Method for screening heart failure oxidized lipid after AMI based on solid-phase extraction, LC-MS / MS and AI

Abstract

The invention discloses a post-AMI heart failure oxidized lipid screening method based on solid phase extraction, LC-MS / MS and AI. The method comprises the steps of blood sample pretreatment, solid phase extraction and purification, LC-MS / MS detection and AI model screening. According to the method, high-sensitivity qualitative and quantitative detection of oxidized lipid is realized through optimized LC-MS / MS parameters, and key oxidized lipid molecules and risk levels of heart failure after AMI emergency treatment blood supply reconstruction can be quickly and accurately screened in combination with a trained random forest AI model; the kit solves the problems that in the prior art, trace oxidized lipid detection sensitivity is low, specificity is poor, and heart failure risk screening lags, has the advantages of being high in detection efficiency, good in accuracy and high in practicability, can provide a scientific basis for early warning and intervention of clinical heart failure after AMI, and has high clinical application value.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biological detection, and particularly relates to an AMI post heart failure oxidized lipid screening method based on solid phase extraction, LC-MS / MS and AI. BACKGROUND

[0002] Acute myocardial infarction (AMI) is a common clinical acute and critical illness, and emergency blood supply reconstruction is the main means for treating AMI at present, which can effectively restore the blood perfusion of ischemic myocardium and reduce the mortality. However, heart failure may still occur in some patients after blood supply reconstruction, which becomes an important factor affecting the prognosis of patients. Therefore, early screening of key markers of heart failure after AMI blood supply reconstruction, early warning and intervention, have important significance for improving the prognosis of patients.

[0003] Oxidized lipids are products of fatty acid oxidative metabolism, and widely participate in the inflammatory response and immune regulation process of the body. Among them, pro-resolving factors (such as Resolvin, Protectin, Maresin, etc.) are a class of oxidized lipids with anti-inflammatory and pro-inflammatory resolution effects, which can affect the repair process after myocardial injury by regulating the inflammatory microenvironment. Studies have shown that oxidized lipids, especially pro-inflammatory resolution factors, have very low content in blood (usually in the order of pg / mL to ng / mL), and their content changes are closely related to the occurrence and development of heart failure after AMI blood supply reconstruction, and can be used as potential biomarkers for early screening of heart failure.

[0004] Due to the low content of oxidized lipids in blood and the complexity of plasma matrix (containing a large amount of protein, lipid, electrolyte and other interfering substances), the sensitivity and specificity of the detection technology are extremely high. At present, the commonly used detection method in the literature is solid phase extraction combined with high sensitivity LC-MS / MS technology, and the solid phase extraction column is mostly Agilent Bond Elut Certify II column. However, the solid phase extraction column has the following defects:

[0005] First, the retention capacity of the solid phase extraction column for polar micro-oxidized lipids such as pro-inflammatory resolution factors is insufficient, resulting in low recovery rate (usually less than 60%);

[0006] Second, the removal effect of the solid phase extraction column on the interfering substances in the plasma matrix is limited, the matrix effect is strong, and the accuracy of the detection is affected;

[0007] Third, the stability of the filler is general, the service life is short, and the detection cost is increased.

[0008] Therefore, how to improve the sensitivity and accuracy of the detection to meet the accurate determination of oxidized lipids in blood is the research difficulty in the current field.

[0009] Moreover, current technologies can only quantitatively detect oxidized lipids, and cannot directly link them to the risk of heart failure after revascularization following acute myocardial infarction (AMI). Clinicians need to make a comprehensive judgment based on a large amount of clinical data, which has problems such as low screening efficiency, strong subjectivity, and lag.

[0010] Artificial intelligence algorithms have powerful data mining and pattern recognition capabilities, and can build predictive models based on multi-dimensional biomarker data to achieve rapid and accurate screening of disease risks.

[0011] However, there are currently no reports on combining solid-phase extraction technology, LC-MS / MS detection, and AI algorithms for screening key oxidized lipids in heart failure after revascularization following acute myocardial infarction (AMI). Summary of the Invention

[0012] The purpose of this invention is to provide a method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and artificial intelligence (AI). This invention offers the advantages of achieving highly sensitive and specific detection of trace amounts of oxidized lipids, facilitating rapid and accurate screening for heart failure risk.

[0013] The technical solution of this invention:

[0014] A method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and artificial intelligence (AI) includes the following steps:

[0015] A. Blood sample pretreatment: Collect blood samples, add anticoagulant and centrifuge to separate plasma; add internal standard solution and protein precipitant to plasma, vortex mix and centrifuge to collect supernatant; the blood sample is peripheral venous blood from patients after emergency revascularization following AMI;

[0016] B. Solid-phase extraction purification: The supernatant obtained in step A is loaded into a solid-phase extraction column, and activation, equilibration, sample loading, rinsing and elution are performed in sequence. The eluent is collected and dried with nitrogen gas, reconstituted with the reconstitution solution and filtered through a membrane to obtain the sample to be tested.

[0017] C. LC-MS / MS detection: The sample to be tested obtained in step B is injected into an ultra-high performance liquid chromatography-tandem mass spectrometer for qualitative and quantitative detection of oxidized lipids, and the peak area and concentration data of each oxidized lipid are obtained.

