Metabolic syndrome phlegm syndrome diagnosis model construction method based on lipid metabolism characteristics

Abstract

The invention relates to a construction method of a metabolic syndrome phlegm syndrome lipid metabolism characteristic diagnosis model. The construction method comprises the following steps: S1, acquiring a serum sample; s2, carrying out lipid metabolite analysis through a full-quantitative lipidomics formula; s3, carrying out primary screening on differential metabolites; s4, optimizing the characteristic indexes by using a random forest algorithm and stepwise regression; and S5, constructing a differential metabolite diagnosis model through Logistic regression analysis. By analyzing differential lipid metabolites of patients with MetS phlegm syndromes and non-phlegm syndromes, the invention reveals that abnormal accumulation of lipid and lipid metabolites may be the core pathological parenchyma of MetS phlegm syndromes. By integrating lipid metabonomics data, using a random forest algorithm, stepwise regression and other methods to screen feature difference lipid metabolites and construct a MetS phlegm syndrome specific diagnosis model, a novel combined biomarker and evidence-based basis are provided for early diagnosis of MetS phlegm syndromes, and a scientific basis can also be provided for objective diagnosis of traditional Chinese medicine phlegm syndromes.

Description

Technical Field

[0001] This invention relates to the field of metabolic syndrome risk prediction technology, and in particular to a method for constructing a diagnostic model of lipid metabolism characteristics in sputum syndrome of metabolic syndrome. Background Technology

[0002] Metabolic syndrome (MetS) is a clinical syndrome characterized by a cluster of metabolic abnormalities, including central obesity, insulin resistance, hypertension, dyslipidemia, and glucose metabolism disorders. It is a common pathological basis for cardiovascular disease and chronic diseases such as type 2 diabetes. Studies have shown that patients with MetS have a 2-3 times higher risk of developing cardiovascular disease than the general population, and MetS has been proven to be an independent risk factor for atherosclerosis.

[0003] Clinical studies have shown that TCM syndrome elements in patients with MetS phlegm syndrome are significantly positively correlated with lipid metabolism disorder markers, including body mass index (BMI), waist circumference, and low-density lipoprotein cholesterol (LDL-C) (all P<0.05).

[0004] Specifically, traditional Chinese medicine theory holds that "phlegm and turbidity accumulation" is the core pathogenesis of MetS. Abnormal lipid metabolism, such as elevated triglycerides and imbalance of low-density lipoprotein cholesterol subtypes, is regarded as the microscopic material basis of "phlegm and blood stasis." Meanwhile, visceral fat accumulation and lipotoxic microenvironment are highly consistent with the syndrome characteristics of "phlegm and dampness trapping the spleen" and "phlegm and blood stasis obstructing the collaterals."

[0005] However, the current clinical identification of MetS phlegm syndrome mainly relies on traditional four diagnostic methods, such as thick and greasy tongue coating and slippery pulse. Although these can reflect the overall metabolic state, they are difficult to accurately quantify the spatiotemporal relationship between phlegm syndrome and lipid metabolism disorder. Summary of the Invention

[0006] The purpose of this invention is to provide a method for constructing a diagnostic model of lipid metabolism characteristics in sputum syndrome of metabolic syndrome, which provides novel combined biomarkers and evidence-based basis for the early diagnosis of sputum syndrome of Metabolic Syndrome (MetS).

[0007] This invention is achieved through the following technical solution: a method for constructing a multi-dimensional perspective MS sputum syndrome nomogram prediction model, which includes the following steps:

[0008] Step 1: Sample Acquisition: Obtain serum from MetS patients, who are divided into a phlegm syndrome group and a non-phlegm syndrome group;

[0009] Step 2: Lipid metabolite analysis using fully quantitative lipidomics: lipid metabolites were isolated from the serum obtained in Step 1, and the isolated lipid metabolites were analyzed and quantified. Then, general data analysis and noise reduction preprocessing were performed.

[0010] Step 3: Initial screening of differential metabolites: Based on the data from Step 2, principal component analysis and orthogonal partial least squares discriminant analysis are used to model and analyze the differential metabolites.

[0011] Step 4, optimize feature indicators: For the differential metabolites screened in Step 3, use the random forest algorithm to screen differential metabolite indicators, and then use stepwise regression to further screen the differential metabolite feature indicators.

[0012] Step 5, Construct a diagnostic model: Construct a diagnostic model from the differential metabolites obtained in Step 4 using Logistic regression analysis.

