An index combination, method and system for predicting obesity-related metabolic syndrome

Abstract

The present application relates to an index combination, method and system for predicting obesity-related metabolic syndrome, which comprises an index combination for predicting obesity-related metabolic syndrome, the index combination comprising the following indexes: NKG2C + Percentage of mitochondrial low membrane potential in NK cells (NKG2C + NK.MMP low Percentage), NKG2D + Percentage of mitochondrial low membrane potential in NKT cells (NKG2D + NKT.MMP low Percentage), NKG2C + Percentage of mitochondrial low membrane potential in NKT cells (NKG2C + NKT.MMP low Percentage), percentage of mitochondrial low membrane potential in monocytes (Mono.MMP low Percentage), NKG2A + Mitochondrial mass in lymphocytes (NKG2A + Lymph.MM), and mean platelet volume (MPV). The index combination obtained by screening has high biological interpretability and clinical relevance, and can effectively reflect obesity-related immune metabolic abnormalities.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical data mining, and in particular to an index combination, method and system for predicting obesity-related metabolic syndrome. BACKGROUND

[0002] Obesity and its related metabolic syndrome have become a major global public health problem, and its pathogenesis is complex, involving the interaction between metabolic disorders and immune system abnormalities. Innate immune cells, especially natural killer T (NKT) cells and natural killer (NK) cells, play a key role in maintaining metabolic homeostasis and regulating inflammatory response.

[0003] Recent studies have shown that there is a significant immune-metabolic dysfunction in obese individuals. For example, it has been reported that CD56 bright The increase in NKG2D activation frequency in NK cells is associated with obesity, suggesting that NK cell activation receptors may be involved in the pathological process of obesity. NKT cells, as a bridge connecting innate immunity and adaptive immunity, their surface activation receptors (such as NKG2D and NKG2C) expression and functional changes in obesity have attracted increasing attention. However, there are still gaps in current research: on the one hand, the specific role of NKT cells in obesity is complex and even contradictory, some studies have shown that their cytotoxic function is impaired, while there is evidence that they can participate in inflammation or maintain metabolic homeostasis by secreting cytokines; on the other hand, there is currently a lack of direct and systematic evidence to elucidate the relationship between the metabolic state of specific NKT cell subgroups (such as NKG2D + or NKG2C + NKT cells) and obesity, and there is no report on the application of such immune-metabolic indicators for disease risk prediction.

[0004] Mitochondria, as the energy factory and metabolic center of cells, their functional status (such as membrane potential) is a key indicator to assess the metabolic activity of cells. Dysfunction of lymphocyte mitochondria is closely related to many chronic inflammatory diseases. Therefore, in-depth study of the mitochondrial function characteristics of immune cells (especially NKT / NK cell subgroups) in obese populations and mining of their specific metabolic indicators related to obesity are of great significance for revealing the disease mechanism and developing new prediction tools. SUMMARY

[0005] The present application combines flow cytometry with multi-dimensional data analysis, aiming to break through the limitations of existing research, not only focusing on the expression of cell surface receptors (such as NKG2D, NKG2C), but also further deepening the internal metabolic state (mitochondrial membrane potential and mass) of its subpopulation, and integrating conventional clinical indicators (such as mean platelet volume MPV), and through advanced machine learning algorithms for high-throughput screening, ultimately screening a new set of index combinations that can accurately reflect obesity-related immune metabolic disorders, and establishing an efficient and objective risk prediction method.

[0006] In a first aspect, the present application provides an index combination for predicting obesity-related metabolic syndrome, the index combination comprising the following indexes: NKG2C + Percentage of low mitochondrial membrane potential in NK cells (NKG2C + NK.MMP low Percentage), NKG2D + Percentage of low mitochondrial membrane potential in NKT cells (NKG2D + NKT.MMP low Percentage), NKG2C + Percentage of low mitochondrial membrane potential in NKT cells (NKG2C + NKT.MMP low Percentage), percentage of low mitochondrial membrane potential in monocytes (Mono.MMP low Percentage), NKG2A + Mitochondrial mass in lymphocytes (NKG2A + Lymph.MM), and mean platelet volume (MPV).

[0007] In a second aspect, the present application provides a method for screening the index combination of the first aspect, the method comprising the following steps: (1) collecting a peripheral blood sample; (2) detecting the blood routine index, cell percentage index, mitochondrial mass index and mitochondrial low membrane potential percentage index of the sample of step (1); (3) performing correlation analysis and difference analysis on the indexes of step (2) to screen BMI-related indexes; (4) combining machine learning technology to analyze and screen the indexes of step (3) to obtain the index combination.

[0008] Optionally, the machine learning method comprises a random forest algorithm.

[0009] In a third aspect, the present application provides a method for predicting obesity-related metabolic syndrome, comprising the following steps:

[0010] a) obtaining the detection value of the index of the first aspect of the subject;

[0011] b) standardizing the detection value;

[0012] c) inputting the normalized data into a pre-trained prediction model, outputting a BMI-related parameter (BRP) value, and determining whether the subject has a risk of obesity-related metabolic syndrome according to the BRP value.

[0013] Optionally, the normalization processing adopts a min-max normalization method.

[0014] Optionally, the prediction model is a principal component analysis (PCA) model, wherein a first principal component (PC1) is used as the BRP value for risk prediction.

[0015] Optionally, the BRP value calculation formula is: BRP value = PC1 = 0.48*MPV - 0.39*Mono.MMP low %

[0016] -0.26*NKG2C + NK.MMP low %+0.36*NKG2C + NKT.MMP low %+0.20*NKG2D + NKT.MMP low %-0.40*NKG2A + Lymph.MM。

[0017] In a fourth aspect, the present application provides a prediction system for obesity-related metabolic syndrome, comprising:

[0018] an input module configured to receive detection values of the indexes according to the first aspect;

[0019] a preprocessing module configured to perform normalization processing on the detection values;

[0020] an analysis module configured to input the normalized data into a pre-trained PCA model and calculate a BRP value;

[0021] an output module configured to output a risk assessment result of obesity-related metabolic syndrome.

[0022] In a fifth aspect, the present application provides a use of a reagent for detecting the indexes according to the first aspect in a product for predicting obesity-related metabolic syndrome, or a use in preparing a kit for the use.