[0018] D. AI Model Screening: Input the concentration data of each oxidized lipid obtained in step C into the artificial intelligence screening model, and output the risk level of heart failure in patients after emergency revascularization of AMI and the corresponding key oxidized lipid molecules; the artificial intelligence screening model is a classification model built based on the random forest algorithm, the model input features are the concentration values ​​of at least 15 oxidized lipids, and the model output is three heart failure risk levels: low, medium and high.

[0019] In the aforementioned method for screening oxidized lipids in heart failure after AMI based on solid-phase extraction, LC-MS / MS and AI, in step A, the anticoagulant is EDTA-K2, and the volume ratio of the anticoagulant to blood is 1:9; the internal standard solution is a deuterated labeled oxidized lipid mixture, the concentration of the internal standard solution is 100 ng / mL, and the amount added is 1 / 10 of the plasma sample volume;

[0020] The protein precipitant is an acetonitrile-methanol mixed solution with a volume ratio of 1:1, and the volume ratio of the protein precipitant to the plasma sample is 3:1.

[0021] In the aforementioned method for screening oxidized lipids in heart failure after AMI based on solid-phase extraction, LC-MS / MS and AI, in step A, the conditions for centrifuging and separating plasma are 4°C, 3500 r / min, and centrifugation for 10 min.

[0022] The vortex mixing conditions are 2500 r / min, vortex for 2 min;

[0023] The conditions for centrifuging to obtain the supernatant were 4℃, 12000 r / min, and centrifugation for 15 min.

[0024] In the aforementioned method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and AI, the specific details of the solid-phase extraction column in step B are as follows:

[0025] The solid-phase extraction column has a column volume of 3 mL and a packing amount of 60 mg.

[0026] The solid-phase extraction column is packed with an amino-modified silica gel-polymer hybrid packing material, with a particle size of 20-40 μm, a pore size of 60-100 Å, and a specific surface area of ​​500-700 m². 2 / g;

[0027] This solid-phase extraction column exhibits excellent retention and purification capabilities for trace amounts of oxidized lipids, significantly improving the recovery rate and accuracy of detection.

[0028] In the aforementioned method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and AI, the preparation method of the amino-modified silica-polymer hybrid filler is as follows:

[0029] The sol-gel method was used, with tetraethyl orthosilicate as the silicon source and 3-aminopropyltriethoxysilane as the amination reagent. The mixture was stirred and hydrolyzed in an ethanol-water mixture at 40°C for 4 hours to obtain an amination silica gel precursor.

[0030] Subsequently, methyl methacrylate monomer and azobisisobutyronitrile initiator were added, and polymerization was carried out at 60°C for 6 hours under nitrogen protection. After washing, filtration, and vacuum drying at 50°C for 12 hours, the product was obtained.

[0031] In the aforementioned method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and AI, the specific details of step B—activation, equilibration, loading, rinsing, elution, nitrogen drying, and reconstitution—are as follows:

[0032] Activation: Add 2 mL of methanol at a flow rate of 1 mL / min and let stand for 1 min;

[0033] Equilibration: Add 2 mL of 0.1% formic acid aqueous solution at a flow rate of 1 mL / min, and let stand for 1 min;

[0034] Sample loading: The sample loading flow rate is 0.5 mL / min;

[0035] Rinsing: Add 2 mL of 0.1% formic acid aqueous solution and 2 mL of 5% methanol aqueous solution sequentially, at a flow rate of 1 mL / min for both.

[0036] Elution: Add 2 mL of a methanol-acetonitrile mixed solution containing 0.1% formic acid, with a methanol to acetonitrile volume ratio of 1:1, at a flow rate of 0.5 mL / min;

[0037] Nitrogen drying conditions were 37℃ and 0.3 MPa;

[0038] The reconstitution solution is a methanol-water mixture containing 0.1% formic acid, with a volume ratio of 3:7, and the reconstitution volume is 100 μL.

[0039] The membrane was a 0.22 μm organic phase filter membrane.

[0040] In the aforementioned method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and AI, the chromatographic conditions in step C are as follows:

[0041] The chromatographic column was a C18 column with dimensions of 2.1 mm × 100 mm, a particle size of 1.7 μm, and a column temperature of 40℃.

[0042] Mobile phase A is an aqueous solution containing 0.1% formic acid, and mobile phase B is an acetonitrile solution containing 0.1% formic acid;

[0043] The gradient elution program was as follows: 0-2 min, 10% B; 2-8 min, 10%-80% B; 8-12 min, 80%-95% B; 12-15 min, 95% B; 15-16 min, 95%-10% B; 16-20 min, 10% B.

[0044] The flow rate was 0.3 mL / min;

[0045] The injection volume was 5 μL.

[0046] In the aforementioned method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and AI, the mass spectrometry conditions in step C are as follows:

[0047] The ion source is an electrospray ionization source (ESI) in negative ion mode;

[0048] The spray voltage is -4500 V;

[0049] The ion source temperature is 550℃;

[0050] The atomizing gas pressure is 50 psi;

[0051] The auxiliary gas pressure is 50 psi;

[0052] The air curtain pressure is 35 psi;

[0053] The impact pressure is 10 psi;

[0054] Multiple response monitoring (MRM) was used for detection.

[0055] In the aforementioned method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and AI, the artificial intelligence screening model described in step D is as follows:

[0056] D1. Sample set construction: Collect clinical data, heart failure occurrence and corresponding lipid peroxidation test data of patients after emergency revascularization in AMI, divide patients into heart failure group and non-heart failure group, and construct sample set; 70% of the sample set is training set and 30% is validation set.

