Metabolically unhealthy obesity identification model construction system, storage medium and kit

Abstract

The present invention relates to the field of medical prediction, and provides a system for building an identification model for metabolically unhealthy obesity, a storage medium, and a kit. The identification model building system includes: an acquisition module, a feature selection module, a training module, and an identification module. Through machine learning of plasma metabolite biomarkers, it is possible to identify patients with metabolically unhealthy obesity early before clinical symptoms appear, helping doctors and patients take intervention measures before cardiovascular risks increase. The identification model building system for metabolically unhealthy obesity of the present invention is simpler than the traditional physical examination process, reducing the burden of patient participation.

Description

Technical Field

[0001] The present invention relates to the field of medical prediction, and in particular to a recognition model construction system, a storage medium, a kit and applications thereof for metabolically unhealthy obesity. Background Art

[0002] Studies have gradually shown that not all obese patients show the same differences in cardiovascular disease risk. Based on these observations, the concepts of metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO) came into being. On the basis of combined obesity, if hypertension and lipid metabolism disorders occur, they are classified as MUO. Such patients usually have visceral fat and ectopic fat deposition. Under physiological and pathological effects, adipose tissue can secrete a large amount of harmful adipose factors, which are widely involved in the progression of various cardiovascular diseases. In contrast, MHO is a benign obesity. Although such patients are overweight, they do not have obvious metabolic disorders such as hypertension, hyperglycemia or dyslipidemia, and their risk of cardiovascular disease is relatively low. MUO patients are accompanied by hypertension, lipid metabolism disorders and other symptoms, and their risk of cardiovascular events is significantly increased.

[0003] Currently, the differential diagnosis of MHO and MUO mainly relies on detecting whether the patient has clinical symptoms such as elevated blood sugar, blood pressure or blood lipid levels. However, by the time these symptoms appear, the risk of cardiovascular disease in MUO patients is already very high. Existing physical examinations require multiple hematological measurements and a series of questionnaires involving medical history and medication history. The process is cumbersome and time-consuming and requires the full cooperation of the participants, and it cannot effectively identify metabolically unhealthy states at an early stage. Summary of the invention

[0004] The object of the present invention is to provide a system, storage medium and kit for building a recognition model of metabolically unhealthy obesity. The system for building a recognition model of metabolically unhealthy obesity includes an acquisition module, a feature selection module, a training module and a recognition module. By using the relative concentration values ​​of specific plasma small molecule metabolic biomarkers to build a recognition model of metabolically unhealthy obesity, patients with metabolically unhealthy obesity can be identified early before clinical symptoms appear, helping doctors and patients take intervention measures before cardiovascular risks increase.

[0005] In order to solve the above problems, the present invention provides the following technical solutions.

[0006] In the first aspect, the present invention provides a recognition model construction system for metabolically unhealthy obesity, which includes: an acquisition module, a feature selection module, a training module and an identification module. In this embodiment, the acquisition module is used to collect plasma samples of each subject with metabolically healthy obesity and metabolically unhealthy obesity, and obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample. The feature selection module is used to obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample, perform feature selection, and screen out plasma small molecule metabolic biomarkers for early identification of metabolically unhealthy obesity as plasma small molecule metabolic biomarkers after feature selection. The training module is used to train the final prediction model for early identification of metabolically unhealthy obesity based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample. The identification module is used to obtain the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the person to be tested, and input the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the person to be tested into the final prediction model to obtain a prediction result of whether the person to be tested belongs to metabolically unhealthy obesity.

[0007] Furthermore, in the above-mentioned identification model construction system for metabolically unhealthy obesity, the acquisition module is used to obtain a first subject group and a second subject group that do not overlap each other, wherein the first subject group and the second subject group both include non-overlapping subjects with metabolically healthy obesity and metabolically unhealthy obesity. The acquisition module can also be used to collect first plasma samples from the first subject group, and obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each first plasma sample as a first set. The acquisition module can also be used to collect second plasma samples from the second subject group, and obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each second plasma sample as a test set.