[0023] In summary, the present application includes at least one of the following beneficial technical effects:

[0024] 1. By multi-dimensional data mining and machine learning screening, it is first determined that the indexes including NKG2C + NK cells, NKG2D + NKT cells, NKG2C+ Percentage of mitochondrial low membrane potential of NKT cells, percentage of mitochondrial low membrane potential of monocytes, NKG2A + Lymphocyte mitochondrial mass, and 6 key indicators including mean platelet volume (MPV); the combination has high biological interpretation and clinical relevance, which can effectively reflect obesity-related immune metabolic abnormalities;

[0025] 2. The obesity-related parameter (BRP) value constructed based on principal component analysis (PCA) was significantly positively correlated with BMI (P<0.001), and the area under the ROC curve (AUC) reached 85.78%, indicating that the model has high prediction accuracy and discrimination ability, and can be used to identify high-risk individuals with abnormal immune cell metabolic expression in obese populations, assisting clinical early intervention and personalized health management;

[0026] 3. Through multiple steps of correlation analysis, differential analysis and random forest algorithm, 6 most representative indicators were selected from 117 initial indicators, which not only ensured the simplicity and interpretability of the model, but also improved its stability and generalization ability;

[0027] 4. This method reveals the close relationship between mitochondrial function status of NKT cells and their subsets and obesity, providing new data support and analysis tools for in-depth study of the role of immune metabolic regulation in obesity and related diseases;

[0028] 5. The indicators involved can be detected by conventional experimental methods such as flow cytometry, and the supporting reagents and detection systems are easy to standardize and industrialize, suitable for popularization and application in physical examination centers, hospitals and research institutions. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1A Mean Decrease Accuracy for Random Forest Analysis of BMI-related Mean;

[0030] Figure 1B Mean Decrease Gini for Random Forest Analysis of BMI-related Mean;

[0031] Figure 1C Distribution Cross-Validation Error for Random Forest Analysis;

[0032] Figure 2A Principal Component Composition Contribution Ratio Chart;

[0033] Figure 2B Paired Chart and PCA Double Chart Reflecting Distribution Differences Between Two Groups;

[0034] Figure 2C Loading Chart, Showing Parameter Composition and Contribution of Each Principal Component;

[0035] Figure 2D The PC1~PC5 of 118 research samples in two groups of receiver operating characteristic curve (ROC) analysis;

[0036] Figure 2E The significant difference of PC1 value between the two groups of 118 research samples;

[0037] Figure 2F The linear analysis graph of BMI and PC1 of 118 research samples;

[0038] Figure 3 The linear analysis graph of BMI and PC1 of 42 validation samples;

[0039] Figure 4 The PC1~PC5 of 42 validation samples in two groups of receiver operating characteristic curve (ROC) analysis. DETAILED DESCRIPTION

[0040] In order to have a clearer understanding of the technical features, objectives and beneficial effects of the present application, the technical solutions of the present application will be described in detail below in conjunction with the following specific examples and drawings of the specification, but it cannot be understood as limiting the scope of the implementation of the present application.

[0041] Example 1 Detection index data collection

[0042] 1.1 Data queue and detection: We collected the gender, BMI, age information of 118 physical examinees, and all the data were in normal distribution. The immune cell data (including: mononuclear cells, lymphocytes and their subgroups T cells, NK cells, NKT cells and γδT cells) and related data of their mitochondrial parameters (including the percentage of lymphocyte subgroups, mitochondrial median fluorescence index (Mito-MFI, also known as MM) and mitochondrial membrane potential reduction percentage (MMP low %) were obtained by flow cytometry, and combined with blood routine data (including mean platelet volume (mean platelet volume, MPV)) for analysis.

[0043] 1.2 Antibodies and indicators: Mitochondrial staining MitoDye-APC (pan-peptide biotechnology (Zhejiang) Co., Ltd.) was used, and the mitochondrial parameters Mito-MFI and MMP lowNKT combination flow cytometry analysis, antibodies include: CD3 FITC, CD45 PerCP-Cy5.5, CD4 PE-Cy7, CD8 APC-Cy7 (Pan-peptide Biotechnology (Zhejiang) Co., Ltd.) and TCRγ / δ BV421, NKG2C PE, NKG2A ECD, CD56 KO525, NKG2D BV610 (BioLegend, San Diego, USA).

[0044] 1.3 Sample collection: Collect 2-5 mL of venous blood with EDTA anticoagulant tube. Immediately mix the blood sample after collection to prevent clotting. Verify that the sample has no clots and hemolysis, and the barcode and basic information are accurate. Sample preparation scheme: a) Incubate 100 μL of peripheral blood sample containing EDTA anticoagulant with mixed antibodies at room temperature for 15 min, add 2 mL of hemolysin to destroy red blood cells. b) Centrifuge the sample at 300xg for 5 min. c) Discard the supernatant and resuspend the precipitate in 200 μL of PBS, transfer to a cuvette containing MitoDye, and incubate at 37°C for 30 min in the dark. d) Finally, transfer it to a flow tube, then count the labeled immune cells by NovoCyte. Use NovoExpress software (Agilent technology, USA) for final analysis and graphical output. The specific detection indicators are shown in Table 1.