[0057] D2. Feature Selection: Using ANOVA and random forest feature importance ranking, feature variables that are significantly related to the occurrence of heart failure are selected from the detected oxidized lipids to determine the input features of the model.

[0058] D3. Model Training: Based on the training set data, a classification model is built using the random forest algorithm. The number of decision trees is set to 100-200, the maximum tree depth is 10-15, the minimum number of sample splits is 2, and the minimum number of sample leaf nodes is 1.

[0059] D4. Model Optimization and Validation: The model is optimized using 5-fold cross-validation, and its performance is verified using the validation set data.

[0060] A screening system for oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and AI, comprising a detection module, an AI analysis module, a display module, and a data storage module;

[0061] The detection module uses the aforementioned LC-MS / MS detection method to obtain oxidized lipid concentration data;

[0062] The AI ​​analysis module has a built-in trained random forest classification model that receives detection data and outputs risk levels and key molecules.

[0063] The display module is used to display the detection results and AI analysis results;

[0064] The data storage module is used to store patient-related data and supports querying and exporting.

[0065] Compared with the prior art, the beneficial effects of this application are as follows:

[0066] 1. Significant advantages of solid phase extraction columns: This invention uses aminated silica-polymer hybrid packing material to replace the traditional Agilent Bond Elut Certify II column. Through unique aminated modification and hybrid structure design, this packing material has good retention capacity for both polar and non-polar oxidized lipids. In particular, the recovery rate of trace amounts of polar oxidized lipids such as pro-inflammatory de-inflammatory factors can be increased to more than 85%, which is significantly higher than that of traditional columns.

[0067] Meanwhile, the packing material has excellent removal effect on interfering substances in the plasma matrix, reducing the matrix effect to below 10%, thus improving the accuracy and repeatability of the detection.

[0068] 2. Excellent LC-MS / MS detection performance: By optimizing the chromatographic column, mobile phase, gradient elution program and mass spectrometry parameters, efficient separation and high-sensitivity detection of various target oxidized lipids are achieved, with detection limits of 0.1-0.5 pg / mL and quantitation limits of 0.3-1.5 pg / mL, meeting the detection requirements for trace amounts of oxidized lipids in blood.

[0069] 3. AI model enables precise screening: By combining lipid peroxidation detection data with clinical indicators, a random forest AI model is constructed, which can quickly and accurately screen for the risk of heart failure after revascularization following acute myocardial infarction (AMI). The model has an accuracy of ≥88%, sensitivity of ≥85%, and specificity of ≥90%. Compared with traditional clinical judgment methods, the screening efficiency and accuracy are significantly improved, enabling early warning of heart failure.

[0070] 4. High practicality: The method of this invention is simple to operate, has a short detection cycle (the entire process can be completed within 4 hours), and requires a small sample volume (only 200 μL of plasma). It can be used for routine screening of patients after revascularization in clinical AMI.

[0071] Meanwhile, the screening results can identify key oxidized lipid molecules, providing targeted targets for clinical intervention, which has significant clinical application value and promising prospects for promotion. Attached Figure Description

[0072] Figure 1 This is a flowchart illustrating the screening method of the present invention. Detailed Implementation

[0073] The present invention will be further described below with reference to the accompanying drawings and embodiments, but this should not be construed as limiting the present invention.

[0074] Implementation.

[0075] A method for screening oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and artificial intelligence (AI), such as Figure 1 As shown, it includes the following steps:

[0076] A. Blood sample pretreatment: Collect 5 mL of peripheral venous blood from patients after emergency revascularization following AMI, add 0.5 mL of LEDTA-K2 anticoagulant, gently invert and mix 3-5 times, centrifuge at 4℃ and 3500 r / min for 10 min, separate the supernatant plasma, and store it in a -80℃ refrigerator for later use.

[0077] Thaw 200 μL of frozen plasma sample to room temperature, add 20 μL of internal standard solution (a mixed solution of deuterated-15-HETE, deuterated-12-HETE, and deuterated-RvD1, all at a concentration of 100 ng / mL), and vortex for 30 s. Then add 600 μL of acetonitrile-methanol mixed protein precipitant (volume ratio 1:1), vortex at 2500 r / min for 2 min, and then centrifuge at 4℃ and 12000 r / min for 15 min. Collect the supernatant for later use.

[0078] B. Solid-phase extraction purification: Purification is performed using a solid-phase extraction column. The specific procedures are as follows:

[0079] B1. Activation: Add 2 mL of methanol to the solid phase extraction column and pass it through the column at a flow rate of 1 mL / min. Let it stand for 1 min to fully activate the packing material.

