Cardiovascular disease risk assessment markers based on metabolomics data and their applications

Abstract

Provided are a cardiovascular disease risk assessment marker based on metabolomics data and its application. Specifically, provided is the application of a reagent for detecting metabolite levels in a sample from an individual in the preparation of a cardiovascular disease risk assessment device, wherein the metabolites include differential metabolites screened based on metabolomics data of cardiovascular disease patients and healthy people, including at least: pretyrosine, 6-phospho-2-dehydro-D-gluconic acid, retinoic acid, clopidogrel carboxylic acid, leukotriene B4 and 7α-hydroxycholesterol.

Description

Technical Field

[0001] The present invention relates to a cardiovascular disease risk assessment marker based on metabolomics data, and an early warning method and device. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention recited in the claims. No admission is made that the description herein is prior art by inclusion in this section.

[0003] Cardiovascular disease mainly refers to coronary artery atherosclerosis, abbreviated as coronary artery disease (CAD). Coronary artery disease is a metabolic disorder caused by the combined effects of genes and the environment. According to its clinical symptoms, arterial blockage and degree of myocardial damage, CAD can be divided into different categories: stable coronary artery disease (sCAD), unstable angina pectoris (UA) and acute myocardial infarction (AMI), among which UA and AMI are also called acute coronary syndrome (ACS). ACS is mostly caused by sudden intraluminal thrombosis caused by rupture or erosion of atherosclerotic (AS) plaques. Atherosclerosis is a lipid-driven inflammatory disease of the arterial intima. Risk factors such as hyperlipidemia and high cholesterol increase the level of oxidized low-density lipoprotein (LDL), leading to damage to the arterial intima. Monocytes on the intima phagocytose lipids to become foam cells, forming atherosclerotic lesions - lipid streaks. Later, under the action of multiple growth factors and pro-inflammatory mediators, lesions are generated, forming fibrous plaques. When the plaques are unstable, they rupture, causing acute thrombosis. Due to the dynamics and complexity of coronary heart disease, the mechanism of formation, development and displacement of inflammatory unstable plaques remains unclear. Therefore, identifying biomarkers of coronary plaque rupture in ACS patients is crucial to prevent the transition from a stable state to an unstable state and to prevent thrombosis.

[0004] At present, genomic research has achieved great success in identifying genetic variations that affect major risk factors such as cholesterol levels, but new specific susceptibility genes, drug targets and biomarkers that predict ACS risk are still needed to increase medical staff's understanding of the pathological mechanism of ACS. Existing diagnostic methods for coronary heart disease have shortcomings such as poor sensitivity and specificity, invasiveness and invasion, making it difficult to achieve early diagnosis of ACS.

[0005] Metabolomics is a new omics technology that emerged after genomics, transcriptomics and proteomics. It has become a powerful tool for studying metabolic processes, identifying biomarkers and revealing metabolic mechanisms. Therefore, the use of metabolomics to deeply explore the dynamic regulatory network of coronary heart disease, explore the basis of key signal pathways, find molecular markers, and establish early diagnosis and treatment methods for coronary heart disease is of great significance for reducing the incidence and mortality of coronary heart disease.

[0006] With the advancement of metabolomics analysis instruments and methods, the number of metabolites detected from biological samples is increasing, and the amount of data is also growing. Therefore, how to fully consider the metabolic characteristics of patients with acute coronary syndrome and determine a model for early warning of acute coronary syndrome based on metabolite data is a technical problem that needs to be solved urgently. Summary of the invention

[0007] One object of the present invention is to provide a group of markers associated with the risk of cardiovascular disease.

[0008] Another object of the present invention is to provide a method for establishing a cardiovascular disease risk assessment model.

[0009] Another object of the present invention is to provide a cardiovascular disease risk assessment model.

[0010] Another object of the present invention is to provide a device for assessing the risk of cardiovascular disease.

[0011] Another object of the present invention is to provide a method for assessing the risk of cardiovascular disease.

[0012] The inventors of this case have identified a group of biomarkers related to the risk of cardiovascular disease through extensive research and actual detection and analysis experiments, which include multiple metabolites. By detecting the relevant information of these metabolites in samples from individuals, the risk of cardiovascular disease in individuals can be well assessed.

[0013] Specifically, on the one hand, the present invention provides the use of a reagent for detecting metabolite levels in a sample from an individual in the preparation of a cardiovascular disease risk assessment device, wherein the metabolites include differential metabolites screened based on metabolomics data of cardiovascular disease patients and healthy people, including at least: protyrosine, 6-phospho-2-dehydro-D-gluconic acid, retinoic acid, clopidogrel carboxylic acid, leukotriene B4 and 7α-hydroxycholesterol.

[0014] According to a specific embodiment of the present invention, in the application of the present invention, the cardiovascular disease is stable coronary heart disease, acute coronary syndrome or acute coronary syndrome for stable coronary heart disease.

[0015] According to a specific embodiment of the present invention, in the application of the present invention:

[0016] When the cardiovascular disease is stable coronary heart disease, the metabolites include: 6-phospho-2-dehydro-D-gluconic acid, 4-phosphopanthenol-L-cysteine, piperidine, chorismic acid, vitamin A, methyl piperidine, biotin, retinoic acid, N-formylkynurenine, 7α-hydroxycholesterol, glycocholic acid, sorbitol, leukotriene A4, protyrosine, desthiobiotin and clopidogrel carboxylic acid;

[0017] When the cardiovascular disease is acute coronary syndrome, the metabolites include: 6-phospho-2-dehydro-D-gluconic acid, retinoic acid, pantothenic acid, protyrosine, 4-phosphopantothenoylcysteine, aconitic acid, N-acetyl-L-phenylalanine, dihydrouracil, dopaquinone, 7α-hydroxycholesterol, 6-phosphoglucose, 2-hydroxybutyric acid, lysine, 2,3-dihydroxybutyric acid, rhamno-1,4-lactone, salicylic acid, clopidogrel carboxylic acid, N-formylkynurenine, leukotriene B4, hydroxyethyllysine, 3-hydroxy-L-kynurenine, succinic acid and galactose;

[0018] When the cardiovascular disease is acute coronary syndrome for stable coronary heart disease, the metabolites include: retinoic acid, 6-phospho-2-dehydro-D-gluconate, succinic acid, vitamin A, sorbitol, 7,8-dihydroxykynurenine, clopidogrel carboxylic acid, leukotriene A4, lysine, threose, piperidine, rhamn-1,4-lactone, dihydrouracil, pantothenic acid, glucose, protocatechuic aldehyde, chorismic acid, 6-phosphoglucose, threonine, aconitic acid, tryptophan, hydroxyethyllysine, riboflavin, dopaquinone, 2-hydroxybutyric acid, pimelic acid, protyrosine, galactose, 2-hydroxyphenylpropionic acid, 7α-hydroxycholesterol and ribose.

[0019] According to a specific embodiment of the present invention, in the application of the present invention, the characteristic importance of each metabolite in assessing the risk of cardiovascular disease is in the following order:

[0020] 6-phospho-2-dehydro-D-gluconic acid > 4-phosphopanthin-L-cysteine ​​> piperidine > chorismic acid = vitamin A > methylpiperidine > biotin = retinoic acid > N-formylkynurenine = 7α-hydroxycholesterol = glycocholic acid > sorbitol > leukotriene A4 = protyrosine > desthiobiotin > clopidogrel carboxylic acid;

[0021] 6-phospho-2-dehydro-D-gluconic acid > retinoic acid > pantothenic acid = protyrosine > 4-phosphopantothenoylcysteine ​​> aconitic acid > N-acetyl-L-phenylalanine > dihydrouracil > dopaquinone > 7α-hydroxycholesterol > 6-phosphoglucose = 2-hydroxybutyric acid > lysine > 2,3-dihydroxybutyric acid > rhamnosyl-1,4-lactone > salicylic acid = clopidogrel carboxylic acid > N-formylkynurenine = leukotriene B4 > hydroxyethyllysine = 3-hydroxy-L-kynurenine = succinic acid = galactose;

[0022] Retinoic acid > 6-phospho-2-dehydro-D-gluconic acid > succinic acid > vitamin A = sorbitol = 7,8-dihydroxykynurenine = clopidogrel carboxylic acid > leukotriene A4 = lysine = threose > piperidine = rhamn-1,4-lactone = dihydrouracil = pantothenic acid > glucose = protocatechuic aldehyde > chorismic acid = 6-phosphoglucose = threonine > aconitic acid = tryptophan = hydroxyethyllysine > riboflavin = dopaquinone = 2-hydroxybutyric acid = pimelic acid = pre-tyrosine = galactose = 2-hydroxyphenylpropionic acid = 7α-hydroxycholesterol = ribose.