[0045] Table 1

[0046] SEQ ID NO: Abbreviation Chinese Name of Index SEQ ID NO: Abbreviation Chinese Name of Index Age Age 59 NKT.MMP low %]]> Percentage of NKG2A-positive lymphocyte mitochondrial low membrane potential 1 WBC White blood cell count 60 NKG2A + L.MMP low %]]> Percentage of NKG2A-positive T cell mitochondrial low membrane potential 2 RBC Red blood cell count 61 NKG2A + T.MMP low %]]> Percentage of NKG2A-positive killer T cell mitochondrial low membrane potential 3 HGB Hemoglobin 62 NKG2A + Tc.MMP low %]]> Percentage of NKG2A-positive natural killer cell mitochondrial low membrane potential 4 HCT Hematocrit 63 NKG2A + NK.MMP low %]]> Percentage of NKG2A-positive natural killer T cell mitochondrial low membrane potential 5 MCV Mean corpuscular volume 64 NKG2A + NKT.MMP low %]]> Percentage of NKG2A-positive γδ T cell mitochondrial low membrane potential 6 MCH Mean corpuscular hemoglobin 65 NKG2A + γδT.MMP low %]]> Percentage of NKG2C-positive neutrophil mitochondrial low membrane potential 7 MCHC Mean corpuscular hemoglobin concentration 66 NKG2C + G.MMP low %]]> Percentage of NKG2C-positive lymphocyte mitochondrial low membrane potential 8 RDW Red blood cell distribution width 67 NKG2C + L.MMP low %]]> Percentage of NKG2C-positive T cell mitochondrial low membrane potential 9 PLT Platelet count 68 NKG2C + T.MMP low %]]> Percentage of NKG2C-positive natural killer cell mitochondrial low membrane potential 10 PCT Platelet hematocrit 69 NKG2C + NK.MMP low %]]> Percentage of NKG2C-positive natural killer T cell mitochondrial low membrane potential 11 MPV Mean platelet volume 70 NKG2C + NKT.MMP low %]]> Percentage of NKG2C-positive γδ T cell mitochondrial low membrane potential 12 PDW Platelet distribution width 71 NKG2C + γδT.MMP low %]]> Percentage of NKG2C-positive γδ T cell mitochondrial low membrane potential 13 Lymph% Lymphocyte percentage 72 NKG2D + L.MMP low %]]> Percentage of NKG2D-positive lymphocyte mitochondrial low membrane potential 14 Mono% Monocyte percentage 73 NKG2D + T.MMP low %]]> Percentage of NKG2D-positive T cell mitochondrial low membrane potential 15 Granu% Neutrophil percentage 74 NKG2D + Tc.MMP low %]]> Percentage of NKG2D-positive killer T cell mitochondrial low membrane potential 16 T% T lymphocyte percentage 75 NKG2D + NK.MMP low %]]> Percentage of NKG2D-positive natural killer cell mitochondrial low membrane potential 17 Th% Helper T cell percentage 76 NKG2D + NKT.MMP low %]]> Percentage of NKG2D-positive natural killer T cell mitochondrial low membrane potential 18 Tc% Cytotoxic T cell percentage 77 NKG2D + γδT.MMP low %]]> Percentage of NKG2D-positive γδ T cell mitochondrial low membrane potential 19 NK% Natural killer cell percentage 78 CD8 + NK.MMP low %]]> Percentage of CD8-positive natural killer cell mitochondrial low membrane potential 20 NKT% Natural killer T cell percentage 79 CD8 + NKT.MMP low %]]> Percentage of CD8-positive natural killer T cell mitochondrial low membrane potential 21 γδ T% γδ T cell percentage 80 CD8 + γδT.MMP low %]]> Percentage of CD8-positive γδ T cell mitochondrial low membrane potential 22 NKG2A + Lymph%]]> Percentage of NKG2A-positive lymphocyte 81 NKG2C + CD8 + NKT.MMP low %]]> Percentage of NKG2C, CD8 double-positive natural killer T cell mitochondrial low membrane potential 23 NKG2A + Mono%]]> Percentage of NKG2A-positive monocyte 82 NKG2D + CD8 + NKT.MMP low %]]> Percentage of NKG2D, CD8 double-positive natural killer T cell mitochondrial low membrane potential 24 NKG2A + Granu% Percentage of NKG2A-positive granulocyte 83 Granu.MM Neutrophil mitochondrial mass 25 NKG2A + T Percentage of NKG2A-positive T cell 84 Lymph.MM Lymphocyte mitochondrial mass 26 NKG2A + Tc Percentage of NKG2A-positive cytotoxic T cell 85 Mono.MM Monocyte mitochondrial mass 27 NKG2A + NK%]]> Percentage of NKG2A-positive natural killer cell 86 T.MM T cell mitochondrial mass 28 NKG2A + NKT Percent NKG2A positive natural killer T cells 87 Th.MM Helper T cell mitochondrial mass 29 NKG2A + γδT Percent NKG2C positive gamma delta T cells 88 Tc.MM Cytotoxic T cell mitochondrial mass 30 NKG2C + Lymph%]]> Percent NKG2C positive lymphocytes 89 Gamma delta T.MM Gamma delta T cell mitochondrial mass 31 NKG2C + Mono%]]> Percent NKG2C positive monocytes 90 NK.MM Natural killer cell mitochondrial mass 32 NKG2C + T Percent NKG2C positive T cells 91 NKT.MM Natural killer T cell mitochondrial mass 33 NKG2C + Tc%]]> Percent NKG2C positive cytotoxic T cells 92 NKG2A + L.MM]] NKG2A positive lymphocyte mitochondrial mass 34 NKG2C + NK%]]> Percent NKG2C positive natural killer cells 93 NKG2A + Mono.MM NKG2A positive monocyte mitochondrial mass 35 NKG2C + NKT Percent NKG2C positive natural killer T cells 94 NKG2A + T.MM]]> NKG2A positive T cell mitochondrial mass 36 NKG2C + γδT cells Percent NKG2C positive gamma delta T cells 95 NKG2A + Tc.MM]] NKG2A positive cytotoxic T cell mitochondrial mass 37 NKG2D + Granu% Percent NKG2D positive granulocytes 96 NKG2A + NK.MM]]> NKG2A positive natural killer cell mitochondrial mass 38 NKG2D + Lymph%]]> Percent NKG2D positive lymphocytes 97 NKG2A + NKT.MM]]> NKG2A positive natural killer T cell mitochondrial mass 39 NKG2D + Mono%]] Percent NKG2D positive monocytes 98 NKG2A + γδT.MM]] NKG2A positive gamma delta T cell mitochondrial mass 40 NKG2D + T Percent NKG2D positive T cells 99 NKG2C + G.MM]]> NKG2C positive neutrophil mitochondrial mass 41 NKG2D + Tc Percent NKG2D positive cytotoxic T cells 100 NKG2C + L.MM]] NKG2C positive lymphocyte mitochondrial mass 42 NKG2D + NK%]]> Percent NKG2D positive natural killer cells 101 NKG2C + Mono.MM NKG2C positive monocyte mitochondrial mass 43 NKG2D + NKT Percent NKG2D positive natural killer T cells 102 NKG2C + T.MM]] NKG2C positive T cell mitochondrial mass 44 NKG2D + γδT Percent NKG2D positive gamma delta T cells 103 NKG2C + Tc.MM]] NKG2C positive cytotoxic T cell mitochondrial mass 45 NKG2C + CD8 + NKT Percent NKG2C positive CD8 positive natural killer T cells 104 NKG2C + NK.MM]]> NKG2C positive natural killer cell mitochondrial mass 46 NKG2D + CD8 + NKT Percent NKG2D positive CD8 positive natural killer T cells 105 NKG2C + NKT.MM]]> NKG2C positive natural killer T cell mitochondrial mass 47 CD8 + NK%]]> Percent CD8 positive natural killer cells 106 NKG2C + γδT cells. MM NKG2C positive gamma delta T cell mitochondrial mass 48 CD8 + NKT Percent CD8 positive natural killer T cells 107 NKG2D + L.MM]] NKG2D positive lymphocyte mitochondrial mass 49 CD8 + γδT Percent CD8 positive gamma delta T cells 108 NKG2D + Mono.MM NKG2D positive monocyte mitochondrial mass 50 Lymph.Abs.Count Lymphocyte absolute count 109 NKG2D + G.MM]]> NKG2D positive neutrophil mitochondrial mass 51 Granu.MMP low %]]> Percent neutrophil mitochondrial low membrane potential 110 NKG2D + T.MM]] NKG2D positive T cell mitochondrial mass 52 Lymph.MMP low %]]> Percent lymphocyte mitochondrial low membrane potential 111 NKG2D + Tc.MM]] NKG2D positive cytotoxic T cell mitochondrial mass 53 Mono.MMP low %]]> Percent monocyte mitochondrial low membrane potential 112 NKG2D + NK.MM]]> NKG2D positive natural killer cell mitochondrial mass 54 T.MMP low %]]> Percent T cell mitochondrial low membrane potential 113 NKG2D + NKT.MM]]> NKG2D positive natural killer T cell mitochondrial mass 55 Th. MMP low %]]> Percent helper T cell mitochondrial low membrane potential 114 NKG2D + γδT.MM]]> NKG2D positive gamma delta T cell mitochondrial mass 56 Tc. MMP low ​ Percent cytotoxic T cell mitochondrial low membrane potential 115 CD8 + NK.MM]]> Percent CD8 positive natural killer cell mitochondrial mass 57 Gammadelta T.MMP low %]]> Percent gamma delta T cell mitochondrial low membrane potential 116 CD8 + NKT.MM]]> Percent CD8 positive natural killer T cell mitochondrial mass 58 NK.MMP low %]]> Percentage of mitochondrial low membrane potential in natural killer cells 117 CD8 + γδT.MM]]> mitochondrial mass of CD8-positive γδT cells