[0080] B2. Equilibration: Add 2 mL of 0.1% formic acid aqueous solution, pass it through the column at a flow rate of 1 mL / min, and let it stand for 1 min to allow the column bed to equilibrate to an aqueous environment;

[0081] B3. Sample loading: Slowly load the supernatant obtained in step A into the solid phase extraction column at a flow rate of 0.5 mL / min to ensure that the target analyte is fully retained on the packing material;

[0082] B4. Eluting: Add 2 mL of 0.1% formic acid aqueous solution and 2 mL of 5% methanol aqueous solution sequentially, passing both through the column at a flow rate of 1 mL / min to remove interfering substances such as proteins and salts adsorbed on the column;

[0083] B5. Elution: Add 2 mL of methanol-acetonitrile mixed eluent (volume ratio 1:1) containing 0.1% formic acid, pass through the column at a flow rate of 0.5 mL / min, and collect the eluent;

[0084] B6. Reconstitution: Place the eluent in a nitrogen evaporator and dry it with nitrogen at 37°C and 0.3 MPa. Add 100 μL of a methanol-water mixed solution containing 0.1% formic acid (volume ratio 3:7), vortex mix for 2 min, and then filter through a 0.22 μm organic phase filter membrane to obtain the sample to be tested. Place it in a sample bottle for LC-MS / MS detection.

[0085] C. LC-MS / MS detection: Detection was performed using an ultra-high performance liquid chromatography-tandem mass spectrometer (AB Sciex, QTRAP6500+). Specific parameters are as follows:

[0086] C1. Chromatographic conditions:

[0087] Chromatographic column: Waters ACQUITY UPLC BEH C18 column (2.1 mm × 100 mm, 1.7 μm);

[0088] Column temperature: 40℃;

[0089] Mobile phase: Phase A is an aqueous solution containing 0.1% formic acid, and Phase B is an acetonitrile solution containing 0.1% formic acid;

[0090] Gradient elution program: 0-2 min, 10% B; 2-8 min, 10%-80% B; 8-12 min, 80%-95% B; 12-15 min, 95% B; 15-16 min, 95%-10% B; 16-20 min, 10% B;

[0091] Flow rate: 0.3 mL / min;

[0092] Injection volume: 5 μL.

[0093] C2. Mass spectrometry conditions:

[0094] Ion source: Electrospray ionization source (ESI), negative ion mode;

[0095] Spray voltage: -4500 V;

[0096] Ion source temperature: 550℃;

[0097] Nebulizer gas pressure: 50 psi;

[0098] Auxiliary gas pressure: 50 psi;

[0099] Air curtain pressure: 35 psi;

[0100] Impact pressure: 10 psi;

[0101] Detection mode: Multiple reaction monitoring (MRM). The MRM parameters for each target oxidized lipid are shown in Table 1.

[0102] Adding 0.1% formic acid in negative ion mode can improve the peak shape of oxidized lipids and enhance detection stability. Experiments have verified that the response intensity of the target analyte is increased by more than 20% under this condition.

[0103] D. AI model screening:

[0104] D1. Model Building:

[0105] D1.1 Sample Set Construction: Clinical samples were collected from 200 patients who underwent emergency revascularization after acute myocardial infarction (AMI), including 30 patients in the heart failure group and 170 patients in the non-heart failure group. Plasma lipid peroxidation data (i.e., the concentrations of the eight lipid peroxidation substances obtained in step C) and clinical data (age, sex, BMI, history of hypertension, history of diabetes, etc.) were collected from all patients to construct the sample dataset. The dataset was randomly divided into a training set (140 cases) and a validation set (60 cases) in a 7:3 ratio.

[0106] D1.2 Feature Selection: Analysis of variance (ANOVA) was used to select variables that were significantly associated with the occurrence of heart failure (P<0.05). Then, the feature importance of each variable was calculated by the random forest algorithm, and the top 15 variables with the highest feature importance were selected as input features of the model, including the concentrations of RvD1, RvE1, PD1, MaR1, 15-HETE, 12-HETE, PGE2, LTB4 and 7 clinical indicators.

[0107] D1.3 Model Training: Based on the training set data, a classification model is constructed using the random forest algorithm. The number of decision trees is set to 150, the maximum tree depth to 12, the minimum number of sample splits to 2, and the minimum number of leaf nodes to 1. Five-fold cross-validation is used to optimize the model to improve its generalization ability.

[0108] D2. Model Validation: The trained model is validated using validation set data. Evaluation metrics include accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and area under the ROC curve (AUC).

[0109] D3. Risk Screening: Input the lipid oxidation concentration data of the patient to be tested into the trained AI model. The model outputs the risk level (low, medium, high) of the patient's heart failure and the corresponding key lipid oxidation molecules (i.e., the lipid oxidation that contributes the most to risk prediction).

[0110] Table 1. MRM parameters of the target oxidized lipids

[0111]

[0112] The solid-phase extraction column described in step B is filled with an aminated modified silica gel-polymer hybrid filler. The synthesis method is as follows: using the sol-gel method, tetraethyl orthosilicate is used as the silicon source and 3-aminopropyltriethoxysilane is used as the amination reagent. In an ethanol-water mixture, the mixture is stirred and hydrolyzed for 4 hours at 40°C to obtain an aminated silica gel precursor. Subsequently, methyl methacrylate monomer (mass ratio of 1:8 to the aminated silica gel precursor) and azobisisobutyronitrile initiator (0.8% of the mass of methyl methacrylate monomer) are added. The mixture is polymerized at 60°C for 6 hours under nitrogen protection. After washing, filtration, and vacuum drying at 50°C for 12 hours, the final product is obtained.

[0113] This solid-phase extraction column exhibits excellent retention and purification capabilities for trace amounts of oxidized lipids, significantly improving the recovery rate and accuracy of detection.