[0023] According to a specific embodiment of the present invention, in the application of the present invention, when the metabolites are used to assess the risk of cardiovascular disease, the weight of each metabolite is determined according to the following characteristic importance values, or the weight ratio of each metabolite is:

[0024] Group 1:

[0025] 6-phospho-2-dehydro-D-gluconic acid 44, 4-phosphopanthin-L-cysteine ​​24, piperidine 20, chorismic acid 12, vitamin A 12, methylpiperidine 8, biotin 7, retinoic acid 7, N-formylkynurenine 6, 7α-hydroxycholesterol 6, glycocholic acid 6, sorbitol 4, leukotriene A4 3, pro-tyrosine 3, desthiobiotin 2, clopidogrel carboxylic acid 1;

[0026] Group 2:

[0027] 6-phospho-2-dehydro-D-gluconic acid 38, retinoic acid 36, pantothenic acid 20, protyrosine 20, 4-phosphopantothenoylcysteine ​​19, aconitic acid 18, N-acetyl-L-phenylalanine 15, dihydrouracil 13, dopaquinone 12, 7α-hydroxycholesterol 11, 6-phosphoglucose 10, 2-hydroxybutyric acid 10, lysine 9, 2,3-dihydroxybutyric acid 6, rhamno-1,4-lactone 4, salicylic acid 3, clopidogrel carboxylic acid 3, N-formylkynurenine 2, leukotriene B4 2, hydroxyethyllysine 1, 3-hydroxy-L-kynurenine 1, succinic acid 1, galactose 1;

[0028] Group 3:

[0029] Retinoic acid 15, 6-phospho-2-dehydro-D-gluconate 13, succinic acid 8, vitamin A 7, sorbitol 7, 7,8-dihydroxykynurenine 7, clopidogrel carboxylic acid 7, leukotriene A4 6, lysine 6, threose 6, piperidine 5, rhamn-1,4-lactone 5, dihydrouracil 5, pantothenic acid 5, glucose 4, protocatechuic aldehyde 4, chorismic acid 3, 6-phosphoglucose 3, threonine 3, aconitic acid 2, tryptophan 2, hydroxyethyllysine 2, riboflavin 1, dopaquinone 1, 2-hydroxybutyric acid 1, pimelic acid 1, pro-tyrosine 1, galactose 1, 2-hydroxyphenylpropionic acid 1, 7α-hydroxycholesterol 1, ribose 1.

[0030] According to a specific embodiment of the present invention, in the application of the present invention, the sample from the individual is a blood sample. Preferably, the blood sample is a serum or plasma sample.

[0031] Any feasible method in the prior art can be used to detect the level of the metabolite in the sample.

[0032] According to a specific embodiment of the present invention, the technology of the present invention is particularly suitable for assessing the risk of cardiovascular disease in individuals from the East Asian population.

[0033] In another aspect, the present invention provides a cardiovascular disease risk assessment device, which includes a detection unit and a data analysis unit, wherein:

[0034] The detection unit is used to detect the level of metabolites in a sample from an individual to obtain a detection result; wherein the metabolites are the same as the metabolites described in any one of claims 1 to 7;

[0035] The data analysis unit is used to analyze and process the detection result of the detection unit.

[0036] In the cardiovascular disease risk assessment device of the present invention, the detection unit includes any reagent material that can obtain the level of each metabolite in the individual sample to be tested, and any feasible method in the prior art can be used to detect the level of each metabolite in the individual sample to be tested.

[0037] Specifically, in the cardiovascular disease risk assessment device of the present invention, the detection unit includes reagent materials for detecting metabolite levels in a blood sample.

[0038] Specifically, in the cardiovascular disease risk assessment device described in the present invention, when the data analysis unit analyzes and processes the detection results of the detection unit, it includes: assigning the detection results of the metabolite levels in the sample from the individual with a weight coefficient to calculate the risk assessment score of the individual to be tested.

[0039] On the other hand, the present invention also provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it is achieved: obtaining an individual cardiovascular disease risk assessment result based on the metabolite level in the individual sample to be tested.

[0040] On the other hand, the present invention also provides a computer-readable storage medium, which stores computer program instructions, and when the computer program instructions are executed, they achieve: obtaining an individual cardiovascular disease risk assessment result based on the metabolite level in the individual sample to be tested.

[0041] In some embodiments of the present invention, a stable coronary heart disease early warning method based on metabolomics data is provided to solve the technical problem that the existing technology cannot realize early warning of stable coronary heart disease based on metabolite data. The method comprises: loading a stable coronary heart disease early warning model, wherein the stable coronary heart disease early warning model is a model containing multiple biomarkers obtained by machine learning the LightGBM model based on the differential metabolite data of normal coronary artery biological individuals and stable coronary heart disease biological individuals; according to the stable coronary heart disease early warning model, obtaining the index value of each biomarker corresponding to the biological individual to be diagnosed; judging whether the index value of each biomarker corresponding to the biological individual to be diagnosed exceeds the preset normal value range; and outputting the stable coronary heart disease early warning information of the biological individual to be diagnosed according to the judgment result.

[0042] The embodiment of the present invention also provides a stable coronary heart disease early warning device based on metabolomics data to solve the technical problem that the existing technology cannot realize early warning of stable coronary heart disease based on metabolite data. The device includes: a model loading module, used to load a stable coronary heart disease early warning model, wherein the stable coronary heart disease early warning model is a model containing multiple biomarkers obtained by machine learning of the LightGBM model based on the differential metabolite data of normal coronary artery biological individuals and stable coronary heart disease biological individuals; an indicator value acquisition module, used to obtain the indicator value of each biomarker corresponding to the biological individual to be diagnosed according to the stable coronary heart disease early warning model; an indicator value detection module, used to determine whether the indicator value of each biomarker corresponding to the biological individual to be diagnosed exceeds a preset normal value range; a stable coronary heart disease early warning module, used to output stable coronary heart disease early warning information of the biological individual to be diagnosed according to the judgment result.

[0043] A computer device is also provided in an embodiment of the present invention to solve the technical problem that the prior art cannot realize early warning of stable coronary heart disease based on metabolite data. The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned stable coronary heart disease early warning method based on metabolomics data is implemented.

[0044] A computer-readable storage medium is also provided in an embodiment of the present invention to solve the technical problem that the prior art cannot realize early warning of stable coronary heart disease based on metabolite data. The computer-readable storage medium stores a computer program for executing the above-mentioned stable coronary heart disease early warning method based on metabolomics data.

[0045] In an embodiment of the present invention, a stable coronary heart disease prediction model pre-trained by machine learning is first loaded, and based on the stable coronary heart disease early warning model, the index value of each biomarker in the stable coronary heart disease early warning model corresponding to the biological individual to be diagnosed is obtained, and it is determined whether the index values ​​of these biomarkers exceed the preset normal value range. Based on the judgment result, it is determined whether the biological individual to be diagnosed suffers from stable coronary heart disease biomarkers, and corresponding early warning information is issued.

[0046] Through the embodiments of the present invention, a non-invasive stable coronary heart disease early warning method can be provided, which can provide a stable coronary heart disease early warning for a patient based on the patient's metabolite data.