[0047] Example 2 Correlation and principal component analysis (PCA)

[0048] According to the body mass index (BMI) value, the healthy donors were divided into control group and obesity group (normal group 60 cases: BMI 19-23.9; obesity group 58 cases: >24 BMI) to construct correlation analysis. First, through correlation analysis, the candidate indicators (17) strongly correlated with BMI were screened out, that is, through flow cytometry detection and blood routine detection output 117 indicators, according to the correlation coefficient ρ value, 117 indicators in Table 1 (including 10 blood routine indicators, 37 cell percentage indicators, 1 lymphocyte absolute count indicator, 32 mitochondrial low membrane potential percentage indicators, 35 mitochondrial mass indicators, same as Table 1) were sorted from low to high, 5 negative correlation indicators (-0.27) and 12 positive correlation indicators (>0.27) were screened out, a total of 17 good correlation indicators (Table 2). Subsequently, we performed significance analysis on the 117 indicators between the two groups (P<0.05) P<0.05, normal group VS obesity group), 43 significant different indicators were selected (Table 3), and finally the intersection data of the two parts was obtained, with a total of 10 indicators.

[0049] Table 2

[0050] Serial Number Indicator abbreviation Correlation coefficient 1 HCT >0.27 2 HGB >0.27 3 NKG2D + Lymph%]]> >0.27 4 NKG2D + Tc >0.27 5 NKG2D + NKT.MMP low %]]> >0.27 6 NKG2D + NK%]]> >0.27 7 NKG2D + NKT >0.27 8 NKG2C + NKT.MMP low %]]> >0.27 9 CD8 + NK%]]> >0.27 10 Mono% >0.27 11 NKG2D + CD8 + NKT >0.27 12 MPV >0.27 13 NKG2C + NK.MMP low %]]> <-0.27 14 NKG2A + Lymph.MM]] <-0.27 15 Lymph.MM <-0.27 16 Lymph.MMP low ​ <-0.27 17 Mono.MMP low %]]> <-0.27

[0051] Table 3

[0052] Serial Number Indicator abbreviation Serial Number Indicator abbreviation 1 RBC 23 NKG2C + T.MMP low %]]> 2 HGB 24 NKG2C + NK.MMP low %]]> 3 HCT 25 NKG2C + NKT.MMP low %]]> 4 MPV 26 NKG2C + γδT.MMP low %]]> 5 PDW 27 NKG2D + NK.MMP low %]]> 6 Lymph% 28 NKG2D + NKT.MMP low %]]> 7 Mono% 29 Lymph.MM 8 NKG2A + Lymph%]]> 30 Mono.MM 9 NKG2A + Mono%]]> 31 T.MM 10 NKG2A + Granu% 32 Th.MM 11 NKG2A + T 33 γδT.MM 12 NKG2A + NKT 34 NK.MM 13 NKG2A + γδT 35 NKG2A + T.MM]] 14 NKG2D + Granu% 36 NKG2A + Tc.MM]] 15 CD8 + NKT 37 NKG2A + NK.MM]]> 16 CD8 + γδT 38 NKG2A + NKT.MM]]> 17 Granu.MMP low %]]> 39 NKG2A + γδT.MM]]> 18 Mono.MMP low %]]> 40 NKG2D + T.MM]] 19 T. MMP low %]]> 41 NKG2D + Tc.MM]] 20 Th. MMP low %]]> 42 NKG2D + NK.MM <!-- 7 -->]]> 21 NKG2A + L.MMP low %]]> 43 NKG2D + L.MM]] 22 NKG2A + L.MM]] / /

[0053] Ten indicators with the highest importance were selected from the combined data set of the four dimensions (blood routine indicators, cell percentage indicators, mitochondrial mass indicators, and mitochondrial low membrane potential percentage indicators), and a random forest analysis model was constructed based on these indicators. The leave-one-out cross-validation method was used to evaluate the machine learning model based on the omics data. The above process was repeated for each sample in the data set to ensure that each sample had the opportunity to be a validation set. Based on the training samples, the Boruta algorithm was used for feature selection, and the random forest analysis was used to select the best indicators with strong correlation with BMI from the 10 indicators when n = 6, which obtained the highest mean decline accuracy. Six best indicators with strong correlation with BMI were selected from the 10 indicators: NKG2C + NK.MMP low %, NKG2D + NKT.MMP low %, NKG2C + NKT.MMP low %, Mono.MMP low %, MPV, NKG2A + Lymph.MM Figures 1A-1C ). Number of decision trees = 5000, number of cross-validation interruptions = 10, number of reduced variables = 1.1, sample size: n = 118. Figure 1A The importance of indicators was ranked from high to low by the mean reduction accuracy in random forest analysis, and six indicators were selected from the 10 indicators that were both relevant and significantly different (blue). Figure 1B Based on the "Gini impurity" splitting criterion of the decision tree, the contribution of the feature to the "node purity improvement" in "each tree split" was quantified, and the more the contribution, the more important the feature. The Gini coefficient indicates that the six selected indicators are still the indicators with the largest contribution. Figure 1C The optimal number of indicators, i.e. n value, was determined by the cross-validation error of random forest analysis. According to the curve, when n = 6, the cross-validation error tends to be flat, and does not decrease significantly with the increase of the number of indicators, so n = 6 is selected as the optimal number of indicators.