[0013] Compared with previous technologies, the beneficial effects of the present invention are as follows:

[0014] This invention analyzes the differential lipid metabolites between patients with and without MetS (Methuselah Syndrome) phlegm syndrome, revealing that abnormal accumulation of lipids and lipid metabolites may be the core pathological essence of MetS phlegm syndrome. By integrating lipid metabolomics data and employing methods such as random forest algorithm and stepwise regression to screen for characteristic differential lipid metabolites and construct a MetS phlegm syndrome-specific diagnostic model, this invention can provide novel combined biomarkers and evidence-based support for the early diagnosis of MetS phlegm syndrome, and also provide a scientific basis for the objective diagnosis of phlegm syndrome in traditional Chinese medicine. Attached Figure Description

[0015] Figure 1 A schematic diagram showing the types and proportions of lipid metabolites;

[0016] Figure 2 This is a graph showing the principal component analysis results of lipid metabolites in the MetS phlegm syndrome group compared to those in the non-phlegm syndrome group.

[0017] Figure 3 The scatter plot of the OPLS-DA model for the MetS phlegm syndrome group versus the non-phlegm syndrome group is obtained;

[0018] Figure 4 The permutation test results of the OPLS-DA model for the MetS phlegm syndrome group versus the non-phlegm syndrome group are shown in a scatter plot.

[0019] Figure 5 A bar chart showing the permutation test results of the OPLS-DA model for the MetS phlegm syndrome group versus the non-phlegm syndrome group;

[0020] Figure 6 A column chart showing the relative enrichment of lipids in the MetS phlegm syndrome group;

[0021] Figure 7 Box plot showing the abundance of lipid metabolites;

[0022] Figure 8 Box plot showing the significant up- and down-regulation of lipid metabolites in the non-phlegm syndrome group compared to the phlegm syndrome group by MetS;

[0023] Figure 9 Volcano plot showing significant up- and down-regulation of lipid metabolites in the non-phlegm syndrome group compared to the phlegm syndrome group in MetS;

[0024] Figure 10 Matchstick plot showing the differences in lipid metabolites between the non-phlegm syndrome group and the phlegm syndrome group in MetS.

[0025] Figure 11 Importance ranking graph for feature selection

[0026] Figure 12 ROC plot for lipid differential metabolite characteristic indicators;

[0027] Figure 13 Decision curves for characteristic indicators of lipid differential metabolites;

[0028] Figure 14 Calibration curves for characteristic indicators of lipid differential metabolites. Detailed Implementation

[0029] The present invention will now be described in detail with reference to the accompanying drawings:

[0030] like Figure 1-14 The following is a method for constructing a diagnostic model of lipid metabolism characteristics in sputum syndrome of metabolic syndrome, which includes the following steps:

[0031] Step 1: Sample Acquisition: Obtain serum from MetS patients, who are divided into a phlegm syndrome group and a non-phlegm syndrome group;

[0032] Step 2: Lipid metabolite analysis using fully quantitative lipidomics: lipid metabolites were isolated from the serum obtained in Step 1, and the isolated lipid metabolites were analyzed and quantified. Then, general data analysis and noise reduction preprocessing were performed.

[0033] Step 3: Initial screening of differential metabolites: Based on the data from Step 2, principal component analysis and orthogonal partial least squares discriminant analysis are used to model and analyze the differential metabolites.

[0034] Step 4, optimize feature indicators: For the differential metabolites screened in Step 3, use the random forest algorithm to screen differential metabolite indicators, and then use stepwise regression to further screen the differential metabolite feature indicators.

[0035] Step 5, Construct a diagnostic model: Construct a diagnostic model from the differential metabolites obtained in Step 4 using Logistic regression analysis.

[0036] To build the model, the following research subjects were selected.

[0037] From March 2023 to February 2024, 96 patients with metabolic syndrome who visited the inpatient, outpatient, and physical examination centers of Fujian Provincial People's Hospital were included in this study. Among them, 54 patients (31 males and 23 females) were in the phlegm syndrome group, and 42 patients (24 males and 18 females) were in the non-phlegm syndrome group. All participants signed informed consent forms. This study was approved by the Ethics Committee of the Affiliated People's Hospital of Fujian University of Traditional Chinese Medicine (ethics approval number: 2023-013-01).

[0038] The criteria for determining MetS patients and sputum syndrome in step 1 are as follows:

[0039] Western medicine diagnostic criteria: Based on the diagnostic criteria for Metabolic Syndrome (MS) in the 2013 edition of the "Guidelines for the Prevention and Treatment of Type 2 Diabetes in China", patients with three or more of the following criteria can be diagnosed as Metabolic Syndrome (MetS): ① Abdominal obesity (waist circumference ≥90cm for men and ≥85cm for women); ② Hyperglycemia: fasting blood glucose ≥6.1mmol / L or 2-hour postprandial blood glucose ≥7.8mmol / L and / or diagnosed with and treated diabetes; ③ Hypertension: blood pressure ≥130 / 85mmHg and / or diagnosed with and treated hypertension; ④ High fasting triglycerides: triglycerides ≥1.70 mmol / L; ⑤ Low fasting high-density lipoprotein cholesterol (HDL-C) <1.04 mmol / L.