[0047] In other specific embodiments of the present invention, a method for early warning of acute coronary syndrome based on metabolomics data is provided to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome based on metabolite data. The method comprises: obtaining metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals; screening out differential metabolite data based on the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals; performing machine learning on the LightGBM model based on the differential metabolite data to obtain an acute coronary syndrome early warning model, wherein the acute coronary syndrome early warning model includes: a plurality of biomarkers for diagnosing acute coronary syndrome; according to the acute coronary syndrome early warning model, obtaining the index value of each biomarker corresponding to the biological individual to be diagnosed, and judging whether the index value of each biomarker corresponding to the biological individual to be diagnosed exceeds a preset normal value range, and outputting the acute coronary syndrome early warning information of the biological individual to be diagnosed based on the judgment result.

[0048] An embodiment of the present invention also provides an acute coronary syndrome early warning device based on metabolomics data to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome based on metabolite data. The device includes: a metabolite data acquisition module, used to obtain metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals; a differential metabolite screening module, used to screen out differential metabolite data based on the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals; a model learning module, used to perform machine learning on the LightGBM model based on the differential metabolite data to obtain an acute coronary syndrome early warning model, wherein the acute coronary syndrome early warning model includes: a plurality of biomarkers for diagnosing acute coronary syndrome; an acute coronary syndrome early warning module, used to obtain the index value of each biomarker corresponding to the biological individual to be diagnosed according to the acute coronary syndrome early warning model, and determine whether the index value of each biomarker corresponding to the biological individual to be diagnosed exceeds a preset normal value range, and output the acute coronary syndrome early warning information of the biological individual to be diagnosed according to the judgment result.

[0049] A computer device is also provided in an embodiment of the present invention to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome based on metabolite data. The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned acute coronary syndrome early warning method based on metabolomics data is implemented.

[0050] A computer-readable storage medium is also provided in an embodiment of the present invention to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome based on metabolite data. The computer-readable storage medium stores a computer program for executing the above-mentioned acute coronary syndrome early warning method based on metabolomics data.

[0051] In an embodiment of the present invention, after obtaining the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals, differential metabolite data are screened out according to the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals, and then based on the screened differential metabolite data, machine learning is performed on the LightGBM model to obtain an acute coronary syndrome early warning model containing multiple markers, so that based on the acute coronary syndrome early warning model, the index value of each biomarker corresponding to the biological individual to be diagnosed is collected, and it is determined whether the index value of each biomarker corresponding to the biological individual to be diagnosed exceeds the preset normal value range, and the corresponding early warning information is output according to the judgment result.

[0052] Through the embodiments of the present invention, a non-invasive early warning method for acute coronary syndrome can be provided, which can provide an early warning for acute coronary syndrome for patients based on the metabolite data of the patients.

[0053] In other embodiments of the present invention, a method for early warning of acute coronary syndrome for patients with stable coronary heart disease is provided to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome for biological individuals to be diagnosed based on metabolite data. The method comprises: collecting plasma samples of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome; using liquid chromatography-mass spectrometry analysis method to analyze the plasma samples of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome to obtain metabolite data of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome; based on the metabolite data of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome, differential metabolite data are screened out; based on the differential metabolite data, LightGBM model is machine-learned to obtain an early warning model for acute coronary syndrome for patients with stable coronary heart disease, wherein the early warning model for acute coronary syndrome for patients with stable coronary heart disease includes: a variety of biomarkers for diagnosing acute coronary syndrome; using the early warning model for acute coronary syndrome for patients with stable coronary heart disease to warn the biological individuals to be diagnosed of acute coronary syndrome.

[0054] The embodiment of the present invention also provides an acute coronary syndrome early warning device for patients with stable coronary heart disease, which is used to solve the technical problem that the existing technology cannot realize the early warning of acute coronary syndrome for the biological individual to be diagnosed based on metabolite data. The device includes: a plasma sample collection module, which is used to collect plasma samples of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome; a plasma sample analysis module, which is used to use a liquid chromatography-mass spectrometry analysis method to analyze the plasma samples of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome to obtain metabolite data of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome; differential metabolites A screening module is used to screen out differential metabolite data based on the metabolite data of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome; a model training module is used to perform machine learning on the LightGBM model based on the differential metabolite data to obtain an acute coronary syndrome early warning model for patients with stable coronary heart disease, wherein the acute coronary syndrome early warning model for patients with stable coronary heart disease includes: a variety of biomarkers for diagnosing whether patients with stable coronary heart disease suffer from acute coronary syndrome; an early warning module is used to use the acute coronary syndrome early warning model for patients with stable coronary heart disease to warn of acute coronary syndrome in the biological individuals to be diagnosed.

[0055] A computer device is also provided in an embodiment of the present invention to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome for a biological individual to be diagnosed based on metabolite data. The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned early warning method for acute coronary syndrome for patients with stable coronary heart disease is implemented.

[0056] A computer-readable storage medium is also provided in an embodiment of the present invention to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome for biological individuals to be diagnosed based on metabolite data. The computer-readable storage medium stores a computer program for executing the above-mentioned early warning method for acute coronary syndrome for patients with stable coronary heart disease.

[0057] In an embodiment of the present invention, after collecting plasma samples from biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome, a liquid chromatography-mass spectrometry analysis method is used to analyze the plasma samples of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome to obtain metabolite data of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome, and then based on the metabolite data of biological individuals with stable coronary heart disease and biological individuals with acute coronary syndrome, differential metabolite data are screened out, and based on the differential metabolite data, the LightGBM model is machine-learned to obtain an acute coronary syndrome early warning model for patients with stable coronary heart disease, and the model is used to provide early warning of acute coronary syndrome for biological individuals to be diagnosed.

[0058] Through the embodiments of the present invention, a non-invasive method for early warning of acute coronary syndrome for a biological individual to be diagnosed can be provided, and early warning of acute coronary syndrome for the biological individual to be diagnosed can be performed based on the metabolite data of patients with stable coronary heart disease and patients with acute coronary syndrome. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0060] Figure 1 The present invention is a flowchart of an acute coronary syndrome early warning method based on metabolomics data in an embodiment of the present invention.

[0061] Figure 2 A machine learning flowchart provided in an embodiment of the present invention.

[0062] Figure 3 The present invention provides a flow chart of metabolite data acquisition in an embodiment of the present invention.

[0063] Figure 4 The AUROC curve diagram of an acute coronary syndrome early warning model provided in an embodiment of the present invention.

[0064] Figure 5 This is a schematic diagram of an acute coronary syndrome early warning result based on metabolomics data in an embodiment of the present invention.

[0065] Figure 6 Schematic diagram of an acute coronary syndrome early warning device based on metabolomics data in an embodiment of the present invention.

[0066] Figure 7 This is a schematic diagram of an optional acute coronary syndrome early warning device based on metabolomics data provided in an embodiment of the present invention.

[0067] Figure 8 A schematic diagram of a computer device provided in an embodiment of the present invention.

[0068] Fig. 9 The figure is an AUROC curve diagram of a stable coronary heart disease prediction model provided in an embodiment of the present invention.

[0069] Fig.10 This is a schematic diagram of a stable coronary heart disease early warning result based on metabolomics data provided in an embodiment of the present invention.

[0070] Fig.11 This is an AUROC curve diagram of an acute coronary syndrome prediction model for patients with stable coronary heart disease provided in an embodiment of the present invention.

[0071] Fig.12 This is a schematic diagram of the prediction results of acute coronary syndrome biomarker information for patients with stable coronary heart disease provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0072] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0073] Implementation Plan 1: Acute Coronary Syndrome Risk Assessment

[0074] The present invention provides an early warning method for acute coronary syndrome based on metabolomics data. Figure 1 Flow chart of an acute coronary syndrome early warning method based on metabolomics data provided in an embodiment of the present invention, such as Figure 1 As shown, the method comprises the following steps:

[0075] S101, obtaining metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals.