[0054] The detection indicators of each donor in each cell subpopulation were taken as independent samples. First, the blood routine indicators, cell percentage indicators, mitochondrial mass indicators, and mitochondrial low membrane potential percentage indicator data were subjected to principal component analysis (PCA), and the first five principal components (accounting for 94.62% of the variance in the data) of all samples were subjected to K-value cluster analysis (k=2), according to which all indicators were divided into two categories: high BMI correlation and low BMI correlation. Min-max standardization must be used before principal component analysis (PCA). Based on the six best indicators strongly correlated with BMI selected, network-based principal component analysis (PCA) dimensionality reduction analysis obtained BMI related parameters (BRP, BMI related parameters), and their significant differences, ROC curve AUC, and related characteristics were analyzed, which can be used for NKT cell-related analysis to distinguish high BMI and low BMI. The performance of the model on the validation sample was evaluated by the area under the receiving operation characteristic curve (AUROC, area under the receiving operation characteristic curve) (Fig. 6). Figures 2A-2F By inputting the existing two modes of detection data, the expression of five principal components PC1-PC5 was obtained by calculating the six main parameters input. Figure 2A The contribution ratio of each principal component (PC1-PC5) in principal component analysis showed that PC1 contributed the most, followed by a gradual decrease, and the contribution ratio of each principal component was 34.8998%, 18.7809%, 17.0358%, 12.6778%, and 9.5987%. Moreover, principal component PC1 was better than other principal components in identifying and distinguishing groups (obese group, control group, or lean group). Figure 2B The PCA Bi-plot is a "sample-variable two-dimensional visualization chart" of principal component analysis. Each point represents an original sample, and its position is determined by the "principal component score" of the sample on PC1 and PC2. The horizontal and vertical coordinate values are the projection values of the sample on the principal component, reflecting the contribution of the sample to the principal component. Each arrow represents an original variable, and its direction, length, and angle correspond to "the correlation between the variable and the principal component", "the explanatory power of the variable to the principal component", and "the correlation between variables", respectively. It is a variable form of "principal component load" visualization. Figure 2C The index composition of each principal component was displayed, and the vertical coordinate was the coefficient value of the formula of each principal component. The specific principal component coefficient table is shown in Table 4.

[0055] Table 4

[0056] Principal component index PC1 PC2 PC3 PC4 PC5 MPV 0.4779 -0.3465 -0.4061 0.3794 0.4365 Mono.MMP low %]]> -0.3903 0.4903 0.2277 0.5708 0.4748 NKG2C + NK.MMP low %]]> -0.2587 -0.5401 -0.1014 0.2420 0.2017 NKG2C + NKT.MMP low %]]> 0.3560 -0.3063 0.8743 0.0959 0.0703 NKG2D + NKT.MMP low %]]> 0.2019 0.0953 -0.0630 0.6718 -0.6959 NKG2A + Lymph.MM]] 0.4022 0.3209 -0.0437 -0.0686 0.1509

[0057] The PC1 calculation formula is PC1=0.48*MPV-0.39*Mono.MMP.low -0.26 * NKG2C + NK.MMP low +0.36 * NKG2C + NKT.MMP low +0.20 * NKG2D + NKT.MMP low -0.40 * NKG2A + Lymph.MM, PC1~PC5 calculated values are shown in Table 5;