[0040] Traditional Chinese Medicine (TCM) Symptoms: A standardized four-diagnostic-inspection information collection scale was used. Qualified TCM professionals with rigorous training collected symptom and sign information through observation, auscultation, inquiry, and palpation. Symptoms and signs were categorized as absent, mild, moderate, and severe, scored as 0, 1, 2, and 3 points respectively. For symptoms and signs difficult to grade, they were categorized as absent and present, scored as 0 and 1 points respectively. The contribution score of symptoms and signs collected from the four-diagnostic-inspection information collection scale to the syndrome element was scored, with a threshold of 100. A syndrome element score ≥100 was considered diagnostic. In this study, a phlegm syndrome score ≥100 was diagnosed as a phlegm syndrome, and <100 was considered a non-phlegm syndrome.

[0041] Physicochemical indicators include, but are not limited to, gender, age, weight, height, heart rate, systolic blood pressure, and diastolic blood pressure.

[0042] The inclusion and exclusion criteria for the study subjects are as follows:

[0043] Inclusion criteria: ① Meeting the above diagnostic criteria; ② No use of antibiotics or probiotics within the past month, and no history of gastrointestinal diseases or hepatobiliary tract infections; ③ Informed consent from the patient; ④ Age 18–75 years. Exclusion criteria: ① Type 1 diabetes, gestational diabetes, secondary hyperlipidemia, secondary obesity, secondary hypertension; ② Accompanied by severe heart, liver, kidney, or other organ diseases; ③ Patients with mental illness.

[0044] Information and sample collection mainly includes the following information collection.

[0045] Information collection for the four diagnostic methods: Referring to the "Differential Significance of 600 Common Symptoms," and combining clinical research and expert opinions, a standardized information collection form for the four diagnostic methods was developed. Two trained TCM practitioners collected and recorded the information according to the standardized methods of traditional Chinese medicine. The "TCM Health Status Identification System" of the TCM Syndrome Research Base of Fujian University of Traditional Chinese Medicine was used to input the information into the system and output the relevant syndrome elements and scores.

[0046] General information collection: Fasting blood samples were collected from the enrolled patients in the morning to collect general information and main physical and chemical indicators of MS patients, including gender, age, weight, height, heart rate, systolic blood pressure, diastolic blood pressure, etc.

[0047] The general data collection and comparison are shown in Table 3. The baseline characteristics of the two groups of patients were analyzed by independent samples t-test. The results showed that the age difference between the two groups was statistically significant (p < 0.05), while the differences in other indicators between the groups were not statistically significant (all p > 0.05).

[0048] Table 3. Distribution of general information of patients with and without phlegm syndrome in MetS.

[0049]

[0050] Note: 1. Categorical data are represented by the difference in frequency distribution between patients with and without phlegm symptoms according to MetS; 2. Quantitative data are expressed as mean ± standard deviation; 3. * indicates statistically significant difference, p<0.05.

[0051] The serum was obtained by collecting fasting venous blood from participants in the early morning. The blood was collected in blood collection tubes containing EDTA, mixed, centrifuged, and then the supernatant was collected and frozen for later use.

[0052] The specific details are as follows: Lipid data collection (serum collection): Participants uniformly underwent fasting venous blood collection in the morning. After collecting blood with blood collection tubes containing EDTA, the blood was immediately gently inverted and mixed. The blood was centrifuged at 3000 rpm for 10 min at 4°C. 200 μL of the supernatant was collected and placed in a suitable numbered 2 mL centrifuge tube. The plasma was then stored frozen at -80°C until the time of testing.

[0053] Step 2: Perform lipid metabolite analysis using a fully quantitative lipidomics method:

[0054] Lipid metabolites were extracted from serum and separated using ultra-high performance liquid chromatography (UHPLC). Phase A consisted of a solution of 40% water and 60% acetonitrile containing 10 mmol / L ammonium acetate; Phase B consisted of a solution of 10% acetonitrile and 90% isopropanol containing 10 mmol / L ammonium acetate.

