include demographic characteristics: age, sex, education level; lifestyle factors: smoking status, alcohol consumption, insomnia symptoms; metabolic risk factors: body mass index, hypertension, diabetes, hyperlipidemia; pulmonary function indicators: peak expiratory flow rate, forced expiratory volume in one second, forced vital capacity; routine blood tests: white blood cell count, red blood cell count, hemoglobin concentration, platelet count; and blood biochemical indicators: serum creatinine, serum albumin, alkaline phosphatase, aspartate aminotransferase / alanine aminotransferase ratio, C-reactive protein, sex hormone-binding globulin, etc.

[0143] Based on the same concept, this application provides a risk assessment system for idiopathic pulmonary hypertension, which is described below in conjunction with... Figure 3 The risk assessment system for idiopathic pulmonary hypertension provided in the embodiments of this application will be described in detail.

[0144] Figure 3This is a structural block diagram of a risk assessment system for idiopathic pulmonary hypertension, as illustrated in an embodiment of this application. Figure 3 As shown, the risk assessment system for idiopathic pulmonary hypertension may include:

[0145] The acquisition module 310 is used to acquire single nucleotide polymorphism data related to idiopathic pulmonary hypertension, as well as protein biomarker data, metabolomics data and clinical data related to idiopathic pulmonary hypertension of the target subjects. Each single nucleotide polymorphism data includes the corresponding genotype and effect value.

[0146] The determination module 320 is used to correct the effect value of single nucleotide polymorphism data using a Bayesian framework based on preset linkage disequilibrium reference information, and to determine the polygenic risk score by combining the genotype of each single nucleotide polymorphism data with the corrected effect value.

[0147] The calculation module 330 is used to obtain a protein expression score based on the expression level of protein markers in the protein marker data and the site effect value of the protein quantitative trait loci that regulate protein markers, and to calculate the target multigene risk score based on the protein expression score and the multigene risk score.

[0148] The determination module 320 is also used to reduce the dimensionality of metabolomics data through sparse coding, determine the sparse coefficient of metabolic pathways related to idiopathic pulmonary hypertension, and convert the sparse coefficient of metabolic pathways into metabolic pathway scores.

[0149] The quantification module 340 is used to quantify clinical indicators in clinical data and obtain a clinical risk score by weighted summation.

[0150] The determination module 320 is also used to determine the weight coefficients corresponding to the target polygenic risk score, metabolic pathway score and clinical risk score based on the clinical parameters in the clinical data, and to determine the risk assessment value by combining the target polygenic risk score, metabolic pathway score and clinical risk score.

[0151] The matching module 350 is used to match the risk assessment value with the preset age-stratified risk threshold range to obtain the risk assessment level information of the target object.

[0152] In one embodiment, the calculation module 330 is further configured to acquire a set of protein biomarkers related to idiopathic pulmonary hypertension for different age groups in the general population; perform time series analysis based on the expression levels of protein biomarkers for each age group in the protein biomarker set to obtain protein biomarker age parameters that characterize the changing trend of protein biomarker expression levels related to idiopathic pulmonary hypertension with age; and obtain an adjusted protein expression score by weighting the protein expression score using the protein biomarker age parameters.

[0153] In one embodiment, the calculation module 330 is specifically used to multiply the expression level of the protein marker and the site effect value of each protein quantitative trait locus regulating the protein marker and then sum them up to obtain the contribution value of each protein marker; to sum up the contribution values ​​of all protein markers to obtain the protein expression score; and to directly add the multigene risk score to the protein expression score to obtain the target multigene risk score.