[0076] It should be noted that the above-mentioned normal coronary artery biological individuals refer to people who do not suffer from acute coronary syndrome (i.e., normal people); the above-mentioned acute coronary syndrome biological individuals refer to people who suffer from acute coronary syndrome (i.e., patients); the metabolite data in the embodiments of the present invention refer to the data of various biological metabolites obtained by analyzing the plasma samples of the research subjects.

[0077] S102, screening out differential metabolite data based on the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals.

[0078] It should be noted that by analyzing the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals, one or more metabolites with differences can be screened out.

[0079] S103, based on the differential metabolite data, the LightGBM model is machine-learned to obtain an early warning model for acute coronary syndrome, wherein the early warning model for acute coronary syndrome includes: a variety of biomarkers for diagnosing acute coronary syndrome.

[0080] It should be noted that the above-mentioned acute coronary syndrome early warning model refers to training a machine learning model to obtain a model that can provide early warning for acute coronary syndrome based on metabolite data.

[0081] In specific implementation, different machine learning algorithms can be used according to experimental conditions, differences in biological samples and applicable conditions, such as linear regression, random forest, etc. In the embodiment of the present invention, the LightGBM (Light Gradient Boosting Machine) model is selected for machine learning training. The model has the advantages of faster training efficiency, low memory usage, higher accuracy, support for parallel learning, and the ability to process large-scale data.

[0082] S104, according to the early warning model for acute coronary syndrome, obtain the index value of each biomarker corresponding to the biological individual to be diagnosed, and determine whether the index value of each biomarker corresponding to the biological individual to be diagnosed exceeds the preset normal value range, and output the early warning information of acute coronary syndrome of the biological individual to be diagnosed according to the determination result.

[0083] It should be noted that the above-mentioned biological individuals to be diagnosed refer to people whose acute coronary syndrome is unknown. The acute coronary syndrome early warning method based on metabolomics data provided in the embodiment of the present invention can output the acute coronary syndrome early warning information of the biological individuals to be diagnosed in any form such as text or voice.

[0084] The acute coronary syndrome early warning method based on metabolomics data provided in the embodiment of the present invention provides an acute coronary syndrome early warning comprising the following biomarkers: 2-hydroxybutyric acid, dihydrouracil, succinic acid, 2,3-dihydroxybutyric acid, salicylic acid, lysine, rhamno-1,4-lactone, dopaquinone, aconitic acid, N-acetyl-L-phenylalanine, hydroxyethyllysine, pantothenic acid, 3-hydroxy-L-kynurenine, N-formylkynurenine, 6-phosphoglucose, protyrosine, 6-phospho-2-dehydro-D-gluconic acid, retinoic acid, clopidogrel carboxylic acid, leukotriene B4, galactose, 7α-hydroxycholesterol, and 4-phosphopantothenoylcysteine.

[0085] like Figure 2 As shown, the acute coronary syndrome early warning method based on metabolomics data provided in the embodiment of the present invention can also be used to train an acute coronary syndrome early warning model through the following steps:

[0086] S201, dividing the differential metabolite data into training data and test data according to a preset ratio;

[0087] S202, training the LightGBM model according to the training data to obtain an acute coronary syndrome early warning model;

[0088] S203, verifying the trained acute coronary syndrome early warning model based on the test data.

[0089] In a specific implementation, 70% of the data in the metabolomics data sample can be used as training data, and 30% of the data can be used as test data.

[0090] It should be noted that, in the early warning method for acute coronary syndrome based on metabolomics data provided in the embodiment of the present invention, the GridSearchCV algorithm and the Hyperopt algorithm can be used to adjust the model parameters during the machine learning of the LightGBM model.

[0091] In one embodiment, Figure 3 As shown, the early warning method for acute coronary syndrome based on metabolomics data provided in the embodiment of the present invention can also obtain metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals through the following steps:

[0092] S301, collecting plasma samples from normal coronary artery biological subjects and acute coronary syndrome biological subjects;

[0093] S302, using a liquid chromatography-mass spectrometry analysis method to analyze the plasma samples of the normal coronary artery biological individuals and the acute coronary syndrome biological individuals to obtain metabolite data of the normal coronary artery biological individuals and the acute coronary syndrome biological individuals.

[0094] In an embodiment of the present invention, a liquid chromatography-mass spectrometry analysis method is used to analyze plasma samples of normal coronary artery biological individuals and acute coronary syndrome biological individuals, and metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals can be obtained.

[0095] In one embodiment, the acute coronary syndrome early warning method based on metabolomics data provided in the embodiments of the present invention may also include: using the Boruta algorithm to determine the importance of different biomarkers in the acute coronary syndrome early warning model.

[0096] It should be noted that the goal of Boruta is to select all feature sets related to the dependent variable, rather than selecting a feature set that minimizes the model cost function for a specific model. The significance of the Boruta algorithm is that it can more comprehensively understand the influencing factors of the dependent variable, so as to perform feature selection better and more efficiently. Boruta is a feature selection package in Python. After installing the package, input the historical information of the relative abundance of differential metabolites to obtain important features suitable for modeling. The specific algorithm steps are as follows: (1) Create shadow features: for each real feature R, randomly shuffle the order to obtain the shadow feature matrix S, which is spliced ​​behind the real feature to form a new feature matrix N = [R, S]; (2) Use the new feature matrix N as input to train the model to obtain real features and shadow features; (3) Take the maximum value of the shadow feature, and record a hit if the real feature is greater than this value; (4) Use the real feature cumulative hits recorded in (3) to mark the feature as important or unimportant; (5) Delete unimportant features and repeat (1) to (4) until all features are marked.

[0097] The acute coronary syndrome early warning method based on metabolomics data provided in the embodiment of the present invention may include the following steps when implemented:

[0098] 1. Metabolomics sample pretreatment: Collect blood samples, centrifuge to obtain plasma within one hour, and immediately store in a -80℃ refrigerator. During sample pretreatment, accurately pipette 100μl plasma into a 10ml glass tube, add 20μl mixed standard, add 3ml MeOH:MTBE (1:1, V / V) to precipitate protein, vortex at 2500rpm for 5min, then centrifuge at 4200rpm, 4℃ for 10min, and pipette the supernatant into another glass tube. Add 3 ml MTBE and 1.2 ml H2O, vortex at 2500 rpm for 15 min, and centrifuge at 4200 rpm and 4°C for 10 min. Transfer the lower aqueous phase to a 2 ml EP tube, blow dry with nitrogen, and add 100 μL ACN:H2O (2:98, V / V) for re-dissolution. Vortex the reconstituted sample at 2500 rpm for 5 min, centrifuge at 4°C and 12500 for 5 min, collect the supernatant, filter through a 96-well plate, and inject for analysis.

[0099] 2. LC-HRMS analysis conditions: Liquid chromatography-mass spectrometry analysis method was used for metabolomics data collection, where the chromatographic column selected was Waters ACQUITY UPLC HSS T3 chromatographic column (2.1×100mm, 1.8μm); the injection volume was 10μL; the column temperature was 35°C; the flow rate was 0.25ml / min. Mobile phase A: H2O (0.1% formic acid); mobile phase B: ACN. The mobile phase gradient is shown in Table 1, and the mass spectrometry conditions are shown in Table 2.