[0055] The specific method is as follows: First, metabolite extraction is performed on the serum using analytical techniques. After the above sample passes quality control, 40 μL of sample is taken, and 160 μL of water and 480 μL of extraction buffer (MTBE:MeOH = 5:1, containing internal standard) are added. The mixture is vortexed for 60 seconds and sonicated in an ice-water bath for 10 minutes. The sample is then centrifuged at 3000 rpm for 15 minutes at 4°C, and 250 μL of supernatant is collected. Then, 250 μL of MTBE is added to the remaining sample, and the above operation is repeated, collecting 250 μL of supernatant. The two supernatants are combined and vacuum dried at 37°C. 100 μL of solution is added to the dried metabolites.

[0056] Reconstitute (DCM:MeOH:H2O=60:30:4.5), vortex for 30s, sonicate in an ice-water bath for 10min, and centrifuge at 4℃ and 12000rpm for 15min; take 35μL of the supernatant into a sample vial for instrumental analysis; take another 10μL of the supernatant from all samples and mix them to form a QC sample for instrumental analysis.

[0057] Subsequent analysis was performed using an ACQUITY Premier ultra-high performance liquid chromatograph, with the target compounds (i.e., 649 metabolites) separated by liquid chromatography column separation. Phase A consisted of a 40% water and 60% acetonitrile solution containing 10 mmol / L ammonium acetate; Phase B consisted of a 10% acetonitrile and 90% isopropanol solution containing 10 mmol / L ammonium acetate. The flow rate was set at 0.3 mL / min, column temperature at 45 °C, sample tray temperature at 10 °C, and injection volume at 2 μL.

[0058] The above experimental reagents and instruments are shown in Table 1.

[0059] Table 1 List of Experimental Reagents

[0060]

[0061] The list of experimental instruments is shown in Table 2.

[0062]

[0063]

[0064] The isolated lipid metabolites were analyzed and quantified using SCIEX Analyst Work Station Software (Version 1.6.3) and DATA DRIVEN FLOW (Version 1.0.1). The absolute content of each lipid relative to the internal standard (IS) was calculated based on the relationship between the peak area and actual concentration of the same type of lipid internal standard (IS).

[0065] Then, general data analysis was performed: SPSS 27.0 software was used for statistical analysis, and quantitative data were analyzed using... The chi-square test was used for comparing count data. If the data followed a normal distribution, the independent samples t-test was used for comparisons between groups; if the data did not follow a normal distribution, the rank-sum test was used for comparisons between groups. Two-tailed tests were used in this study, with a significance level set at 0.05. A p-value less than 0.05 was considered statistically significant.

[0066] Finally, the raw lipid metabolite data were preprocessed, with individual peaks filtered to remove noise. Offset values ​​were filtered based on the interquartile range. Peak areas were retained only in groups with no more than 50% null values ​​or across all groups. Missing value recoding was performed on the raw data. Numerical simulation methods were used to impute the missing values ​​by multiplying the minimum value by a random number between 0.1 and 0.5.

[0067] Analysis of lipid metabolite composition obtained from fully quantitative lipidomics

[0068] The MetS phlegm syndrome group consisted of 54 cases, and the MetS non-phlegm syndrome group consisted of 42 cases (a total of 96 cases). A total of 649 metabolites were obtained after quantitative lipidomics analysis. These metabolites were mainly distributed among 22 categories, including TAG, SM, PI, PG, PE, PC, LPG, LPE, LPC, LCER, HCE, FFA, DCER, DAG, CER, CE, and BMP. Among these, TAG accounted for 59.78%, PE for 9.09%, PC for 7.86%, DAG for 6.32%, and the remaining metabolites accounted for 16.95%. (See...) Figure 1 ).

[0069] In step 3, differential metabolites are initially screened. Based on the data from step 2, principal component analysis and orthogonal partial least squares discriminant analysis are used to model and analyze the differential metabolites.

[0070] Principal component analysis and orthogonal partial least squares-discriminant analysis (OPLS-DA) modeling analysis were performed on the data using SIMCA software (V18.0.1, Sartorius Stedim Data Analytics AB, Umea, Sweden). Differential metabolites were screened based on a t-test p-value less than 0.05 and a variable importance in the projection (VIP) greater than 1 for the first principal component of the OPLS-DA model.

[0071] Principal component analysis (PCA) of lipid metabolites from the MetS phlegm syndrome group and the non-phlegm syndrome group clarified the overall metabolic differences between the two groups and the magnitude of differences within each group. PCA results showed that the phlegm syndrome group had a wide data distribution and significant internal variability; the non-phlegm syndrome group had a concentrated data distribution and high internal similarity. While there was some overlap between the two groups, there was a certain tendency for separation along the PC2 direction. Figure 2 As can be seen in the figure (PS = MetS for the phlegm syndrome group; non_PS = MetS for the non-phlegm syndrome group), the contribution rates of principal components PC1 and PC2 are 30.9% and 11%, respectively, indicating that lipid metabolism can be distinguished between the phlegm syndrome group and the non-phlegm syndrome group based on the first principal component. The PCA measurement results are highly reliable and can be used for subsequent analysis of differential metabolites.