[0154] In one embodiment, the determining module 320 is specifically used to construct a metabolic correlation matrix based on a preset idiopathic pulmonary hypertension-related metabolic pathway database. The rows and columns of the correlation matrix correspond to the metabolic pathways related to idiopathic pulmonary hypertension and the metabolites in the metabolomics data, respectively. The concentration value of each metabolite in the metabolomics data is multiplied by the correlation matrix to generate an initial metabolic pathway activity matrix, and the initial metabolic pathway activity matrix is ​​decomposed into a metabolic pathway weight matrix and a sparse coefficient matrix. The metabolic pathway weight matrix and the sparse coefficient matrix are iteratively updated using an alternating direction optimization method. In each iteration, when the metabolic pathway weight matrix is ​​fixed, an L1 norm constraint is applied to the sparse coefficient matrix to determine key metabolic pathways. When the sparse coefficient matrix is ​​fixed, the update direction of the metabolic pathway weight matrix is ​​limited by the correlation matrix. When the update difference of the metabolic pathway weight matrix is ​​less than a preset convergence threshold, the sparse coefficients corresponding to the metabolic pathways included in the metabolic pathway database are screened from the sparse coefficient matrix, and the metabolic pathway score is obtained by summing the sparse coefficients corresponding to all metabolic pathways included in the metabolic pathway database.

[0155] In one embodiment, the determining module 330 is specifically used for: clinical parameters including age parameters and lung function measurements; matching a preset initial weight set based on the age parameters, the initial weight set including the baseline weights of the target polygenic risk score, metabolic pathway score, and clinical risk score; adjusting the baseline weights of the clinical risk score in the initial weight set according to the difference between the lung function measurements and preset measurement thresholds, the adjustment range of the baseline weights of the clinical risk score being proportional to the difference; keeping the sum of the baseline weights of the target polygenic risk score and the metabolic pathway score unchanged, readjusting the allocation ratio of the baseline weights of the target polygenic risk score and the metabolic pathway score, the allocation ratio of the baseline weights of the target polygenic risk score being proportional to the difference; and generating a risk assessment value by weighting and summing the corresponding target polygenic risk score, metabolic pathway score, and clinical risk score based on the adjusted baseline weights corresponding to the target polygenic risk score, metabolic pathway score, and clinical risk score.

[0156] In one embodiment, the determining module 320 is specifically used to construct a local linkage disequilibrium matrix by dividing the genotype data region of the single nucleotide polymorphism data in the reference panel of the linkage disequilibrium reference information using a sliding window based on preset linkage disequilibrium reference information; based on the effect value of the single nucleotide polymorphism data and the local linkage disequilibrium matrix, the posterior mean effect value of each single nucleotide polymorphism data is calculated using a point normal mixture Bayesian model and Markov chain Monte Carlo sampling; the posterior mean effect value is used as the corrected effect value of each single nucleotide polymorphism data; and the corrected effect value of each single nucleotide polymorphism data is multiplied and summed with the corresponding genotype to generate a polygenic risk score.

[0157] In one embodiment, clinical data includes physiological indicators, medical history, and demographic information.

[0158] Figure 3 Each module in the system shown has an implementation Figures 1 to 2 The functions of each step in the process and their corresponding technical effects are described in detail here for the sake of brevity.

[0159] Figure 4 A schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application is shown.

[0160] The electronic device may include a processor 410 and a memory 420 storing computer program instructions.

[0161] Specifically, the processor 410 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0162] Memory 420 may include mass storage for data or instructions. For example, and not limitingly, memory 420 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 420 may include removable or non-removable (or fixed) media. Where appropriate, memory 420 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 420 is non-volatile solid-state memory.

[0163] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to the first aspect of this disclosure.

[0164] The processor 410 reads and executes computer program instructions stored in the memory 420 to implement any of the risk assessment methods for idiopathic pulmonary hypertension in the above embodiments.

[0165] In one example, the electronic device may also include a communication interface 430 and a bus 440. Wherein, such as Figure 4 As shown, the processor 410, memory 420, and communication interface 430 are connected through bus 440 and complete communication with each other.

[0166] The communication interface 430 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0167] Bus 440 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 440 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0168] The electronic device can perform the risk assessment method for idiopathic pulmonary hypertension as described in the embodiments of this application, thereby achieving a combination of Figures 1 to 2 The risk assessment method for idiopathic pulmonary hypertension is described.

[0169] Furthermore, in conjunction with the risk assessment methods for idiopathic pulmonary hypertension described in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the risk assessment methods for idiopathic pulmonary hypertension described in the above embodiments.