[0008] Furthermore, in the above-mentioned metabolically unhealthy obesity identification model construction system, the feature selection module is used to perform feature selection on all plasma small molecule metabolic biomarkers in the first set, such as 303 plasma small molecule metabolic biomarkers, using the Lasso feature selection algorithm and five-fold cross validation on the R (4.3.1) software. The implementation of the Lasso feature selection algorithm depends on the glmnet package, wherein the glmnet package generates a sequence of 100 values ​​corresponding to the hyperparameter lambda through a built-in program, which is used to control the intensity of the L1 regularization penalty on the model coefficients and determine the lambda.min when the error rate is the lowest in the five-fold cross validation. The final calculated lambda.min = 0.02, and the metabolite biomarkers are screened out based on lambda.min.

[0009] Furthermore, in the above-mentioned metabolically unhealthy obesity recognition model construction system, the feature selection module is used to divide the first set into 5 non-overlapping data, and each time use one of the 5 non-overlapping data that has not been selected as an internal validation set, and each time use the remaining 4 data as training sets; wherein the internal validation set and the training set both include non-overlapping subjects with metabolically healthy obesity and metabolically unhealthy obesity. The feature selection module can also be used to use the Lasso feature selection algorithm on the R (4.3.1) software based on each training set, and cyclically train the feature selection model for 5 rounds to obtain the corresponding 5 feature selection models.

[0010] The feature selection module can also be used for each value in the sequence of 100 values ​​corresponding to the hyperparameter lambda, input a corresponding feature selection model in the 5 feature selection models in each round of the corresponding internal validation set, and obtain a misclassification rate corresponding to each value on each internal validation set; from the 5 misclassification rates corresponding to each value, an average misclassification rate corresponding to each value is obtained; based on the minimum average misclassification rate among all values, lambda.min with the lowest corresponding misclassification rate is selected, where lambda is one of the hyperparameters; based on lambda.min with the lowest misclassification rate, the corresponding metabolite marker is obtained as the plasma small molecule metabolic biomarker after feature selection.

[0011] Furthermore, in the above-mentioned metabolically unhealthy obesity identification model construction system, the plasma small molecule metabolic biomarkers after feature selection include: thioethylamine (Cysteamine), 4-methylpentanal (4-Methylpentanal), 3-hydroxybutyric acid ((R)-3-Hydroxybutyric acid), aminophenol (4-Aminophenol), pyruvic acid (Pyruvic acid), threonine (L-Threonine), succinic anhydride (Succinic anhydride), threitol (D-Threitol), glycerol (Glycerol), hypotaurine (Hypotaurine), creatinine (Creatinine), adenine (Adenine), valine (L-Valine), cysteine ​​(L-Cysteine), nicotinic acid (Picolinic acid), taurine (Taurine), glutamic acid (L-Glutamic acid), L-Methionine, Citrulline, D-Glucose, L-Arginine, Ascorbic acid, Lipoic acid, Propionylcarnitine, Glycerophosphocholine, Stearic acid and Sphingosine.

[0012] Furthermore, in the above-mentioned metabolically unhealthy obesity identification model construction system, the training module can be used to train a prediction model for early identification of metabolically unhealthy obesity based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample.

[0013] The training module can also be used to perform performance evaluation on the trained prediction model for early identification of metabolically unhealthy obesity, and use the prediction model for early identification of metabolically unhealthy obesity whose performance evaluation meets preset requirements as the final prediction model.

[0014] Furthermore, in the above-mentioned metabolically unhealthy obesity identification model construction system, the training module is used to use the elastic network algorithm on the R (4.3.1) software and based on the first set to train a prediction model for early identification of metabolically unhealthy obesity. The implementation of the elastic network algorithm depends on the glmnet package, wherein the ElasticNet algorithm hyperparameter alpha for controlling the ratio of L1 regularization and L2 regularization is set to 0.5; and s for controlling the penalty intensity of the parameter is set to 0.01.