[0072] like Figure 3 As shown: Based on the features obtained from the samples, the first principal component was analyzed by OPLS-DA modeling. The horizontal axis t[1]P in the figure shows the differences between sample groups, and the vertical axis t[1]O shows the differences within sample groups. The differences between the phlegm syndrome group and the non-phlegm syndrome group were large, and the repeatability within the groups was good, showing good discrimination.

[0073] like Figure 4 As shown: By randomly permuting the order of the categorical variable Y, a 200-time OPLS-DA model was constructed to obtain the R-values ​​for the MetS phlegm syndrome group and the non-phlegm syndrome group. 2 Y(cum) = (0, 0.36), Q 2 (cum) = (0, -0.2). R 2 The value Q indicates the model's explanatory power for Y. 2 The value reflects predictive power. The dashed line represents R. 2 Y and Q 2 The regression line is used to evaluate the effectiveness of the model.

[0074] like Figure 5 As shown: MetS phlegm syndrome group vs. non-phlegm syndrome group Q 2 =0.002, P>0.05(17 / 200), indicating that in the permutation test, 17 randomized models had better predictive power than the original model; R 2 Y = 0.338, P > 0.05 (110 / 200) indicates that in the permutation test, 110 randomized models have better explanatory power than the original model.

[0075] like Figure 6As shown: Observations revealed significant lipid metabolism heterogeneity in the serum of patients with MetS phlegm syndrome. The relative percentage difference for TAG was very high, indicating high expression of TAG in the phlegm syndrome group. DAG, PE, and PC also showed some relative differences, but these were far lower than those for TAG. The relative percentage differences for most other metabolites were low. Figure 7 As can be seen, the relative abundance of DAG, PC, PE, SM, and TAG in the MetS phlegm syndrome group was significantly different from that in the non-phlegm syndrome group.

[0076] like Figure 8 As shown in Table 4: Following the above analysis, and combining the results of univariate and multivariate statistical analysis, differentially expressed metabolites with p < 0.05 and VIP > 1 were screened using the t-test and OPLS-DA model. The results revealed 22 differentially expressed metabolite molecules in the non-phlegm syndrome group compared to the phlegm syndrome group, of which 20 were significantly downregulated and 2 were significantly upregulated. These metabolic changes may be closely related to the pathophysiological mechanisms of phlegm syndrome.

[0077] Table 4. Statistical table of differences in lipid metabolites between the non-phlegm syndrome group and the phlegm syndrome group in MetS.

[0078]

[0079] Screening results for differentially expressed metabolites were obtained using a volcano plot. Figure 9 Presented. Figure 9 Each dot in the graph represents a metabolite, and the size of the dot reflects the VIP value of the OPLS-DA model. Red and blue represent significantly upregulated and downregulated metabolites, respectively, while metabolites with no significant difference are shown in gray. For example... Figure 9 Compared with the phlegm syndrome group, the top 10 downregulated metabolites in the non-phlegm syndrome group were: TAG(48:5)_FA18:3, TAG(46:3)_FA18:3, TAG(48:3)_FA18:3, TAG(52:1)_FA16:1, TAG(52:2)_FA16:1, TAG(50:4)_FA20:3, TAG(50:4)_FA18:1, TAG(46:3)_FA14:0, TAG(48:4)_FA18:3, and PE(16:0 / 20:4), and the two upregulated metabolites were: SM(14:0) and PC(18:0 / 20:1).

[0080] By calculating the quantitative values ​​of the differentially regulated metabolites, the ratios were determined and converted to base 2 logarithms. The top 10 changes in both upregulation and downregulation were then presented for comparison. Compared to the MetS phlegm syndrome group, the fold changes in downregulated metabolites in the MetS non-phlegm syndrome group were as follows: TAG(48:5)_FA18:3,

[0081] TAG(46:3)_FA18:3, TAG(48:3)_FA18:3, TAG(46:3)_FA14:0, TAG(52:1)_FA16:1, TAG(50:3)_FA20:3, TAG(48:4)_FA18:3, TAG(50:4)_FA20:3, TAG(52:2)_FA16:1, TAG(50:4)_FA18:1; The fold change of the upregulated metabolites is as follows: SM(14:0),

[0082] PC (18:0 / 20:1) (as shown) Figure 10 (As shown).