[0170] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0171] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0172] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0173] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0174] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A risk assessment method for idiopathic pulmonary hypertension, characterized in that, include: Acquire single nucleotide polymorphism data associated with idiopathic pulmonary hypertension, as well as protein biomarker data, metabolomics data, and clinical data of the target subjects associated with idiopathic pulmonary hypertension, wherein each single nucleotide polymorphism data includes the corresponding genotype and effect size; Based on preset linkage disequilibrium reference information, the effect values ​​of the single nucleotide polymorphism data are corrected using a Bayesian framework, and a polygenic risk score is determined by combining the genotype of each single nucleotide polymorphism data with the corrected effect value. Based on the expression levels of protein biomarkers in the protein biomarker data and the site effect values ​​of preset protein quantitative trait loci regulating the protein biomarkers, a protein expression score is obtained, and a target polygenic risk score is calculated based on the protein expression score and the polygenic risk score. The metabolomics data were dimensionality reduced by sparse coding to determine the sparse coefficients of metabolic pathways related to the idiopathic pulmonary hypertension, and the sparse coefficients of the metabolic pathways were converted into metabolic pathway scores. The clinical indicators in the clinical data are quantified and weighted to obtain a clinical risk score; Based on the clinical parameters in the clinical data, determine the weighting coefficients corresponding to the target polygenic risk score, the metabolic pathway score, and the clinical risk score, and combine the target polygenic risk score, the metabolic pathway score, and the clinical risk score to determine the risk assessment value; The risk assessment value is matched with a preset age-stratified risk threshold range to obtain the risk assessment level information of the target object.

2. The method according to claim 1, characterized in that, The method further includes: Obtain a set of protein biomarker data related to the idiopathic pulmonary hypertension in different age groups of the general population; Based on the expression levels of protein biomarkers in each age group in the protein biomarker dataset, time series analysis is performed to obtain protein biomarker age parameters that characterize the changing trend of protein biomarker expression levels associated with idiopathic pulmonary hypertension with age. The protein expression score is adjusted by weighting the protein expression score using the age parameter of the protein biomarker.

3. The method according to claim 1, characterized in that, The process involves obtaining a protein expression score based on the expression levels of protein biomarkers in the protein biomarker data and the site effect values ​​of preset protein quantitative trait loci regulating the protein biomarkers, and calculating a target polygenic risk score based on the protein expression score and the polygenic risk score, including: The contribution value of each protein biomarker is obtained by multiplying the expression level of the protein biomarker and the site effect value of each protein quantitative trait site that regulates the protein biomarker, and then summing them. The contribution values ​​of all the protein markers are summed to obtain the protein expression score; The target polygenic risk score is obtained by directly adding the polygenic risk score to the protein expression score.

4. The method according to claim 1, characterized in that, The step of dimensionality reduction of the metabolomics data through sparse coding to determine the sparse coefficients of metabolic pathways related to idiopathic pulmonary hypertension, and converting the sparse coefficients into metabolic pathway scores, includes: Based on a pre-defined database of metabolic pathways related to idiopathic pulmonary hypertension, a metabolic association matrix is ​​constructed, wherein the rows and columns of the association matrix correspond to the metabolic pathways related to idiopathic pulmonary hypertension and the metabolites in the metabolomics data, respectively. The concentration value of each metabolite in the metabolomics data is multiplied with the correlation matrix to generate an initial metabolic pathway activity matrix, and the initial metabolic pathway activity matrix is ​​decomposed into a metabolic pathway weight matrix and a sparse coefficient matrix. The metabolic pathway weight matrix and the sparse coefficient matrix are iteratively updated using an alternating direction optimization method. In each iteration, when the metabolic pathway weight matrix is ​​fixed, an L1 norm constraint is applied to the sparse coefficient matrix to determine key metabolic pathways. When the sparse coefficient matrix is ​​fixed, the update direction of the metabolic pathway weight matrix is ​​limited by the correlation matrix. When the update difference of the metabolic pathway weight matrix is ​​less than a preset convergence threshold, the sparse coefficients corresponding to the metabolic pathways included in the metabolic pathway database are selected from the sparse coefficient matrix, and the metabolic pathway score is obtained by summing the sparse coefficients corresponding to all metabolic pathways included in the metabolic pathway database.