[0015] Furthermore, in the above-mentioned metabolically unhealthy obesity identification model construction system, the training module is used to obtain the trained metabolically unhealthy obesity prediction model for early identification on the R (4.3.1) software, and the recognition performance on the first set and the recognition performance on the test set respectively; if the recognition performance of the trained metabolically unhealthy obesity prediction model for early identification in the training set and the test set differs by less than a preset difference threshold, the trained metabolically unhealthy obesity prediction model for early identification is used as the final prediction model.

[0016] In a second aspect, the present application provides a computer-readable storage medium having computer-executable instructions stored thereon, wherein when the computer-executable instructions are executed by a processor, the processor is caused to perform the following steps:

[0017] Step S1, the collection module collects plasma samples of each subject with metabolically healthy obesity and metabolically unhealthy obesity, and obtains plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample;

[0018] Step S2, the feature selection module obtains plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample, performs feature selection, and screens out plasma small molecule metabolic biomarkers for early identification of metabolically unhealthy obesity as plasma small molecule metabolic biomarkers after feature selection;

[0019] Step S3, the training module trains a final prediction model for early identification of metabolically unhealthy obesity based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample;

[0020] Step S4, the recognition module obtains the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested, and inputs the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested into the final prediction model to obtain a prediction result of whether the subject to be tested belongs to metabolically unhealthy obesity.

[0021] In this embodiment, the plasma small molecule metabolic biomarkers after feature selection include: mercaptoethylamine, 4-methylvaleraldehyde, 3-hydroxybutyric acid, aminophenol, pyruvate, threonine, succinic anhydride, threitol, glycerol, hypotaurine, creatinine, adenine, valine, cysteine, isonicotinic acid, taurine, glutamate, methionine, citrulline, glucose, arginine, ascorbic acid, lipoic acid, acetylcarnitine, glycerophosphocholine, stearic acid and sphingosine.

[0022] In a third aspect, the present application provides a detection kit for early identification of metabolically unhealthy obesity, the detection kit comprising metabolite biomarkers, which metabolite biomarkers include at least the following substances: mercaptoethylamine, 4-methylvaleraldehyde, 3-hydroxybutyric acid, aminophenol, pyruvate, threonine, succinic anhydride, threitol, glycerol, hypotaurine, creatinine, adenine, valine, cysteine, isonicotinic acid, taurine, glutamate, methionine, citrulline, glucose, arginine, ascorbic acid, lipoic acid, acetylcarnitine, glycerophosphocholine, stearic acid and sphingosine.

[0023] Specifically, the beneficial effects of the present invention are as follows.

[0024] First, it has strong early identification capabilities: through metabolomics, a comprehensive analysis of metabolites in plasma can be conducted to promptly identify metabolic abnormalities in metabolically unhealthy obese patients before they show clinical symptoms such as hyperglycemia and hypertension, thereby providing a scientific basis for early intervention.

[0025] Secondly, high throughput and high sensitivity: The identification model construction system for metabolically unhealthy obesity is based on high-throughput mass spectrometry technologies such as nanoparticle enhanced laser desorption / ionization time-of-flight mass spectrometry (MALDI-TOF-MS). It can detect a large number of metabolites at the same time and has extremely high sensitivity and specificity, ensuring the accuracy of identifying metabolically unhealthy obesity.

[0026] In addition, convenience and scalability: plasma sample collection is easy, the detection process is simplified, and it can be applied to the screening of large populations, suitable for a variety of scenarios such as clinical, scientific research and physical examinations.

[0027] Finally, personalized health management support: Based on the identification model and metabolite combination of the present invention, it is possible to provide data support for personalized obesity management, help formulate personalized intervention plans, and reduce the risk of cardiovascular diseases related to obesity.

[0028] Therefore, the present invention can not only provide an efficient and reliable technical solution for the early identification of metabolically unhealthy obesity, but also provide important basic data and application prospects for cardiovascular disease prevention and obesity intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 1 is a graph of relative concentration values ​​of 303 metabolite biomarkers in the plasma of metabolically healthy obese (MHO) and metabolically unhealthy obese (MUO) subjects according to an embodiment of the present invention.