[0083] Step 4, optimize feature indicators: For the differential metabolites screened in Step 3, use the random forest algorithm to screen differential metabolite indicators, and then use stepwise regression to further screen the differential metabolite feature indicators.

[0084] Specifically, in step 4, the random forest algorithm is used, and based on the Mean Decrease Accuracy and statistical efficiency, the top-ranked differential metabolite characteristic indicators are selected; then stepwise regression is used to perform multiple rounds of screening on the differential metabolite characteristic indicators.

[0085] The analysis was performed using R4.4.3 software. The random forest algorithm was used to screen differentially expressed metabolite indicators. Based on the mean decrease accuracy and statistical efficiency, the top 8 differentially expressed metabolite indicators were selected: TAG(46:3)_FA18:3, TAG(48:3)_FA18:3, ...

[0086] TAG(48:5)_FA18:3, PE(16:0 / 16:1), TAG(50:4)_FA20:3, TAG(50:3)_FA20:3, TAG(48:4)_FA18:3, TAG(46:3)_FA14:0 (see Figure 11 ).

[0087] The eight selected feature indicators were further refined using stepwise regression. After six rounds of refinement, two indicators remained. TAG(48:5)_FA18:3 and TAG(48:3)_FA18:3 had a significant impact on group classification (see Table 6). Subsequently, the variables TAG(50:3)_FA20:3 and TAG(50:3)_FA20:3 were gradually removed.

[0088] For TAG(46:3)_FA18:3, TAG(46:3)_FA14:0, TAG(50:4)_FA20:3, and TAG(48:4)_FA18:3, the AIC decreased from 127.95 to 119.03, indicating a significant improvement in model simplification (see Table 5). For TAG(48:5)_FA18:3...

[0089] (p<0.05) may serve as a key metabolic biomarker for distinguishing between phlegm syndrome and non-phlegm syndrome. Although TAG(48:3)_FA18:3 did not reach statistical significance, it was retained for use in constructing the diagnostic model due to its contribution to the overall model optimization (AIC) and potential biological significance.

[0090] Table 5. Variable Selection and AIC Changes

[0091]

[0092] Table 6. Significance of variables in the final model

[0093]

[0094] Note: AIC = AIC of the current step - AIC of the previous step, negative values ​​indicate model optimization; significance markers: ***P<0.001, **P<0.01, *P<0.05; the final model AIC = 119.03, a decrease of 8.92 compared to the initial model, indicating that variable selection significantly improved model fit.

[0095] Step 5, Construct a diagnostic model: Construct a diagnostic model from the differential metabolites obtained in Step 4 using Logistic regression analysis.

[0096] like Figure 12 As shown, a Logistic regression diagnostic model was constructed using the two selected feature indicators. The area under the curve was 0.72, and the 95% confidence interval was 0.619–0.821. This indicates that the feature indicators can effectively distinguish between phlegm-related and non-phlegm-related symptoms in MetS.

[0097] In addition, the diagnostic model is validated in step 6: Based on the diagnostic model in step 5, the area under receiver operating characteristics (ROC) curve (AUC) is used to evaluate the discriminative performance of the feature indicators; Bootstrap resampling is applied to verify the model's fit and calibration; and decision curve analysis (DCA) is used to evaluate the model's practical value in clinical practice.

[0098] Specifically, the Bootstrap method was used for 1000 internal sampling validations. The error between the model's predicted probability and the actual value was 0.043, indicating that the MetS prediction probability for phlegm syndrome has good consistency with the actual probability (e.g., Figure 13(As shown). Decision curve analysis further shows that the net return of the predictive model is higher than the other two extreme curves, indicating that the model has practicality and value in clinical practice. Figure 14 (As shown).

[0099] In summary, the basic research of this invention, through lipidomics analysis, found that the serum TAG content of patients with MetS phlegm syndrome was significantly higher than that of the non-phlegm syndrome group. At the same time, some DAG, PE and various TAG molecules with specific structures, such as TAG(48:5)_FA18:3 and TAG(46:3)_FA18:3, were significantly upregulated, suggesting that lipid metabolism disorder may be the core pathological mechanism of phlegm syndrome formation.