[0058] Table 5

[0059] Sample number Grouping Body Mass Index PC1 PC2 PC3 PC4 PC5 S001 Normal / Underweight Group 17.28 -0.0702 1.0607 0.5162 0.8551 -0.0958 S002 Normal / Underweight Group 18.26 0.0293 0.8013 0.4857 0.9293 -0.0798 S003 Normal / Underweight Group 18.37 -0.2784 0.6378 -0.2071 0.7271 -0.3154 S004 Normal / Underweight Group 18.66 -0.2484 0.9273 0.7579 0.8670 0.0005 S005 Normal / Underweight Group 18.82 0.1772 0.9264 0.4034 1.0246 -0.0193 S006 Normal / Underweight Group 19.27 0.1653 1.1234 0.1957 0.8721 -0.0112 S007 Normal / Underweight Group 19.36 0.2061 0.9345 0.3643 0.9759 0.0512 S008 Normal / Underweight Group 19.61 -0.0434 0.9331 0.6324 0.8946 0.1071 S009 Normal / Underweight Group 19.93 0.2160 1.1730 0.1371 0.8276 -0.1912 S0010 Normal / Underweight Group 20.18 -0.0631 1.0850 0.2881 0.8117 0.1956 S0011 Normal / Underweight Group 20.25 -0.0281 0.9328 0.1089 0.8423 -0.0659 S0012 Normal / Underweight Group 20.40 -0.1323 0.9349 0.5704 0.8774 -0.1891 S0013 Normal / Underweight Group 20.46 0.0431 0.8869 -0.0189 0.8481 -0.0428 S0014 Normal / Underweight Group 20.50 -0.0673 1.2565 -0.2171 0.7242 0.3100 S0015 Normal / Underweight Group 20.69 0.0979 1.1496 -0.0006 0.8741 -0.0003 S0016 Normal / Underweight Group 20.78 -0.0254 1.0860 0.1529 0.7513 -0.1385 S0017 Normal / Underweight Group 20.79 -0.2198 0.8024 0.3992 0.8651 0.2730 S0018 Normal / Underweight Group 21.09 0.0656 1.5499 -0.0896 0.7050 -0.1011 S0019 Normal / Underweight Group 21.14 -0.2668 0.7941 0.4006 0.8087 0.3368 S0020 Normal / Underweight Group 21.17 -0.3729 0.7876 0.7255 0.8158 0.2362 S0021 Normal / Underweight Group 21.40 0.0767 0.8124 0.2477 0.9292 0.1888 S0022 Normal / Underweight Group 21.44 0.0876 0.9124 0.5499 0.9137 -0.0435 S0023 Normal / Underweight Group 21.45 -0.2611 1.2715 0.2476 0.6793 0.1571 S0024 Normal / Underweight Group 21.67 0.0171 1.0287 0.6057 0.9069 -0.0968 S0025 Normal / Underweight Group 21.71 0.0598 1.4494 -0.1621 0.7374 0.0496 S0026 Normal / Underweight Group 21.89 0.0864 1.0681 0.1858 0.8731 -0.0888 S0027 Normal / Underweight Group 22.07 0.2434 1.1586 0.1752 0.9832 0.0406 S0028 Normal / Underweight Group 22.10 0.0331 0.9141 0.1132 0.8381 -0.0814 S0029 Normal / Underweight Group 22.13 0.0252 1.2802 -0.2530 0.7172 -0.0864 S0030 Normal / Underweight Group 22.21 -0.3309 0.4380 -0.0884 0.8006 -0.2100 S0031 Normal / Underweight Group 22.25 -0.6463 0.4299 0.2696 0.7431 -0.0947 S0032 Normal / Underweight Group 22.25 0.0262 0.9139 0.4779 0.9783 0.0331 S0033 Normal / Underweight Group 22.73 0.0550 1.1160 0.4945 0.8569 -0.0905 S0034 Normal / Underweight Group 22.85 0.1859 1.2389 0.2371 0.8344 -0.0656 S0035 Normal / Underweight Group 22.89 -0.0317 0.8108 0.5035 0.9157 0.0375 S0036 Normal / Underweight Group 22.93 0.1423 1.2418 0.2569 0.8339 -0.1211 S0037 Normal / Underweight Group 23.00 -0.0539 0.9682 0.4212 0.8811 -0.2059 S0038 Normal / Underweight Group 23.02 0.1525 1.0570 0.2581 0.9272 0.0184 S0039 Normal / Underweight Group 23.23 -0.0529 0.8768 0.5590 0.8898 0.0256 S0040 Normal / Underweight Group 23.50 -0.0042 0.9217 0.6115 0.8933 0.0928 S0041 Normal / Underweight Group 23.53 -0.0374 0.9873 0.2283 0.7482 0.0236 S0042 Normal / Underweight Group 23.53 0.1103 0.9728 0.6352 0.8482 -0.1936 S0043 Normal / Underweight Group 18.55 -0.3396 1.1652 -0.3956 0.5729 0.1152 S0044 Normal / Underweight Group 19.21 -0.1390 0.8067 0.5528 0.8375 -0.0361 S0045 Normal / Underweight Group 20.10 0.4095 0.9807 -0.0784 1.1274 0.2184 S0046 Normal / Underweight Group 20.60 0.3510 0.6904 0.1002 1.0562 0.0633 S0047 Normal / Underweight Group 21.00 0.0693 0.4318 -0.0054 0.9683 -0.2792 S0048 Normal / Underweight Group 21.07 -0.0540 0.9404 0.4764 0.9195 0.2515 S0049 Normal / Underweight Group 21.21 0.0108 1.2048 -0.0402 0.7907 0.1175 S0050 Normal / Underweight Group 21.60 0.2285 0.3898 -0.0626 1.0749 -0.2172 S0051 Normal / Underweight Group 22.53 -0.1980 1.1394 -0.0081 0.7133 0.1095 S0052 Normal / Underweight Group 22.63 -0.5724 0.5668 -0.2683 0.6602 0.0634 S0053 Normal / Underweight Group 22.64 -0.4828 0.6404 0.4006 0.8133 0.4545 S0054 Normal / Underweight Group 23.03 -0.2063 1.2449 -0.1654 0.6310 0.0527 S0055 Normal / Underweight Group 23.21 -0.2236 0.5835 -0.0613 0.8654 0.3765 S0056 Normal / Underweight Group 23.21 -0.3536 0.8067 -0.3463 0.7160 0.3994 S0057 Normal / Underweight Group 23.32 -0.3652 0.7853 0.1198 0.7672 0.3544 S0058 Normal / Underweight Group 23.37 -0.6314 0.9312 0.2590 0.6251 0.5884 S0059 Normal / Underweight Group 23.49 -0.5129 0.3626 0.3899 0.8397 0.2211 S0060 Normal / Underweight Group 23.50 0.0299 0.8390 0.5625 0.9927 0.0485 S0061 Obese group 25.00 0.2224 1.1863 0.1189 0.8097 -0.1137 S0062 Obese group 25.20 0.3029 1.0462 0.2803 0.9912 -0.2725 S0063 Obese group 25.20 0.2729 1.2742 -0.1064 0.8736 0.0743 S0064 Obese group 25.22 0.5413 1.0396 0.2404 1.1367 -0.0350 S0065 Obese group 25.48 0.5029 1.0524 0.2601 1.0256 0.1065 S0066 Obese group 25.48 0.1691 1.1398 0.4464 0.8036 0.0315 S0067 Obese group 25.54 -0.0349 1.0385 0.0655 0.7714 0.0360 S0068 Obese group 25.56 0.0735 0.9146 0.3298 0.9042 -0.0291 S0069 Obese group 25.94 0.0660 1.2134 0.5965 0.8261 0.0269 S0070 Obese group 26.10 0.4133 1.0569 0.2522 1.0349 0.1531 S0071 Obese group 26.14 0.0301 1.3789 -0.2897 0.7069 -0.0737 S0072 Obese group 26.24 0.1875 1.0685 0.0667 0.9054 -0.0377 S0073 Obese group 26.80 0.2998 0.6881 -0.1360 1.1139 -0.2745 S0074 Obese group 27.10 0.9050 0.8019 0.3057 0.3951 0.1291 S0075 Obese group 27.40 0.9533 0.6069 0.2683 0.3709 0.0869 S0076 Obese group 27.51 0.3595 0.9089 0.3189 1.0600 0.1534 S0077 Obese group 28.10 0.9228 0.6810 0.2694 0.3848 0.0312 S0078 Obese group 28.40 0.3444 1.1303 0.0461 0.9035 0.0053 S0079 Obese group 29.09 0.0222 0.9956 0.3611 0.8381 -0.0977 S0080 Obese group 29.20 0.8329 0.9458 0.3556 0.4986 0.0705 S0081 Obese group 29.20 0.5766 0.8831 0.2603 0.6426 -0.0031 S0082 Obese group 29.50 0.7434 0.8875 0.3198 0.5375 0.0887 S0083 Obese group 29.90 0.5305 0.4654 0.1052 0.5270 -0.4060 S0084 Obese group 30.41 0.4292 0.8203 0.1267 1.1395 0.0481 S0085 Obese group 31.90 0.4577 0.9247 0.1364 1.0663 0.0712 S0086 Obese group 31.91 0.1360 0.9988 0.4135 0.7968 -0.1091 S0087 Obese group 25.00 0.1678 1.0741 0.3057 0.8510 -0.0502 S0088 Obese group 25.00 -0.0122 0.9809 0.7760 0.8912 -0.0592 S0089 Obese group 25.00 0.0532 1.2874 -0.0162 0.7727 0.0266 S0090 Obese group 25.00 0.5074 1.1395 0.2606 1.0888 0.0184 S0091 Obese group 25.10 0.3733 0.9950 0.1967 1.0732 0.2480 S0092 Obese group 25.40 0.2345 0.7305 0.1514 1.0119 -0.5215 S0093 Obese group 25.40 0.5780 1.0527 0.2660 1.1013 0.0827 S0094 Obese group 25.50 0.4205 0.8220 0.0958 1.1226 -0.3104 S0095 Obese group 25.80 0.1369 0.2179 -0.1105 1.0188 0.4394 S0096 Obese group 25.82 0.1372 1.5479 -0.2268 0.7089 0.0232 S0097 Obese group 26.00 -0.0587 1.2852 -0.1171 0.7225 -0.0138 S0098 Obese group 26.03 0.1302 0.9333 0.5326 0.9228 -0.0300 S0099 Obese group 26.13 0.0321 1.0555 0.4954 0.8073 -0.0187 S00100 Obese group 26.15 -0.0128 0.7616 0.5469 0.8824 0.0610 S00101 Obese group 26.23 0.1636 1.1930 0.3531 0.8790 0.0547 S00102 Obese group 26.30 0.6149 0.8713 0.2040 0.8802 0.0885 S00103 Obese group 26.40 0.4515 0.9105 0.1384 1.0955 0.3083 S00104 Obese group 26.49 0.2709 1.2077 0.1972 0.9406 0.0869 S00105 Obese group 26.51 -0.1169 0.9989 0.3562 0.8363 0.0260 S00106 Obese group 26.59 -0.0310 1.0008 0.3809 0.8097 0.0119 S00107 Obese group 26.80 0.5345 1.1466 0.3167 1.1193 -0.0137 S00108 Obese group 26.80 0.0356 1.3084 -0.2680 0.6677 -0.0406 S00109 Obese group 26.97 0.3139 1.0104 0.1912 1.0221 -0.1402 S00110 Obese group 27.01 0.5868 1.0133 0.2341 0.9999 -0.0033 S00111 Obese group 28.41 0.0124 1.0412 0.5839 0.7810 0.0235 S00112 Obese group 28.88 0.3207 1.1952 0.1167 0.8566 0.0604 S00113 Obese group 29.48 0.1162 1.0169 0.2144 0.8234 -0.0554 S00114 Obese group 29.80 0.6155 0.8619 0.1384 1.1722 0.1924 S00115 Obese group 30.00 0.3620 0.8773 0.2003 0.9923 -0.0404 S00116 Obese group 30.63 0.1188 1.0096 0.1978 0.9294 0.1716 S00117 Obese group 31.00 0.7403 0.7289 0.0319 1.2596 0.1534 S00118 Obese group 33.52 0.2073 1.0804 0.3947 0.8391 -0.0264