[0100] Table 1 Mobile phase gradient

[0101] Total time(min) A(v%) B (v%) 0 98 2 9 40 60 18 40 60 20 0 100 30 0 100

[0102] Table 2 Mass spectrometry conditions

[0103] parameter Numeric Ion source ESI Ionization mode Positive ion mode / negative ion mode Scan Mode Full Scan Scan range 70-1000 Resolution 70000 Automatic gain control target number 3e6 Maximum injection time (ms) 100 Ion source temperature (℃) 350℃ Sheath gas flow rate (arb) 40 / 45 Spray voltage(KV) 3.5 / -3.2 Auxiliary gas flow rate (arb) 11 / 10 Auxiliary gas heating temperature (℃) 220

[0104] 3. Metabolomics data processing and differential metabolite screening: The raw data obtained by LC-HRMS analysis were imported into Progenesis QI for peak identification, peak alignment, and data normalization to obtain a two-dimensional data matrix containing information such as mass-to-charge ratio, retention time, original peak area, and normalized peak area. QC samples were used for data quality assessment. For the quality-controlled data, a metabolomics processing method based on metabolic pathway mapping of the Mummichog algorithm was used to quickly match the molecular weight information in the metabolite database with the detected ion information with an appropriate mass-to-charge ratio tolerance, and the metabolites were "tentatively labeled" with high throughput and directly mapped to the metabolic pathway based on the KEGG database. Through the overall changes in metabolic pathways and the correlation of metabolic reactions, the differences between groups were modularly analyzed, so as to efficiently discover and identify reliable biomarkers with biological significance. The differential metabolites obtained by preliminary identification were identified by LC-MS / MS technology for metabolite structure identification, and the metabolites with available standards were compared with the standards to further confirm the structure. For metabolites whose structures cannot be determined due to low abundance or not included in the secondary database and the inability to obtain standards, the preliminary identification results were temporarily used. Then, a model was established for prediction based on the final differential metabolite abundance information.

[0105] 4. Based on the LightGBM machine learning algorithm, build a prediction model and use external data to verify the model:

[0106] (1) Use Boruta algorithm for feature selection.

[0107] (2) GridSearchCV (grid search) adjusts parameters, that is, within the specified parameter range, adjusts the parameters in sequence according to the step size, uses the adjusted parameters to train the learner, and finds the parameters with the highest accuracy on the validation set from all the parameters. This is actually a cycle and comparison process.

[0108] (3) LightGBM modeling: LightGBM is a more powerful and faster model than Xgboost, with greatly improved performance. Compared with traditional algorithms, it has the following advantages: faster training efficiency, low memory usage, higher accuracy, support for parallel learning, and the ability to process large-scale data.

[0109] (4) Select important features in the model and remodel it;

[0110] (5) Use Hyperopt to further tune the parameters of the new model. Hyperopt is a tool that adjusts parameters through Bayesian optimization. This method is fast and has good results. In addition, Hyperopt can be combined with MongoDB to perform distributed parameter adjustment and quickly find relatively optimal parameters;

[0111] (6) LightGBM modeling prediction;

[0112] (7) Re-acquire a batch of external data that has never been involved in modeling, use the constructed model to predict this batch of data, and use AUROC to judge the quality of the prediction model.

[0113] 5. Based on the above algorithm, an early warning model for acute coronary syndrome for the normal population was obtained. The marker information used in the model is shown in Table 3.

[0114] Table 3 Biomarker information

[0115]

[0116] In the above Table 3, “*” represents the 2 Identified metabolites; “#” represents metabolites identified by comparison with standard substances; others are preliminary identifications by primary mass spectrometry.

[0117] Figure 4 The AUROC curve of the early warning model for acute coronary syndrome is shown in FIG. Figure 4 As shown, the AUROC of the training data is 0.98491, and the AUROC of the test data is 0.97166. Figure 5 Multiple acute coronary syndrome biomarkers screened from metabolomics data and their importance are shown.

[0118] The embodiment of the present invention is described below with reference to a specific example:

[0119] (I) Clinical inclusion criteria: Based on the clinical characteristics of coronary atherosclerotic heart disease, participants were divided into two groups, including: ① ST-segment elevation acute myocardial infarction (STEMI, unstable plaque rupture group, myocardial necrosis); non-ST-segment elevation acute myocardial infarction (NSTEMI, unstable plaque partial rupture group, a small amount of myocardial necrosis) and unstable angina pectoris (UAP, plaque on the verge of rupture or unstable before rupture group, myocardial micronecrosis), namely ACS group, N = 102; ② normal control group without atherosclerotic plaque, namely normal coronary artery group, NCA, N = 93. Based on the collection of clinical information, blood samples were collected from each group, and plasma was obtained by centrifugation for metabolomics analysis.

[0120] (1) Research subjects:

[0121] Inclusion criteria: stable coronary heart disease (old myocardial infarction, history of PCI, stable angina pectoris or "healthy people" without clinical ischemic symptoms, and coronary artery stenosis lesions >50% found by coronary CT / angiography).

[0122] Exclusion criteria:

[0123] 1) Myocardial infarction type 2-5 was diagnosed according to the international common definition of myocardial infarction;

[0124] 2) severe heart failure / cardiogenic shock (Killip>2 or NYHA>2);

[0125] 3) the presence of mechanical complications (ventricular septal perforation, free wall rupture, papillary muscle rupture, etc.);

[0126] 4) Cardiac arrest and / or cardiopulmonary resuscitation occurred after the onset of the disease;

[0127] 5) Taking any oral or intravenous antibiotics for ≥ 1 week within 3 months;

[0128] 6) Acute coronary syndrome (ACS) or coronary artery revascularization (including PCI and CABG) within 3 months;

[0129] 7) Trauma or surgery within 3 months;

[0130] 8) History of cerebrovascular disease within 3 months (including cerebral infarction or cerebral hemorrhage);

[0131] 9) Upper or lower gastrointestinal bleeding within 3 months;

[0132] 10) Confirmed infection within 3 months (including digestive tract, respiratory tract, body surface infection, etc.);

[0133] 11) Chronic intestinal diseases (such as Crohn's disease, ulcerative colitis, etc.);

[0134] 12) Any tumor;

[0135] 13) Rheumatic and immune diseases;

[0136] 14) Chronic kidney disease, including after kidney transplantation.

[0137] (2) Research subject selection and case information collection process

[0138] 1) Informed consent;

[0139] 2) inclusion / exclusion criteria;

[0140] 3) clinical data of patients’ lifestyle questionnaire;

[0141] 4) Based on the clinical information collected, blood and fresh or properly frozen feces were collected from each group of people for omics analysis.

[0142] (3) This clinical study was carried out in compliance with the World Medical Association Declaration of Helsinki and relevant national laws and regulations. All clinical patients participating in the experiment have signed the Informed Consent Form for this project.

[0143] (II) Implementation methods:

[0144] (1) Clinical sample collection and pre-processing: Metabolomics data of 195 participants (biological individuals) were collected. According to the diagnostic guidelines and exclusion criteria, the collected metabolomics data were divided into the following two groups: NCA group (N = 93) and ACS group (N = 102).

[0145] Blood samples were collected from each participant in the morning when the fasting time was greater than 10 hours, and the hospital completed the relevant clinical routine biomarker tests. All tests were performed in accordance with international standard methods. After the test, the samples were collected in a tube containing sodium heparin, cooled to 4°C, and centrifuged within 1 hour to obtain plasma (1,350rpm, centrifugation for 12 minutes), divided into aliquots and stored in a -80°C refrigerator.

[0146] The collected samples were pre-treated before metabolomics analysis. The pre-treatment steps are as follows:

[0147] Accurately pipette 100μl plasma into a 10ml glass tube, add 20μl mixed standard, add 3ml MeOH:MTBE (1:1, V / V) to precipitate protein, vortex at 2500rpm for 5min, centrifuge at 4200rpm, 4℃ for 10min, and pipette the supernatant into another glass tube. Add 3ml MTBE and 1.2ml H2O, vortex at 2500rpm for 15min, centrifuge at 4200rpm, 4℃ for 10min, transfer the lower aqueous phase to a 2ml EP tube, blow dry with nitrogen, add 100μL ACN:H2O (2:98, V / V) to re-dissolve, vortex the re-dissolved sample at 2500rpm for 5min, centrifuge at 4℃, 12500 for 5min, take the supernatant, filter through a 96-well plate, and inject for analysis.

[0148] (2) LC-HRMS analysis conditions: Liquid chromatography-mass spectrometry analysis method was used for metabolomics data acquisition, wherein the liquid chromatography system selected Waters two-dimensional ACQUITY UPLC system, including binary ultra-high pressure gradient pump, quaternary ultra-high pressure gradient pump, online degasser, autosampler, column oven, 4-40°C constant temperature oven, diode array detector and evaporative light scattering detector, and equipped with EmpowerTM workstation software, and the chromatographic column selected Waters ACQUITY UPLC HSS T3 chromatographic column (2.1×100mm, 1.8μm); injection volume was 10μL; column temperature was 35°C; flow rate was 0.25ml / min. Mobile phase A: H2O (0.1% formic acid); mobile phase B: ACN. The mobile phase gradient is shown in Table 4.