[0030] Figure 2 1 is a graph showing the performance of a combination of 27 metabolic biomarkers in identifying MUO in the first set and an independent test set according to an embodiment of the present invention.

[0031] Figures 3a to 3c is a 10-year cardiovascular event risk comparison based on an independent test set in one embodiment of the present invention, Figure 3a The middle is the number of MHO and MUO in the independent test set based on clinical symptoms / metabolomics prediction; Figure 3b Middle is the comparison of 10-year cardiovascular event risk between MHO and MUO patients predicted by metabolomics in an independent test set; Figure 3c Figure 3 Comparison of 10-year cardiovascular event risk between MHO and MUO patients predicted by metabolomics in individuals diagnosed with MHO based on clinical symptoms in an independent test set. DETAILED DESCRIPTION

[0032] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0033] In a typical configuration of the present application, the terminal, the device of the service network and the trusted party all include one or more processors (CPU), input / output interfaces, network interfaces and memories.

[0034] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0035] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0036] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail in combination with the embodiments below. The experimental methods without specific conditions in the embodiments are usually carried out under conventional conditions, such as those described in textbooks and experimental guides, or under the conditions recommended by the manufacturer. The supporting software is operated under the recommended conditions.

[0037] The present invention provides a recognition model construction system for metabolically unhealthy obesity, the system comprising a collection module, a feature selection module, a training module and a recognition module.

[0038] In one embodiment, the collection module can be used to collect plasma samples from subjects with metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO), and obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample.

[0039] Here, metabolomics, as a cutting-edge omics technology, can detect metabolic disorders in the body before disease symptoms appear by performing high-throughput, global analysis of metabolites in biological samples. Therefore, it shows great potential in the early diagnosis of diseases. Plasma, as an easily accessible and information-rich biological fluid, is an important source of metabolites in the body. Plasma-based metabolomics analysis can provide a valuable resource for the early identification of MUO patients and promote the prevention and intervention of obesity-related cardiovascular diseases.

[0040] First, the acquisition module can obtain a first subject group and a second subject group that do not overlap each other, wherein the first subject group and the second subject group both include non-overlapping subjects with metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO).

[0041] Specifically, the first subject group includes metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO); the second subject group includes metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO); the first subject group and the second subject group do not overlap.

[0042] Secondly, the collection module can collect first plasma samples of the first subject group, and obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each first plasma sample as a first set;

[0043] Finally, the acquisition module can collect second plasma samples from the second subject group, and obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each second plasma sample as a test set;

[0044] Specifically, with the approval of a hospital ethics committee, subjects from three different communities from January 2019 to March 2019 can be recruited to obtain the first set and the test set. The first set includes: 167 metabolically healthy obese (MHO) and 167 metabolically unhealthy obese (MUO) subjects from the first and second communities; the first set can be split into a training set and an internal validation set in subsequent feature screening. It should be noted that the feature screening model requires an internal validation set, but the obesity model does not need an internal validation set.

[0045] The test set included 39 metabolically healthy obese (MHO) and 39 metabolically unhealthy obese (MUO) cases from a third community, which were used to independently verify the accuracy and clinical value of the metabolomics-based model.

[0046] The inclusion criteria for the study population may be as follows:

[0047] Obesity criteria: According to the 2004 “Expert Consensus on Obesity Prevention and Treatment among Chinese Residents” issued by the Chinese Nutrition Society, subjects with a BMI ≥ 28 kg / m² are defined as obese.