[0114] A screening system for oxidized lipids in heart failure after acute myocardial infarction (AMI) based on solid-phase extraction, LC-MS / MS, and AI, comprising a detection module, an AI analysis module, a display module, and a data storage module;

[0115] The detection module uses the above-mentioned LC-MS / MS detection method to obtain oxidized lipid concentration data;

[0116] The AI ​​analysis module has a built-in trained random forest classification model that receives detection data and outputs risk levels and key molecules.

[0117] The display module is used to display the detection results and AI analysis results;

[0118] The data storage module is used to store patient-related data and supports querying and exporting.

[0119] Verification Experiment

[0120] Experiment 1: Performance Comparison Experiment of the Solid Phase Extraction Column of the Present Invention and the Traditional Solid Phase Extraction Column

[0121] 1. Experimental materials

[0122] The solid-phase extraction column of this invention has a column volume of 3 mL and a packing material of 60 mg, which is an aminated silica-polymer hybrid packing material. The synthesis method involves a sol-gel method using tetraethyl orthosilicate as the silicon source and 3-aminopropyltriethoxysilane as the amination reagent. Hydrolysis and condensation are carried out in an ethanol-water mixture at 40°C for 4 h to obtain an aminated silica precursor. Subsequently, methyl methacrylate monomer and azobisisobutyronitrile initiator are added, and polymerization is performed at 60°C for 6 h under nitrogen protection. After washing, filtration, and vacuum drying at 50°C for 12 h, the resulting product has a particle size of 20-40 μm, a pore size of 60-100 Å, and a specific surface area of ​​500-700 m². 2 / g;

[0123] Traditional solid-phase extraction column: Agilent Bond Elut Certify II, Agilent Technologies, column volume 3 mL, packing material 60 mg;

[0124] Eight oxidized lipid standards (purity ≥98%, Sigma-Aldrich);

[0125] Deuterated internal standard (purity ≥98%, Cayman).

[0126] Methanol and acetonitrile (chromatographic grade, Merck).

[0127] Formic acid (chromatographic grade, Thermo Fisher Scientific);

[0128] Ultrapure water (prepared by Millipore ultrapure water system);

[0129] Plasma samples from healthy individuals (sourced from a hospital's health checkup center, excluding cardiovascular and cerebrovascular diseases, inflammatory diseases, etc.).

[0130] 2. Experimental Methods

[0131] 2.1 Preparation of standard solutions: Accurately weigh each oxidized lipid standard, dissolve and dilute to volume with methanol to prepare a standard stock solution with a concentration of 1 mg / mL; dilute the standard stock solution with methanol to obtain standard working solutions with concentrations of 0.1, 0.5, 1, 5, 10, 50, and 100 ng / mL;

[0132] Internal standard solution preparation: concentration 100 ng / mL.

[0133] 2.2 Sample preparation: Take 200 μL of plasma from healthy individuals, add 20 μL of standard working solution (5 ng / mL) and 20 μL of internal standard solution, and vortex to mix for 30 s; add 600 μL of acetonitrile-methanol mixed protein precipitant, vortex, centrifuge, and take the supernatant. Perform solid-phase extraction purification using the solid-phase extraction column of this invention and a conventional solid-phase extraction column, respectively, following the same operation steps as in step B; perform LC-MS / MS detection on the purified sample, using the same detection parameters as described in this invention.

[0134] 2.3 Recovery rate calculation: The spiked recovery method was adopted. Recovery rate = (spiked sample concentration - blank sample concentration) / spiked concentration × 100%;

[0135] For each concentration level, the assay was performed in parallel six times, and the average recovery rate and relative standard deviation (RSD) were calculated.

[0136] 2.4 Evaluation of matrix effect: Matrix effect = (Slope of matrix-matched standard curve / Slope of pure solvent standard curve) × 100%;

[0137] A matrix effect in the range of 85%-115% indicates a small matrix effect.

[0138] 3. Experimental Results

[0139] The experimental results are shown in Table 2. As can be seen from Table 2, the average recovery rate of the solid-phase extraction column of this invention for eight oxidized lipids was 85.6%-92.3%, with an RSD of 2.1%-4.5%.

[0140] Traditional solid-phase extraction columns have an average recovery rate of only 52.4%-58.7% and an RSD of 5.8%-8.3%.

[0141] Regarding the matrix effect, the matrix effect of the solid phase extraction column of this invention is 92%-105%, while the matrix effect of the traditional solid phase extraction column is 72%-83%.

[0142] The results show that the recovery rate of the solid phase extraction column of the present invention is significantly higher than that of the traditional column, and the matrix effect is smaller, resulting in better purification effect.

[0143] Table 2 Recovery Rate Results

[0144]

[0145] Experiment 2 Methodological validation of LC-MS / MS detection method

[0146] 1. Experimental materials

[0147] Similar to Experiment 1, a blank plasma sample (a plasma sample without the target oxidized lipids) was added.

[0148] 2. Experimental Methods

[0149] 2.1 Standard curve plotting: Take 200 μL of blank plasma, add 20 μL of standard working solution of different concentrations and 20 μL of internal standard solution, and perform blood sample pretreatment and solid phase extraction purification operations according to steps A and B, and then perform LC-MS / MS detection;

[0150] A standard curve was plotted with the concentration of the target oxidized lipids on the x-axis and the peak area ratio of the target oxidized lipids to the internal standard on the y-axis. The regression equation and correlation coefficient (r) were then calculated.