[0060] By ROC curve drawing for the obesity group and control / slim group data of Table 5, the results are shown in Figure 2D PC1 as a BMI-related parameter (BRP) is the best in the ROC, AUC = 85.78, sensitivity is 75.86%, and specificity is 81.67%. Figure 2E To show the difference of BMI-related parameters (BRP) in normal group VS obesity group, the cut-off value is 0.113 (0.111~0.116), less than the value is negative, greater than the value is positive, and the PC1 value of normal group VS obesity group has extremely significant difference (t= - 5. 59, P < 0. 001). P Figure 2F The correlation of BMI-related parameters (BRP) and BMI is plotted (obesity group n = 58, control / slim group n = 60), and the results are shown in

[0061] ​Seventeen indicators with strong correlation with BMI were screened out by correlation analysis, and six indicators were screened out by random forest analysis. The BMI related parameters (BRP) values obtained by PCA can represent the mitochondrial immunometabolic characteristics of NKT cells in obese or overweight populations. Currently, in the immune evaluation of obesity-related metabolic syndrome, the percentage and number of immune cells obtained by blood routine test are often used to judge, among which natural killer (NK) cells show significant correlation with obesity. The core pathological feature of obesity is chronic low-grade inflammation in white adipose tissue (WAT), often accompanied by insulin resistance, hyperglycemia, hyperlipidemia and other metabolic disorders. As a key effector cell of innate immunity, NK cells play an important role in the occurrence and development of obesity by regulating the microenvironment of adipose tissue, participating in inflammatory response and metabolic regulation. The prediction method of the present application covers the key markers of NK cell functional status, i.e. activating receptor NKG2D and inhibitory receptor NKG2A, comprehensively reflecting its immune function from two dimensions of positive and negative regulation. We are committed to screening more accurate immune indicators and combining multi-index joint analysis strategy to establish a new evaluation system based on immune mechanism to more accurately identify obesity-related metabolic syndrome.

[0062] Verification of methodology

[0063] We randomly collected 42 cases of physical examination as verification samples, and detected NKG2C + NK.MMP low %, NKG2D + NKT.MMP low %, NKG2C + NKT.MMP low %, Mono.MMP low %, MPV, NKG2A + Lymph.MM of the 42 cases of physical examination by the method of embodiment 1. The detection data were standardized by min-max standardization, and the standardized data were calculated to obtain PC1 (BRP) values by the calculation formula of embodiment 2. The standardized values of each index and PC1 values are shown in Table 6. The correlation analysis of PC and BMI values in Table 6 is shown in Figure 3 p ​<0.001). Due to the strong correlation of BRP value with BMI, it can represent the mitochondrial immunometabolic characteristics of NKT cells in obese or overweight population. That is, not all obese population with BMI > 24 will have abnormal mitochondrial indicators of NKT cells. We can further distinguish the population with abnormal expression of immune cell metabolism in obese population by this method (such as mitochondrial dysfunction cannot normally perform oxidative phosphorylation, and / or ROS abnormally increased, and mitochondrial autophagy in mitochondrial metabolism cannot normally proceed, etc.), to screen the population with abnormal immune cells that urgently need to lose weight or restrict diet. The increase of this mitochondrial metabolism indicator can lead to decreased mitochondrial metabolic activity and mitochondrial dysfunction. NKG2C + The decrease of NK cell mitochondrial low membrane potential suggests that the expression of some cell types in NK cells can depend on the metabolic mode of TCA cycle, and the mitochondrial metabolism of obese population is more active than that of control group. This feature is consistent with NKG2A + The characteristics of lymphocytes and monocytes are consistent.