5. The method according to claim 1, characterized in that, The clinical parameters include age and lung function measurements; The step involves determining the weighting coefficients corresponding to the target polygenic risk score, the metabolic pathway score, and the clinical risk score based on the clinical parameters in the clinical data, and combining these factors to determine a risk assessment value, including: Based on the age parameter, a preset initial weight set is matched, the initial weight set including the target polygenic risk score, the metabolic pathway score and the baseline weight of the clinical risk score; Based on the difference between the lung function measurement value and the preset measurement threshold, the baseline weight of the clinical risk score in the initial weight set is adjusted, and the adjustment range of the baseline weight of the clinical risk score is directly proportional to the difference. Keeping the total baseline weight of the target polygenic risk score and the metabolic pathway score unchanged, the allocation ratio of the baseline weight of the target polygenic risk score and the metabolic pathway score is readjusted, and the allocation ratio of the baseline weight of the target polygenic risk score is directly proportional to the difference. Based on the adjusted baseline weights corresponding to the target polygenic risk score, the metabolic pathway score, and the clinical risk score, the risk assessment value is generated by weighted summation of the corresponding target polygenic risk score, the metabolic pathway score, and the clinical risk score.

6. The method according to claim 1, characterized in that, Based on preset linkage disequilibrium reference information, the effect values ​​of the single nucleotide polymorphism (SNP) data are corrected using a Bayesian framework. A polygenic risk score is then determined by combining the genotype and corrected effect value for each SNP, including: Based on the preset linkage disequilibrium reference information, the genotype data region of the single nucleotide polymorphism data in the reference panel of the linkage disequilibrium reference information is divided by a sliding window to construct a local linkage disequilibrium matrix. Based on the effect values ​​of the single nucleotide polymorphism data and the local linkage disequilibrium matrix, the posterior mean effect value of each single nucleotide polymorphism data is calculated using a point normal mixture Bayesian model and Markov chain Monte Carlo sampling. The posterior mean effect value is used as the corrected effect value for each single nucleotide polymorphism data point; The corrected effect value of each single nucleotide polymorphism data point is multiplied by the corresponding genotype and summed to generate the polygenic risk score.

7. The method according to claim 1, characterized in that, The clinical data includes physiological indicators, medical history, and demographic information.

8. A risk assessment system for idiopathic pulmonary hypertension, characterized in that, The system includes: The acquisition module is used to acquire single nucleotide polymorphism data related to idiopathic pulmonary hypertension, as well as protein biomarker data, metabolomics data and clinical data of the target subjects related to the idiopathic pulmonary hypertension. Each single nucleotide polymorphism data includes the corresponding genotype and effect value. The determination module is used to correct the effect value of the single nucleotide polymorphism data using a Bayesian framework based on preset linkage disequilibrium reference information, and to determine the polygenic risk score by combining the genotype of each single nucleotide polymorphism data and the corrected effect value. The calculation module is used to obtain a protein expression score based on the expression level of the protein biomarker in the protein biomarker data and the site effect value of the protein quantitative trait locus that regulates the protein biomarker, and to calculate a target polygenic risk score based on the protein expression score and the polygenic risk score. The determination module is also used to reduce the dimensionality of the metabolomics data through sparse coding, determine the sparse coefficient of the metabolic pathways related to the idiopathic pulmonary hypertension, and convert the sparse coefficient of the metabolic pathways into metabolic pathway scores. The quantification module is used to quantify the clinical indicators in the clinical data and obtain a clinical risk score by weighted summation. The determination module is further configured to determine the weighting coefficients corresponding to the target polygenic risk score, the metabolic pathway score, and the clinical risk score based on the clinical parameters in the clinical data, and to determine the risk assessment value by combining the target polygenic risk score, the metabolic pathway score, and the clinical risk score; The matching module is used to match the risk assessment value with a preset age-stratified risk threshold range to obtain the risk assessment level information of the target object.

9. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements a risk assessment method for idiopathic pulmonary hypertension as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement a risk assessment method for idiopathic pulmonary hypertension as described in any one of claims 1-7.

Images