[0151] 2.2 Limit of Detection (LOD) and Limit of Quantification (LOQ): The limit of detection is the concentration corresponding to a signal-to-noise ratio (S / N) of 3, and the limit of quantification is the concentration corresponding to an S / N of 10.

[0152] 2.3 Precision Validation: Standard working solutions at three concentration levels (0.5, 10, and 50 ng / mL) were selected and tested according to the above method. Six parallel measurements were performed within each concentration level for three consecutive days, and the intra-day precision (RSD) was calculated. r ) and daytime precision (RSD) R ).

[0153] 2.4 Stability verification: The prepared test samples were placed at room temperature (25℃) for 0, 2, 4, 8, and 12 hours, and frozen in a -80℃ refrigerator for 0, 7, 14, and 30 days. The concentration of the target oxidized lipids was measured, the RSD was calculated, and the short-term and long-term stability of the samples were evaluated.

[0154] 3. Experimental Results

[0155] 3.1 Standard curve, limit of detection and limit of quantitation: The eight oxidized lipids showed good linearity in the concentration range of 0.1-100 ng / mL, with correlation coefficients (r) ≥0.998. The limits of detection were 0.1-0.5 pg / mL and the limits of quantitation were 0.3-1.5 pg / mL. The specific results are shown in Table 3.

[0156] Table 3. Results of Standard Curve, Limit of Detection, and Limit of Quantification

[0157]

[0158] 3.2 Precision: Intra-day precision RSD at low, medium, and high concentration levels r The precision ranges from 1.8% to 4.2%, with a daytime RSD of 1.8%. R The percentages were 2.5%-5.3%, all less than 6%, indicating that the detection method has good precision (see Table 4).

[0159] Table 4 Precision Results

[0160]

[0161] 3.3 Stability: After being placed at room temperature (25℃) for 0, 2, 4, 8, and 12 h, the RSD of each lipid oxidization concentration was 2.1%-4.6%; after being frozen at -80℃ for 0, 7, 14, and 30 days, the RSD of each lipid oxidization concentration was 2.3%-4.8%, all less than 5%, indicating that the samples have good stability under the above storage conditions and can meet the sample preservation requirements for clinical testing. The specific results are shown in Table 5.

[0162] Table 5 Stability Results

[0163]

[0164] Experiment 3: Construction and Validation of an Artificial Intelligence Screening Model

[0165] 1. Experimental materials

[0166] Clinical Sample: Clinical samples were collected from 200 patients who underwent emergency revascularization after acute myocardial infarction (AMI) at a tertiary hospital between January 2023 and January 2024. Among them, 112 were male and 88 were female, aged 45-78 years, with a mean age of (62.3±8.5) years. Based on the occurrence of heart failure 6 months post-surgery, patients were divided into a heart failure group (30 cases) and a non-heart failure group (170 cases). There were no significant differences between the two groups in baseline characteristics such as age, sex, and BMI (P>0.05), making them comparable.

[0167] Experimental equipment: computer (configured with Intel Core i7 processor, 16GB memory), data analysis software Python 3.9, machine learning library Scikit-learn 1.2.2.

[0168] 2. Experimental Methods

[0169] 2.1 Sample Data Collection: Plasma lipid peroxidation detection data (concentrations of 8 lipid peroxidation types detected using the method of this invention) and clinical data were collected from all patients. Clinical data included age, gender, BMI, history of hypertension (yes / no), history of diabetes (yes / no), smoking history (yes / no), infarction site (anterior wall / inferior wall / other), and postoperative LVEF value.

[0170] 2.2 Data Preprocessing: The collected raw data were preprocessed, including missing value imputation (using the mean to imput numerical data and the mode to imput categorical data), outlier removal (using the 3σ criterion), and data standardization (using Z-score standardization).

[0171] 2.3 Sample set division: The preprocessed dataset was randomly divided into a training set (140 cases, including 21 cases in the heart failure group and 119 cases in the non-heart failure group) and a validation set (60 cases, including 9 cases in the heart failure group and 51 cases in the non-heart failure group) in a 7:3 ratio.

[0172] 2.4 Feature selection: First, analysis of variance (ANOVA) was used to preliminarily screen all variables and select variables that were significantly associated with the occurrence of heart failure (P<0.05); then, the feature importance of each variable was calculated by the random forest algorithm, and the top 15 variables in terms of feature importance were selected as the input features of the model.

[0173] 2.5 Model Construction and Optimization: Based on the training set data, a classification model was constructed using the random forest algorithm, and hyperparameters were optimized using GridSearchCV. The optimized hyperparameters included the number of decision trees (100, 150, 200), the maximum tree depth (10, 12, 15), the minimum number of sample splits (2, 5), and the minimum number of sample leaf nodes (1, 2).

[0174] Five-fold cross-validation was used to train and optimize the model, and the optimal combination of hyperparameters was selected using accuracy as the evaluation metric.

[0175] 2.6 Model Validation: The trained optimal model is validated using validation set data. Evaluation metrics include accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and area under the ROC curve (AUC).

[0176] Meanwhile, the performance of the random forest model constructed in this invention is compared with that of the logistic regression model and the support vector machine model.