[0149] Table 4 Mobile phase gradient

[0150]

[0151]

[0152] After liquid phase analysis, the samples were subjected to mass spectrometry detection using a Q-Orbitrap high-resolution mass spectrometer (Q-Exactive model) in full scan acquisition mode, which was equipped with an Xcalibur 3.0 data acquisition system. The detailed parameter settings are shown in Table 5.

[0153] Table 5 Setting parameter information

[0154] parameter Numeric Ion source ESI Ionization mode Positive ion mode / negative ion mode Scan Mode Full Scan Scan range 70-1000 Resolution 70000 Automatic gain control target number 3e6 Maximum injection time (ms) 100 Ion source temperature (℃) 350℃ Sheath gas flow rate (arb) 40 / 45 Spray voltage(KV) 3.5 / -3.2 Auxiliary gas flow rate (arb) 11 / 10 Auxiliary gas heating temperature (℃) 220

[0155] (3) Metabolomics data processing and differential metabolite screening:

[0156] The raw data obtained by LC-MS analysis were imported into Progenesis QI for peak identification, peak alignment, and data normalization to obtain a two-dimensional data matrix containing information such as mass-to-charge ratio, retention time, original peak area, and normalized peak area. QC samples were used for data quality assessment. For the quality-controlled data, the metabolomics processing method of metabolic pathway mapping based on the Mummichog algorithm was used. After setting the appropriate mass-to-nuclear ratio tolerance (<5ppm) and the summed ion form, the m / z (mass-to-charge ratio, i.e., the ratio of the number of protons to the number of charges) obtained by the high-resolution mass spectrometry data was tentatively annotated with high-throughput to obtain metabolite information. Then, pathway analysis and module analysis were performed in combination with pathway databases such as KEGG and UCSD BiGG. Finally, 12 metabolic pathways with significant changes were obtained, including phenylalanine metabolism, tryptophan metabolism, arachidonic acid metabolism, tyrosine metabolism, valine, leucine and isoleucine biosynthesis, TCA cycle, etc. The metabolites in the pathway were subjected to t-tests to obtain significantly changed differential metabolites. The differential metabolites obtained by preliminary identification were identified by LC-MS / MS technology, and the metabolites with available standards were compared with the standards to further confirm the structure. For metabolites whose structures could not be determined due to low abundance or lack of inclusion in the secondary database and inability to obtain standards, the preliminary identification results were temporarily used. Finally, 49 differential metabolites were obtained.

[0157] (4) Use machine learning methods to establish a prediction model: Use the LightGBM machine learning method to build a model and the ten-by-ten cross-validation method to randomly divide the differential metabolite data into a training set and a test set. First, use the Boruta algorithm for feature selection. Use GridSearchCV (grid search) and Hyperopt to continuously adjust the parameters and select the optimal parameters. Reacquire a batch of external data that has never been involved in modeling, use the constructed model to predict this batch of data, and use AUROC to judge the quality of the prediction model. The importance of the feature is expressed by its contribution to the model. All analyses use Python's Scikit-learn package.

[0158] As can be seen from the above, the acute coronary syndrome early warning method based on metabolomics data provided in the embodiment of the present invention is aimed at the problem that the current coronary heart disease diagnosis method has poor sensitivity and specificity, is invasive and invasive, and is difficult to achieve early diagnosis of ACS. The metabolomics data of normal coronary arteries and acute coronary syndrome patients are obtained by using LC-HRMS technology, and after screening differential metabolites, biomarkers that can be used to predict and monitor acute coronary syndrome are screened from complex and cumbersome biological big data based on machine learning methods, which can provide a non-invasive screening and early warning method to fill the gap in clinical early warning of acute coronary syndrome. Based on metabolomics biological big data, biomarkers are screened using machine learning methods to make the screened biomarkers more reliable.

[0159] Based on the same inventive concept, an acute coronary syndrome early warning device based on metabolomics data is also provided in the embodiments of the present invention, such as the following embodiments. Since the principle of solving the problem by the device is similar to that of the acute coronary syndrome early warning method based on metabolomics data, the implementation of the device can refer to the implementation of the acute coronary syndrome early warning method based on metabolomics data, and the repeated parts will not be repeated.

[0160] Figure 6 Schematic diagram of an acute coronary syndrome early warning device based on metabolomics data provided in an embodiment of the present invention, such as Figure 6 As shown, the device includes: a metabolite data acquisition module 61, a differential metabolite screening module 62, a model learning module 63 and an acute coronary syndrome early warning module 64.

[0161] Among them, the metabolite data acquisition module 61 is used to obtain the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals; the differential metabolite screening module 62 is used to screen out differential metabolite data based on the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals; the model learning module 63 is used to perform machine learning on the LightGBM model based on the differential metabolite data to obtain an acute coronary syndrome early warning model, wherein the acute coronary syndrome early warning model includes: a variety of biomarkers for diagnosing acute coronary syndrome; the acute coronary syndrome early warning module 64 is used to obtain the index value of each biomarker corresponding to the biological individual to be diagnosed according to the acute coronary syndrome early warning model, and determine whether the index value of each biomarker corresponding to the biological individual to be diagnosed exceeds the preset normal value range, and output the acute coronary syndrome early warning information of the biological individual to be diagnosed according to the judgment result.

[0162] Optionally, the metabolite data acquisition module 61 is further used to: obtain plasma samples of normal coronary artery biological individuals and acute coronary syndrome biological individuals; use a liquid chromatography-mass spectrometry analysis method to analyze the plasma samples of normal coronary artery biological individuals and acute coronary syndrome biological individuals to obtain metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals.

[0163] In one embodiment, in the acute coronary syndrome early warning device based on metabolomics data provided in the embodiments of the present invention, the model learning module 63 is also used to adjust the model parameters using the GridSearchCV algorithm and the Hyperopt algorithm during the process of machine learning of the LightGBM model.

[0164] In one embodiment, Figure 7 As shown, in the acute coronary syndrome early warning device based on metabolomics data provided in an embodiment of the present invention, the model learning module 63 specifically includes: a sample data division module 631, a model training module 632 and a model verification module 633.

[0165] Among them, the sample data division module 631 is used to divide the differential metabolite data into training data and test data according to a preset ratio; the model training module 632 is used to train the LightGBM model according to the training data to obtain an acute coronary syndrome early warning model; the model verification module 633 is used to verify the trained acute coronary syndrome early warning model according to the test data.

[0166] In one embodiment, Figure 7As shown, the acute coronary syndrome early warning device based on metabolomics data provided in the embodiment of the present invention also includes: a biomarker analysis module 65, which is used to use the Boruta algorithm to determine the importance of different biomarkers in the acute coronary syndrome early warning model.

[0167] Based on the same inventive concept, a computer device is also provided in an embodiment of the present invention to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome based on metabolite data. Figure 8 A schematic diagram of a computer device provided in an embodiment of the present invention is shown in FIG. Figure 8 As shown, the computer device 80 includes a memory 801, a processor 802, and a computer program stored in the memory 801 and executable on the processor 802. When the processor 802 executes the computer program, the above-mentioned acute coronary syndrome early warning method based on metabolomics data is implemented.

[0168] Based on the same inventive concept, a computer-readable storage medium is also provided in an embodiment of the present invention to solve the technical problem that the prior art cannot realize early warning of acute coronary syndrome based on metabolite data. The computer-readable storage medium stores a computer program for executing the above-mentioned acute coronary syndrome early warning method based on metabolomics data.