[0064] Table 6

[0065] Validation Sample BMI age MPV Mono.MMP low %]]> NKG2C + NK.MMP low %]]> NKG2C + NKT.MMP low %]]> NKG2D + NKT.MMP low %]]> NKG2A + Lymph.MM]] PC1(BRP) SuppS01 23.73 56 0.0309 1.0000 0.5224 0.2500 0.4668 0.3891 -0.3971 SuppS02 25.75 31 0.0389 0.0095 0.2081 0.1250 0.0687 0.5688 -0.2136 SuppS03 21.06 15 0.0241 0.0206 0.1517 0.1250 0.0806 0.4877 -0.1649 SuppS04 20.19 44 0.0183 0.0364 0.2792 0.1429 0.4396 0.6311 -0.1173 SuppS05 28.93 40 0.0275 0.0499 0.0000 1.0000 0.3277 0.8640 0.1812 SuppS06 20.86 49 0.0321 0.0245 0.3079 0.0857 0.2430 0.8804 -0.3231 SuppS07 23.61 51 0.0424 0.0111 0.3292 0.2115 0.2831 0.5385 -0.1235 SuppS08 22.79 28 0.0332 0.0150 0.4182 0.0000 0.4232 0.5620 -0.1937 SuppS09 26.31 63 0.0275 0.0055 0.2482 0.0000 0.0000 0.7896 -0.3976 SuppS10 19.47 28 0.0286 0.0625 0.6274 0.1034 0.6519 0.5555 -0.1557 SuppS11 19.65 58 0.0103 0.0214 0.3642 0.0476 0.1849 1.0000 -0.4417 SuppS12 21.33 36 0.0137 0.0491 0.1954 0.0690 0.1334 0.6756 -0.2690 SuppS13 24.45 28 0.0389 0.0483 0.2023 0.1359 0.3944 0.5899 -0.0867 SuppS14 25.06 29 0.0080 0.0245 0.6899 0.2286 0.6173 0.6224 -0.1714 SuppS15 29.97 32 0.0195 0.0435 0.6328 0.2667 0.6061 0.3426 -0.0307 SuppS16 24.22 43 0.0630 0.0816 0.4604 0.2105 0.7169 0.4234 0.0425 SuppS17 22.48 67 0.0367 0.0198 0.5747 0.2857 0.6097 0.4453 -0.0219 SuppS18 23.24 60 0.0275 0.0261 0.9078 0.4500 0.4451 0.1228 -0.0475 SuppS19 30.80 72 0.0263 0.0222 0.4701 0.5172 1.0000 0.4371 0.2762 SuppS20 22.94 56 0.0298 0.0499 0.7116 0.1667 0.4706 0.4187 -0.1867 SuppS21 27.05 54 0.0298 0.0649 0.2869 0.0000 0.2017 0.3518 -0.1740 SuppS22 27.34 49 0.0309 0.0744 1.0000 0.5714 0.9857 0.2876 0.1178 SuppS23 21.09 39 0.0275 0.0736 0.9617 0.7500 0.3021 0.2662 -0.0904 SuppS24 24.41 37 0.0458 0.0166 0.9185 0.6000 0.8926 0.1525 0.2009 SuppS25 23.23 65 0.0424 0.0348 0.8414 0.4167 0.6257 0.2027 0.0186 SuppS26 23.41 41 0.0538 0.0182 0.7819 0.7143 0.7300 0.2280 0.2026 SuppS27 20.25 24 0.0619 0.0380 0.2909 0.0909 0.3005 0.6260 -0.1803 SuppS28 25.28 30 0.0000 0.0206 0.4884 0.1714 0.4543 0.4953 -0.1354 SuppS29 22.62 20 0.0481 0.0158 0.4215 0.1591 0.2244 0.4259 -0.1615 SuppS30 28.07 51 0.0401 0.0333 0.4549 0.1429 0.6136 0.3777 -0.0023 Supp S31 28.91 39 0.0401 0.0420 0.8883 0.4286 0.5971 0.1883 -0.0063 Supp S32 27.97 66 1.0000 0.0000 0.9948 0.3333 0.8296 0.0000 0.5625 Supp S33 23.31 51 0.0538 0.0269 0.2785 0.3750 0.5036 0.3546 0.1270 Supp S34 18.60 25 0.0321 0.0063 0.2833 0.0345 0.1436 0.4536 -0.2031 Supp S35 31.14 67 0.0195 0.0380 0.9901 0.6000 0.7922 0.1650 0.1031 Supp S36 26.23 30 0.0195 0.0198 0.9244 0.0000 0.9583 0.1905 -0.0290 Supp S37 22.89 40 0.0252 0.0491 0.3078 0.1176 0.4893 0.5293 -0.0782 Supp S38 23.15 29 0.0424 0.0325 0.4269 0.1389 0.4659 0.4467 -0.0825 Supp S39 26.53 37 0.0172 0.0467 0.5226 0.4118 0.4185 0.3925 -0.0344 Supp S40 26.99 40 0.0195 0.0475 0.5567 0.2857 0.8362 0.3231 0.1141 Supp S41 20.43 51 0.0126 0.0293 0.9112 0.2800 0.6893 0.1734 -0.0347 Supp S42 27.99 55 0.0172 0.0301 0.8694 0.5000 0.6347 0.1953 0.0353

[0066] By drawing ROC curve for obese group and control / slim group data, the results are shown in Table 7 and Figure 4 PC1 as BMI related parameter (BRP) is the best in ROC, AUC = 82.27%, sensitivity is 75.86%, and specificity is 81.67%.

[0067] Table 7

[0068] Feature \ Indicator PC1 (BRP) PC2 PC3 PC4 PC5 Area under the curve 82.27% 78.86% 56.14% 67.50% 66.43% Standard error 0.06546 0.07237 0.09085 0.08739 0.08726 95% confidence interval CI 95% ]]> 69.44%~95.10% 64.68%~93.05% 38.33%~73.94% 50.37%~84.63% 49.33%~83.53% Significant difference value 0.0003 0.0014 0.4965 0.0525 0.0719 Accuracy 80.95% 76.19% 64.29% 76.19% 69.05% Precision 83.33% 100.00% 59.26% 69.57% 64.00% Sensitivity 75.00% 50.00% 80.00% 84.21% 80.00% Specificity 86.36% 100.00% 50.00% 69.57% 59.09%

[0069] Of course, the above only describes specific embodiments of the present application, and is not intended to limit the scope of the present application. Any equivalent changes or modifications made in accordance with the features and principles described in the patent application of the present application shall be included in the patent application of the present application.

Claims

1. A kit for predicting obesity-related metabolic syndrome, characterized by, The kit comprises reagents for detecting the following indicators: NKG2C + Percentage of mitochondrial low membrane potential in NK cells (NKG2C + NK.MMP low Percentage of mitochondrial low membrane potential in NK cells (NKG2C + NKT.MMP + Percentage of mitochondrial low membrane potential in NK cells (NKG2C low NKT.MMP + Percentage of mitochondrial low membrane potential in NK cells (NKG2C + NKT.MMP low Percentage of mitochondrial low membrane potential in monocytes (Mono.MMP low Percentage of mitochondrial low membrane potential in monocytes (Mono.MMP + Percentage of mitochondrial low membrane potential in monocytes (Mono.MMP + Lymph.MM) and mean platelet volume (MPV).

2. A method for predicting obesity-related metabolic syndrome, characterized by, The method comprises the following steps: a) obtaining the index detection value of the subject using the kit of claim 1; b) standardizing the index detection value; c) inputting the standardized data into a pre-trained prediction model to output a BMI-related parameter (BRP) value, and determining whether the subject has a risk of obesity-related metabolic syndrome according to the BRP value.

3. The method of claim 2, wherein, The standardization adopts a min-max standardization method.

4. The method of claim 3, wherein, The prediction model is a principal component analysis (PCA) model, wherein a first principal component (PC1) is used as the BRP value for risk prediction.

5. The method of claim 4, wherein, The BRP value calculation formula is: BRP value = PC1 = 0.48*MPV-0.39*Mono.MMP low -0.26*NKG2C + NK.MMP low +0.36*NKG2C + NKT.MMP low +0.20*NKG2D + NKT.MMP low -0.40*NKG2A + Lymph.MM.

6. A system for predicting obesity related metabolic syndrome, characterized in that, The method comprises: an input module for receiving the index detection value obtained by the kit of claim 1; a preprocessing module for standardizing the index detection value; an analysis module for inputting the standardized data into a pre-trained PCA model to calculate a BMI-related parameter (BRP) value; and an output module for outputting the risk assessment result of obesity-related metabolic syndrome.

7. Use of the kit of claim 1 in a product for predicting obesity-related metabolic syndrome.

Images