[0177] 3. Experimental Results

[0178] 3.1 Feature screening results: Analysis of variance initially screened out 12 variables that were significantly associated with the occurrence of heart failure (P<0.05), namely RvD1 concentration, RvE1 concentration, PD1 concentration, MaR1 concentration, 15-HETE concentration, 12-HETE concentration, PGE2 concentration, LTB4 concentration, age, BMI, history of hypertension, and postoperative LVEF value;

[0179] By ranking the importance of features using random forest, the top 15 feature variables (including some interaction features) were finally selected as model inputs, including 8 types of lipid oxidation concentrations and 7 clinically relevant features (age, BMI, history of hypertension, history of diabetes, postoperative LVEF value, infarction site, and smoking history).

[0180] 3.2 Optimal Hyperparameter Combination: After optimization by GridSearchCV, the optimal hyperparameter combination for the random forest model is: 150 decision trees, maximum tree depth 12, minimum number of sample splits 2, and minimum number of sample leaf nodes 1.

[0181] 3.3 Model performance validation results: The validation set data validation results show that the random forest model constructed in this invention has an accuracy of 88.0%, a sensitivity of 85.0%, a specificity of 90.0%, a positive predictive value of 78.0%, a negative predictive value of 93.0%, and an area under the ROC curve (AUC) of 0.89.

[0182] Compared with other models, the random forest model of this invention has better performance, as shown in Table 6.

[0183] Table 6 Model Performance Table

[0184]

[0185] Experiment 4: Clinical application verification experiment of the method of the present invention

[0186] 1. Experimental materials

[0187] Clinical Samples: Peripheral venous blood samples were collected from 50 patients who underwent emergency revascularization after acute myocardial infarction (AMI) at a tertiary hospital between February and June 2024. There were 28 males and 22 females, aged 46-79 years, with a mean age of (63.1±7.9) years. All patients signed informed consent forms, and this study was approved by the hospital's ethics committee.

[0188] Experimental equipment and reagents: Same as Experiments 1 and 2.

[0189] 2. Experimental Methods

[0190] 2.1 Sample testing: Plasma samples from 50 patients were tested using the screening method of this invention to obtain the risk level (low, medium, high) of heart failure and key oxidized lipid molecules for each patient.

[0191] 2.2 Clinical follow-up: Fifty patients were followed up clinically for 6 months after surgery. The actual occurrence of heart failure was recorded, and the clinical diagnosis results were used as the gold standard.

[0192] 2.3 Result Comparison: The screening results of the present invention are compared with the actual clinical situation to calculate the diagnostic accuracy, positive predictive value, and negative predictive value of the method of the present invention.

[0193] 3. Experimental Results

[0194] Clinical follow-up results showed that among the 50 patients, 10 developed heart failure within 6 months post-surgery, while 40 did not. The screening results using the method of this invention were: high-risk 8 cases (7 of which actually developed heart failure), medium-risk 6 cases (2 of which actually developed heart failure), and low-risk 36 cases (1 of which actually developed heart failure). Calculations showed that the positive predictive value of the method of this invention was 64.3% (9 / 14), and the negative predictive value was 97.2% (35 / 36), indicating good consistency between the screening results and the actual clinical situation. Specific results are shown in Table 7.

[0195] Table 7 Clinical Follow-up Experiment Results

[0196]

[0197] Meanwhile, analysis of the key oxidized lipid molecules screened out revealed that the concentrations of RvD1, RvE1, PD1, and MaR1 (pro-inflammatory remission factors) in heart failure patients were significantly lower than those in non-heart failure patients (P<0.01), while the concentrations of 15-HETE, 12-HETE, PGE2, and LTB4 (pro-inflammatory oxidized lipids) were significantly higher in heart failure patients (P<0.01). Among these, decreased RvD1 concentration and increased 15-HETE concentration were the two most critical molecular markers for predicting the occurrence of heart failure.

[0198] Experimental results show that the method of the present invention can accurately screen for the risk of heart failure and key oxidized lipid molecules after emergency revascularization in acute myocardial infarction (AMI), with a high diagnostic accuracy and good reliability in clinical application.

Claims

1. A method for post-AMI heart failure oxidized lipid screening based on solid phase extraction, LC-MS / MS and AI, characterized in that, Comprising the following steps: A, blood sample pretreatment: collect blood samples, add anticoagulant and centrifugal separation of plasma; add internal standard solution and protein precipitant to the plasma, vortex mix and centrifugal take supernatant; B, solid phase extraction purification: the supernatant obtained in step A is loaded into a solid phase extraction column, and is sequentially activated, balanced, loaded, eluted, eluted, and collected, and is dried with nitrogen, and is dissolved with a re-dissolution solution and filtered to obtain a sample to be detected; C, LC-MS / MS detection: the sample to be detected obtained in step B is injected into an ultra-high performance liquid chromatography-tandem mass spectrometer for qualitative and quantitative detection of oxidized lipids, and peak area and concentration data of each oxidized lipid are obtained; D, AI model screening: input the concentration data of each oxidized lipid obtained in step C into the artificial intelligence screening model, and output the risk level of heart failure after AMI revascularization and the corresponding key oxidized lipid molecules; the artificial intelligence screening model is a classification model based on a random forest algorithm, the model input features are the concentration values of multiple oxidized lipids, and the model output is three heart failure risk levels of low, medium and high.