[0169] In summary, an embodiment of the present invention provides an acute coronary syndrome early warning method, apparatus, computer equipment and computer-readable storage medium based on metabolomics data. After obtaining the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals, differential metabolite data are screened out according to the metabolite data of normal coronary artery biological individuals and acute coronary syndrome biological individuals, and then the LightGBM model is machine-learned based on the screened differential metabolite data to obtain an acute coronary syndrome early warning model containing multiple biomarkers, so that based on the acute coronary syndrome early warning model, the index value of each biomarker corresponding to the biological individual to be diagnosed is collected, and it is determined whether the index value of each biomarker corresponding to the biological individual to be diagnosed exceeds the preset normal value range, and the corresponding early warning information is output according to the judgment result.

[0170] Through the embodiments of the present invention, a non-invasive method for screening biomarkers for acute coronary syndrome can be provided, and an acute coronary syndrome early warning can be performed on patients based on their metabolite data.

[0171] Implementation Plan 2: Stable Coronary Heart Disease Risk Assessment

[0172] The embodiment of the present invention provides a stable coronary heart disease early warning method based on metabolomics data.

[0173] (I) Clinical inclusion criteria: Based on the clinical characteristics of coronary atherosclerotic heart disease, patients were divided into two groups, including: (1) stable CAD group (plaque stability group), i.e., stable CAD group, sCAD, N = 92; (2) normal coronary artery group, NCA, N = 93. Based on the collection of clinical information, blood samples were collected from each group, and plasma was obtained by centrifugation for metabolomics analysis.

[0174] (II) Implementation methods:

[0175] Clinical sample collection and pre-processing: Metabolomics data of 185 participants (biological individuals) were collected. According to the diagnostic guidelines and exclusion criteria, the collected metabolomics data were divided into the following two groups: NCA group (N = 93) and sCAD group (N = 92).

[0176] Blood samples were collected from each participant in the morning when the fasting time was greater than 10 hours, and the hospital completed the relevant clinical routine biochemical index tests. All tests were performed in accordance with international standard methods. After the test, the samples were collected in a test tube containing sodium heparin, cooled to 4°C, and centrifuged within 1 hour to obtain plasma (1,350rpm, centrifugation for 12 minutes), divided into aliquots and stored in a -80°C refrigerator.

[0177] The metabolomics sample pretreatment and LC-HRMS analysis conditions are the same as those in the first embodiment described above.

[0178] Metabolomics data processing and differential metabolite screening: The raw data obtained from LC-MS analysis were imported into Progenesis QI for peak identification, peak alignment, and data normalization to obtain a two-dimensional data matrix containing information such as mass-to-charge ratio, retention time, original peak area, and normalized peak area. QC samples were used for data quality assessment. For the quality-controlled data, a metabolomics processing method based on metabolic pathway mapping of the Mummichog algorithm was used to quickly match the molecular weight information in the metabolite database with the detected ion information with an appropriate mass-to-charge ratio tolerance, and the metabolites were "tentatively labeled" with high throughput and directly mapped to the metabolic pathway based on the KEGG database. Through the overall changes in metabolic pathways and the correlation of metabolic reactions, the differences between groups were modularly analyzed, so as to efficiently discover and identify reliable biomarkers with biological significance. The differential metabolites obtained by preliminary identification were identified by LC-MS / MS technology for metabolite structure identification, and the metabolites with available standards were compared with the standards to further confirm the structure. For metabolites whose structures could not be determined due to low abundance or not included in the secondary database and the inability to obtain standards, the preliminary identification results were temporarily used. Then, a model was established for prediction based on the final differential metabolite abundance information.

[0179] Based on the LightGBM machine learning algorithm, build a prediction model and use external data to verify the model:

[0180] (1) Use the Boruta algorithm for feature selection. The goal of Boruta is to select all feature sets related to the dependent variable, rather than selecting a feature set that minimizes the model cost function for a specific model. The significance of the Boruta algorithm is that it can help us understand the influencing factors of the dependent variable more comprehensively, so as to perform feature selection better and more efficiently.

[0181] (2) GridSearchCV (grid search) adjusts parameters, that is, within the specified parameter range, adjusts the parameters in sequence according to the step size, uses the adjusted parameters to train the learner, and finds the parameters with the highest accuracy on the validation set from all the parameters. This is actually a cycle and comparison process.

[0182] (3) LightGBM modeling: LightGBM is a more powerful and faster model than Xgboost, with greatly improved performance. Compared with traditional algorithms, it has the following advantages: faster training efficiency, low memory usage, higher accuracy, support for parallel learning, and the ability to process large-scale data.

[0183] (4) Select important features in the model and remodel it;

[0184] (5) Use Hyperopt to further tune the parameters of the new model. Hyperopt is a tool that adjusts parameters through Bayesian optimization. This method is fast and has good results. In addition, Hyperopt can be combined with MongoDB to perform distributed parameter adjustment and quickly find relatively optimal parameters;

[0185] (6) LightGBM modeling prediction;

[0186] (7) Re-acquire a batch of external data that has never been involved in modeling, use the constructed model to predict this batch of data, and use AUROC to judge the quality of the prediction model.

[0187] Based on the above algorithm, a stable coronary heart disease prediction model for the normal population was obtained. The marker information used in the model is shown in Table 6.

[0188] Table 6 Biomarker information

[0189]

[0190] In the above Table 6, “*” represents the 2 Identified metabolites; others were preliminarily identified by primary mass spectrometry.

[0191] Fig. 9 The AUROC curve of the stable coronary heart disease prediction model is shown in FIG. Fig. 9 As shown, the AUROC of the training data sample is 0.99883, and the AUROC of the test data sample is 0.91967. Fig.10 Shown are multiple stable coronary heart disease markers screened from metabolomics data and their importance.

[0192] Implementation Plan 3: Acute Coronary Syndrome Risk Assessment for Patients with Stable Coronary Heart Disease

[0193] An embodiment of the present invention provides an acute coronary syndrome early warning method for patients with stable coronary heart disease.

[0194] (I) Clinical inclusion criteria: Based on the clinical characteristics of coronary artery atherosclerotic heart disease, patients were divided into two groups, including: (1) ST-segment elevation acute myocardial infarction (STEMI, unstable plaque rupture group, myocardial necrosis); non-ST-segment elevation acute myocardial infarction (NSTEMI, unstable plaque partial rupture group, a small amount of myocardial necrosis) and unstable angina pectoris (UAP, plaque on the verge of rupture or unstable before rupture group, myocardial micronecrosis), namely ACS group, N = 102; (2) stable CAD group (plaque stable group), namely stable CAD (sCAD) group, N = 92. Based on the collection of clinical information, blood samples were collected from each group, and plasma was obtained by centrifugation for metabolomics analysis.

[0195] (II) Implementation methods:

[0196] Clinical sample collection and pre-processing: Metabolomics data of 194 participants (biological individuals) were collected. According to the diagnostic guidelines and exclusion criteria, the collected metabolomics data were divided into the following two groups: sCAD group (N=92) and ACS group (N=102).

[0197] Blood samples were collected from each participant in the morning when the fasting time was greater than 10 hours, and the hospital completed the relevant clinical routine biochemical index tests. All tests were performed in accordance with international standard methods. After the test, the samples were collected in a test tube containing sodium heparin, cooled to 4°C, and centrifuged within 1 hour to obtain plasma (1,350rpm, centrifugation for 12 minutes), divided into aliquots and stored in a -80°C refrigerator.

[0198] The process of sample pretreatment before metabolomics analysis, C-HRMS analysis conditions, metabolomics data processing and differential metabolite screening, building a prediction model based on the LightGBM machine learning algorithm, and using external data for model verification is basically the same as the above-mentioned Implementation Plan 2. In this implementation plan, an acute coronary syndrome prediction model for patients with stable coronary heart disease is obtained, and the marker information used in the model is shown in Table 7.

[0199] Table 7 Biomarker information

[0200]

[0201]

[0202] In the above Table 7, “*” represents the 2 Identified metabolites; “#” represents metabolites identified by comparison with standard substances; others are preliminary identifications by primary mass spectrometry.