[0100] By integrating random forest algorithm and stepwise regression analysis, the system screened and validated two lipid metabolites, TAG(48:5)_FA18:3 and TAG(48:3)_FA18:3, as specific biomarkers for phlegm syndrome. Among them, TAG(48:5)_FA18:3 showed abnormal accumulation in the phlegm syndrome group, suggesting that the imbalance of TAG molecular metabolism may be the pathological basis for the formation of MetS phlegm syndrome. The combination of biomarkers conforms to the multi-target diagnostic characteristics of the holistic view of traditional Chinese medicine and can also significantly improve the differential efficacy of phlegm syndrome. Its AUC reached 0.72, with a 95% confidence interval of 0.619-0.821, indicating good diagnostic performance. Bootstrap resampling validation and DCA further confirmed that the model's classification error rate was 0.043. Within the clinical decision range with a threshold probability >10%, its net benefit was higher than the other two extreme curves, indicating that the model has good clinical translation potential.

[0101] It should be noted that principal component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA) modeling are common techniques in the industry. PCA is a statistical method that transforms a set of potentially correlated variables into linearly uncorrelated variables (i.e., principal components) through orthogonal transformation. PCA can reveal the internal structure of data, thus better interpreting data variables. Metabolomics data can be considered a multivariate dataset that can be displayed in a high-dimensional data space coordinate system. PCA can then provide a relatively low-dimensional image (two-dimensional or three-dimensional), which is essentially a "projection" of the original object onto the points containing the most information, effectively reducing the dimensionality of the data by utilizing a small number of principal components. Using SIMCA software (V18.0.1, Sartorius Stedim Data Analytics AB, Umea, Sweden), the data was subjected to logarithmic (LOG) transformation and centering (CTR) formatting, followed by automated modeling analysis.

[0102] Metabolomics data are characterized by high dimensionality (many types of metabolites detected) and small sample size (relatively small sample size). These variables include both differentially expressed variables related to categorical variables and a large number of indifferent variables that may be correlated with each other. This means that if we use PCA or PLS models for analysis, the differentially expressed variables will be scattered across more principal components due to the influence of correlated variables, hindering better visualization and subsequent analysis. Therefore, we employ orthogonal partial least squares-discriminant analysis (OPLS-DA) to analyze the results. Through OPLS-DA analysis, we can filter out orthogonal variables in the metabolite data that are not correlated with categorical variables, and analyze non-orthogonal and orthogonal variables separately, thereby obtaining more reliable information on the inter-group differences in metabolites and the correlation between experimental groups. The data was processed using SIMCA software (V18.0.1, Sartorius Stedim Data Analytics AB, Umea, Sweden) with logarithmic transformation and UV formatting. First, OPLS-DA modeling analysis was performed on the first principal component, and the model quality was tested using 7-fold cross-validation. Then, the RY (interpretability of the model for the categorical variable Y) and Q (predictability of the model) obtained after cross-validation were used to evaluate model effectiveness. Finally, a permutation test was conducted, randomly changing the order of the categorical variable Y multiple times to obtain different random Q values, further verifying model effectiveness. Therefore, this application will not elaborate further.

[0103] In summary, this invention, by analyzing the differential lipid metabolites between patients with and without MetS (MetS phlegm syndrome), reveals that abnormal accumulation of lipids and lipid metabolites may be the core pathological essence of MetS phlegm syndrome. By integrating lipid metabolomics data and employing methods such as random forest algorithm and stepwise regression to screen for characteristic differential lipid metabolites and construct a MetS phlegm syndrome-specific diagnostic model, this invention can provide novel combined biomarkers and evidence-based support for the early diagnosis of MetS phlegm syndrome, and also provide a scientific basis for the objective diagnosis of phlegm syndrome in traditional Chinese medicine.

[0104] By employing fully quantitative metabolomics to screen for differentially expressed lipid metabolites between patients with and without MetS (MetS) sputum syndrome, and using multivariate statistical methods to construct a diagnostic model for MetS sputum syndrome, this study aims to identify characteristic diagnostic indicators and provide novel combined biomarkers for the early molecular diagnosis of MetS sputum syndrome. This model can be applied to the early molecular diagnosis of MetS sputum syndrome.

[0105] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a diagnostic model of lipid metabolism characteristics in sputum syndrome of metabolic syndrome, characterized by: It includes the following steps: Step 1: Sample Acquisition: Obtain serum from MetS patients, who are divided into a phlegm syndrome group and a non-phlegm syndrome group; Step 2: Lipid metabolite analysis using fully quantitative lipidomics: lipid metabolites were isolated from the serum obtained in Step 1, and the isolated lipid metabolites were analyzed and quantified. Then, general data analysis and noise reduction preprocessing were performed. Step 3: Initial screening of differential metabolites: Based on the data from Step 2, principal component analysis and orthogonal partial least squares discriminant analysis are used to model and analyze the differential metabolites. Step 4, optimize feature metrics: For the differential metabolites screened in step 3, the random forest algorithm was used to screen differential metabolite indicators, and then stepwise regression was used to further screen the differential metabolite characteristic indicators. Step 5, Construct a diagnostic model: Construct a diagnostic model from the differentially metabolites identified in Step 4 using logistic regression analysis.

2. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: The criteria for determining MetS patients and sputum syndrome in step 1 are as follows: Western medicine diagnostic criteria: A patient can be diagnosed with MetS if they meet three or more of the following criteria: ① Abdominal obesity (waist circumference ≥90cm for men, ≥85cm for women); ② Hyperglycemia: fasting blood glucose ≥6.1mmol / L or 2-hour postprandial blood glucose ≥7.8mmol / L and / or have been diagnosed with and treated for diabetes; ③ Hypertension: blood pressure ≥130 / 85mmHg and / or have been diagnosed with and treated for hypertension; ④ High fasting triglycerides: triglycerides ≥1.70mmol / L; ⑤ Low fasting high-density lipoprotein cholesterol <1.04mmol / L. Traditional Chinese Medicine (TCM) Symptoms: A standardized four-diagnostic-inspection information collection scale was used. Qualified TCM professionals with rigorous training collected symptom and sign information through observation, auscultation, inquiry, and palpation. Symptoms and signs were categorized as absent, mild, moderate, and severe, scored as 0, 1, 2, and 3 points respectively. For symptoms and signs difficult to grade, they were categorized as absent and present, scored as 0 and 1 points respectively. The contribution score of symptoms and signs collected from the four-diagnostic-inspection information collection scale to the syndrome element was scored, with a threshold of 100. A syndrome element score ≥100 was considered diagnostic. In this study, a phlegm syndrome score ≥100 was diagnosed as a phlegm syndrome, and <100 was considered a non-phlegm syndrome. Physicochemical indicators include, but are not limited to, gender, age, weight, height, heart rate, systolic blood pressure, and diastolic blood pressure.

3. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: In step 1, the serum was obtained by having participants collect fasting venous blood in the morning. The blood was collected using blood collection tubes containing EDTA, mixed, centrifuged, and then the supernatant was collected and frozen for later use.

4. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: In step 2, lipid metabolites were extracted from serum and separated using ultra-high performance liquid chromatography (UHPLC). Phase A consisted of a solution of 40% water and 60% acetonitrile containing 10 mmol / L ammonium acetate; Phase B consisted of a solution of 10% acetonitrile and 90% isopropanol containing 10 mmol / L ammonium acetate.

5. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: In step 2, the separation of lipid metabolites was analyzed and quantified using CIEX Analyst Work Station Software and DATA DRIVEN FLOW. The absolute content of each lipid relative to the IS was calculated based on the relationship between the peak area and actual concentration of the same type of lipid.

6. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: Step 2, general data analysis, includes statistical analysis using SPSS 27.0 software, with continuous data presented as follows: The chi-square test was used for comparing count data. If the data followed a normal distribution, the independent samples t-test was used for comparisons between groups; if the data did not follow a normal distribution, the rank-sum test was used for comparisons between groups. Two-tailed tests were used in this study, with a significance level set at 0.05; a p-value less than 0.05 was considered statistically significant.

7. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: In step 2, the noise reduction preprocessing is as follows: filtering individual peaks to remove noise; filtering outliers based on interquartile range; retaining peak area data with no more than 50% null values ​​in a single group or no more than 50% null values ​​in all groups; simulating missing values ​​in the original data; and using the numerical simulation method of imputing the minimum value by multiplying it by a random number between (0.1 and 0.5).

8. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: In step 3, after noise reduction, principal component analysis and orthogonal partial least squares-discriminant analysis were performed on the data using SIMCA software. Differential metabolites were screened based on the p-value of less than 0.05 in the t-test and the variable projection importance of the first principal component of the orthogonal partial least squares-discriminant analysis model being greater than 1.

9. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: In step 4, the random forest algorithm is used, and based on the Mean Decrease Accuracy and statistical efficiency, the top-ranked differential metabolite characteristic indicators are selected; then stepwise regression is used to perform multiple rounds of screening on the differential metabolite characteristic indicators.

10. The method for constructing a diagnostic model for lipid metabolism characteristics of sputum syndrome according to claim 1, characterized in that: Also includes Step 6: Diagnostic Model Validation: Based on the diagnostic model in Step 5, the area under receiver operating characteristics (ROC) curve (AUC) is used to evaluate the discriminative performance of the feature indicators; Bootstrap resampling is applied to validate the model's fit and calibration; and decision curve analysis (DCA) is used to assess the model's practical value in clinical practice.

Images