2. The AMI heart failure oxidized lipid screening method based on solid phase extraction, LC-MS / MS and AI according to claim 1, characterized in that: In step A, the anticoagulant is EDTA-K2, and the volume ratio of anticoagulant to blood is 1:9; the internal standard solution is a deuterium-labeled oxidized lipid mixed solution, the internal standard solution concentration is 100 ng / mL, and the addition amount is 1 / 10 of the volume of the plasma sample; The protein precipitant is an acetonitrile-methanol mixed solution with a volume ratio of 1:1, and the volume ratio of protein precipitant to plasma sample is 3:

1.

3. The AMI heart failure oxidized lipid screening method based on solid phase extraction, LC-MS / MS and AI according to claim 1, characterized in that: In step A, the centrifugal separation of plasma is at 4℃, 3500 r / min, and centrifugal for 10 min; The vortex mixing condition is 2500 r / min, vortex for 2 min; The centrifugal separation of plasma is at 4℃, 12000 r / min, and centrifugal for 15 min.

4. The method of claim 1, wherein the method is a solid phase extraction, LC-MS / MS and AI based post-AMI heart failure oxidized lipid screening method. The specific content of the solid phase extraction column in step B is as follows: The column volume of the solid phase extraction column is 3 mL, and the filler filling amount is 60 mg; The filler of the solid phase extraction column is an aminomodified silica gel-polymer hybrid filler, the particle size of the filler is 20-40 μm, the pore size is 60-100 Å, and the specific surface area is 500-700 m 2 / g.

5. The method of claim 4, wherein the method is a solid phase extraction, LC-MS / MS and AI based post-AMI heart failure oxidized lipid screening method. The preparation method of the amino-modified silica gel-polymer hybrid filler is as follows: A sol-gel method is used, tetraethyl orthosilicate is used as a silicon source, 3-aminopropyl triethoxysilane is used as an amination reagent, in an ethanol-water mixed system, hydrolysis and condensation is carried out at 40℃ for 4h to obtain an amino-modified silica gel precursor; Then, methyl methacrylate monomer and azobisisobutyronitrile initiator are added, and polymerization is carried out at 60℃ for 6h under nitrogen protection, and after washing, filtering and vacuum drying at 50℃ for 12h, the amino-modified silica gel-polymer hybrid filler is prepared.

6. The method of claim 1, wherein the method is a solid phase extraction, LC-MS / MS and AI based post-AMI heart failure oxidized lipid screening method. The specific content of the activation, balance, loading, elution, elution, nitrogen drying and re-dissolution in step B is as follows: Activation: add 2 mL of methanol at a flow rate of 1 mL / min, and stand for 1 min; Equilibration: 2 mL of 0.1% formic acid in water was added at a flow rate of 1 mL / min, and was allowed to stand for 1 min; Loading: The loading flow rate was 0.5 mL / min; Elution: 2 mL of 0.1% formic acid in methanol-acetonitrile (1:1, by volume) was added at a flow rate of 0.5 mL / min; Nitrogen blowing drying conditions were 37°C and 0.3 MPa; The redissolving solution was a methanol-water mixture containing 0.1% formic acid, with a volume ratio of 3:7, and the redissolving volume was 100 μL; The membrane was passed through a 0.22 μm organic phase filter membrane. Chromatographic conditions in Step C:

7. The method of claim 1, wherein the method is a method of post-AMI heart failure oxidized lipid screening based on solid phase extraction, LC-MS / MS and AI. The chromatographic column was a C18 column with a size of 2.1 mm x 100 mm and a particle size of 1.7 μm, and the column temperature was 40°C; The mobile phase A was 0.1% formic acid in water, and the mobile phase B was 0.1% formic acid in acetonitrile; The gradient elution program was as follows: 0-2 min, 10% B; 2-8 min, 10%-80% B; 8-12 min, 80%-95% B; 12-15 min, 95% B; 15-16 min, 95%-10% B; 16-20 min, 10% B; The flow rate was 0.3 mL / min; The injection volume was 5 μL. Mass spectrometry conditions in Step C:

8. The method of claim 1, wherein the method is a method of post-AMI heart failure oxidized lipid screening based on solid phase extraction, LC-MS / MS, and AI. The ion source was an electrospray ionization source (ESI) in negative ion mode; The spray voltage was -4500 V; The ion source temperature was 550°C; The nebulization gas pressure was 50 psi; The auxiliary gas pressure was 50 psi; The curtain gas pressure was 35 psi; The collision gas pressure was 10 psi; Detection was performed using multiple reaction monitoring mode (MRM). The artificial intelligence screening model described in Step D has the following specific content:

9. The method of claim 1, wherein the method is a method of post-AMI heart failure oxidized lipid screening based on solid phase extraction, LC-MS / MS and AI. D1, sample set construction: collect the clinical data, heart failure occurrence, and corresponding oxidized lipid detection data of patients after AMI emergency revascularization, divide the patients into heart failure group and non-heart failure group, and construct a sample set; 70% of the sample set is a training set, and 30% is a validation set; D2, feature selection: using variance analysis and random forest feature importance sorting, the feature variables significantly related to heart failure occurrence are selected from the detected oxidized lipids to determine the model input features; D3, model training: based on the training set data, a classification model is constructed using the random forest algorithm, with the number of decision trees set to 100-200, the maximum tree depth set to 10-15, the minimum sample partition number set to 2, and the minimum sample leaf node number set to 1; D4, model optimization and verification: the model is optimized using 5-fold cross-validation, and the model performance is verified using the validation set data. ​

Images