[0203] Fig.11 The AUROC curve of the prediction model for acute coronary syndrome in patients with stable coronary heart disease is shown in FIG.

[0204] Fig.11 As shown, the AUROC of the training data sample is 0.99981, and the AUROC of the test data sample is 0.914332.

[0205] Fig.12 Several acute coronary syndrome markers screened from metabolite data and their importance are shown.

[0206] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0207] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, 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, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0208] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0209] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0210] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. Application of reagents for detecting metabolite levels in samples from individuals in the preparation of cardiovascular disease risk assessment devices; in, The cardiovascular disease is stable coronary heart disease, acute coronary syndrome or acute coronary syndrome for stable coronary heart disease; in: The cardiovascular disease is stable coronary heart disease, and the metabolites include: 6-phospho-2-dehydro-D-gluconic acid, 4-phosphopanthenol-L-cysteine, piperidine, chorismic acid, vitamin A, methyl piperidine, biotin, retinoic acid, N-formylkynurenine, 7α-hydroxycholesterol, glycocholic acid, sorbitol, leukotriene A4, protyrosine, desthiobiotin and clopidogrel carboxylic acid; or The cardiovascular disease is acute coronary syndrome, and the metabolites include: 6-phospho-2-dehydro-D-gluconic acid, retinoic acid, pantothenic acid, protyrosine, 4-phosphopantothenoylcysteine, aconitic acid, N-acetyl-L-phenylalanine, dihydrouracil, dopaquinone, 7α-hydroxycholesterol, 6-phosphoglucose, 2-hydroxybutyric acid, lysine, 2,3-dihydroxybutyric acid, rhamno-1,4-lactone, salicylic acid, clopidogrel carboxylic acid, N-formylkynurenine, leukotriene B4, hydroxyethyllysine, 3-hydroxy-L-kynurenine, succinic acid and galactose; or The cardiovascular disease is acute coronary syndrome for stable coronary heart disease, and the metabolites include: retinoic acid, 6-phospho-2-dehydro-D-gluconic acid, succinic acid, vitamin A, sorbitol, 7,8-dihydroxykynurenine, clopidogrel carboxylic acid, leukotriene A4, lysine, threose, piperidine, rhamn-1,4-lactone, dihydrouracil, pantothenic acid, glucose, protocatechuic aldehyde, chorismic acid, 6-phosphoglucose, threonine, aconitic acid, tryptophan, hydroxyethyllysine, riboflavin, dopaquinone, 2-hydroxybutyric acid, pimelic acid, protyrosine, galactose, 2-hydroxyphenylpropionic acid, 7α-hydroxycholesterol and ribose.

2. The use according to claim 1, wherein: The characteristic importance of each metabolite in assessing the risk of cardiovascular disease is in the following order: 6-phospho-2-dehydro-D-gluconic acid > 4-phosphopanthin-L-cysteine ​​> piperidine > chorismic acid = vitamin A > methylpiperidine > biotin = retinoic acid > N-formylkynurenine = 7α-hydroxycholesterol = glycocholic acid > sorbitol > leukotriene A4 = protyrosine > desthiobiotin > clopidogrel carboxylic acid; 6-phospho-2-dehydro-D-gluconic acid ﹥ retinoic acid ﹥ pantothenic acid = protyrosine ﹥ 4-phosphopantothenoylcysteine ​​﹥ aconitic acid ﹥ N-acetyl-L-phenylalanine ﹥ dihydrouracil ﹥ dopaquinone ﹥ 7α-hydroxycholesterol ﹥ 6-phosphoglucose = 2-hydroxybutyric acid ﹥ lysine ﹥ 2,3-dihydroxybutyric acid ﹥ rhamnosyl-1,4-lactone ﹥ salicylic acid = clopidogrel carboxylic acid ﹥ N-formylkynurenine = leukotriene B4 ﹥ hydroxyethyllysine = 3-hydroxy-L-kynurenine = succinic acid = galactose; Retinoic acid ﹥ 6-phospho-2-dehydro-D-gluconic acid ﹥ Succinic acid ﹥ Vitamin A = Sorbitol = 7,8-dihydroxykynurenine = Clopidogrel carboxylic acid ﹥ Leukotriene A4 = Lysine = Threonose ﹥ Piperidine = Rhamn-1,4-lactone = Dihydrouracil = Pantothenic acid ﹥ Glucose = Protocatechuic aldehyde ﹥ Choroidal acid = 6-phosphoglucose = Threonine ﹥ Aconitic acid = Tryptophan = Hydroxyethyllysine ﹥ Riboflavin = Dopaquinone = 2-hydroxybutyric acid = Pimelic acid = Protyrosine = Galactose = 2-hydroxyphenylpropionic acid = 7α-hydroxycholesterol = Ribose.

3. The use according to claim 1, wherein: When the metabolites are used to assess the risk of cardiovascular disease, the weight of each metabolite is determined according to the following characteristic importance values, or the weight ratio of each metabolite is: Group 1: 6-phospho-2-dehydro-D-gluconic acid 44, 4-phosphopanthin-L-cysteine ​​24, piperidine 20, chorismic acid 12, vitamin A 12, methylpiperidine 8, biotin 7, retinoic acid 7, N-formylkynurenine 6, 7α-hydroxycholesterol 6, glycocholic acid 6, sorbitol 4, leukotriene A4 3, pro-tyrosine 3, desthiobiotin 2, clopidogrel carboxylic acid 1; Group 2: 6-phospho-2-dehydro-D-gluconic acid 38, retinoic acid 36, pantothenic acid 20, protyrosine 20, 4-phosphopantothenoylcysteine ​​19, aconitic acid 18, N-acetyl-L-phenylalanine 15, dihydrouracil 13, dopaquinone 12, 7α-hydroxycholesterol 11, 6-phosphoglucose 10, 2-hydroxybutyric acid 10, lysine 9, 2,3-dihydroxybutyric acid 6, rhamno-1,4-lactone 4, salicylic acid 3, clopidogrel carboxylic acid 3, N-formylkynurenine 2, leukotriene B4 2, hydroxyethyllysine 1, 3-hydroxy-L-kynurenine 1, succinic acid 1, galactose 1; Group 3: Retinoic acid 15, 6-phospho-2-dehydro-D-gluconate 13, succinic acid 8, vitamin A 7, sorbitol 7, 7,8-dihydroxykynurenine 7, clopidogrel carboxylic acid 7, leukotriene A4 6, lysine 6, threose 6, piperidine 5, rhamn-1,4-lactone 5, dihydrouracil 5, pantothenic acid 5, glucose 4, protocatechuic aldehyde 4, chorismic acid 3, 6-phosphoglucose 3, threonine 3, aconitic acid 2, tryptophan 2, hydroxyethyllysine 2, riboflavin 1, dopaquinone 1, 2-hydroxybutyric acid 1, pimelic acid 1, pro-tyrosine 1, galactose 1, 2-hydroxyphenylpropionic acid 1, 7α-hydroxycholesterol 1, ribose 1.

4. The use according to claim 1, wherein: The sample from the individual is a blood sample.

5. The use according to claim 4, wherein: The blood sample is a serum or plasma sample.

6. A cardiovascular disease risk assessment device, comprising a detection unit and a data analysis unit, wherein: The detection unit is used to detect the level of metabolites in a sample from an individual to obtain a detection result; wherein the metabolites are the same as the metabolites described in any one of claims 1 to 5; The data analysis unit is used to analyze and process the detection result of the detection unit.

7. The cardiovascular disease risk assessment device according to claim 6, wherein: The detection unit includes reagent materials for detecting the level of metabolites in a blood sample.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the following are achieved: obtaining an individual cardiovascular disease risk assessment result based on the metabolite level in the individual sample to be tested; Wherein, the metabolite is the same as the metabolite according to any one of claims 1-5.

9. A computer-readable storage medium storing computer program instructions, which when executed realize: obtaining an individual cardiovascular disease risk assessment result based on the metabolite level in the individual sample to be tested; in, The metabolite is the same as the metabolite according to any one of claims 1 to 5.

Images