[0048] Metabolic abnormality criteria: According to the diagnostic criteria for metabolic syndrome of the Chinese Medical Association Diabetes Branch in 2004, it includes the following three items:

[0049] Hyperglycemia: fasting blood sugar ≥6.1mmol / L (110mg / dl) or blood sugar ≥7.8mmol / L (140mg / dl) two hours after a meal, or those who have been diagnosed with diabetes and are receiving treatment;

[0050] Hypertension: systolic / diastolic blood pressure ≥ 140 / 90 mmHg, or those who have been diagnosed with hypertension and received treatment;

[0051] Dyslipidemia: fasting serum triglycerides ≥1.7mmol / L (150mg / dl), or high-density lipoprotein cholesterol (HDL-C) <0.9mmol / L (male) or <1.0mmol / L (female).

[0052] MHO refers to obese people without the above metabolic abnormalities, and MUO refers to obese people with at least one or more of the above metabolic abnormalities.

[0053] Exclusion criteria: Subjects with acute or infectious clinical symptoms (such as fever, cough, sore throat, etc.) within three weeks before sampling were excluded to ensure that metabolic markers were not affected by acute diseases.

[0054] Obtaining a plasma sample and a relative concentration value from a subject may include steps S121 to S128 as described below.

[0055] Step S121 includes: treating the plasma sample of peripheral venous blood of the subject in the early morning fasting state with EDTA anticoagulation and storing it at -80°C. In this step, 1 ml of peripheral venous blood of the subject can be collected in the early morning fasting state and treated with EDTA anticoagulation tube. After plasma separation, the sample is stored at -80°C to ensure the stability of metabolites.

[0056] Step S122 includes: preparation of instruments and reagents: preparation of deionized water, matrix (inorganic nanoparticles), and nanoparticle enhanced laser desorption / ionization time-of-flight mass spectrometer (MALDI-TOF-MS, Bruker, Germany).

[0057] Step S123 includes: diluting the plasma sample: diluting the plasma sample with deionized water at a standard ratio to obtain a diluted plasma sample, ensuring that the metabolite concentration is suitable for mass spectrometry analysis. Specifically, 100 nL of the plasma sample can be diluted 10 times with deionized water to obtain a diluted plasma sample to adjust the metabolite concentration to be suitable for mass spectrometry analysis.

[0058] Step S124 includes: preparing inorganic nanoparticles into a nanoparticle matrix solution for enhancing mass spectrometry signals. Specifically, the matrix (inorganic nanoparticles) is prepared into a 1 mg / mL matrix solution with deionized water to ensure that the metabolites are well combined with the matrix to obtain a clear mass spectrometry signal.

[0059] Step S125 includes: performing sample preparation on the mass spectrometry target plate: spotting the diluted plasma sample onto the mass spectrometry target plate and drying it at room temperature.

[0060] Specifically, the sample preparation can be performed on a mass spectrometry target plate, and 500 nL of each diluted plasma sample is spotted and dried at room temperature.

[0061] Step S126 includes: performing matrix preparation on the mass spectrometry target plate: spotting the nanoparticle matrix solution on the mass spectrometry target plate, ensuring that the nanoparticle matrix solution evenly covers the plasma sample spot to obtain a sample. Specifically, the matrix preparation can be performed on the mass spectrometry target plate, and 500 nL of each matrix solution is spotted on each plasma sample spot and dried at room temperature.

[0062] Step S127 includes: using a nanoparticle enhanced laser desorption / ionization time-of-flight mass spectrometer (MALDI-TOF-MS) to collect data in a positive ion mode, using a 1000 Hz repetition rate, a 20 kV acceleration voltage, and a 250 ns delay time for mass spectrometry analysis, and recording 2000 laser emission results in each analysis to obtain a metabolite peak map in the sample.

[0063] Specifically, data acquisition was performed in a nanoparticle enhanced laser desorption / ionization time-of-flight mass spectrometer, and the acquired data were extracted in positive ion mode with delayed extraction, a repetition rate of 1000 Hz, an acceleration voltage of 20 kV, a delay time of 250 ns, and 2000 laser shots per analysis.

[0064] Step S128 includes: based on the metabolite peak graph in the sample, under the conventional standardized conditions of the software supporting the nanoparticle enhanced laser desorption / ionization time-of-flight mass spectrometer, running the built-in data preprocessing pipeline, including: spectral line smoothing, baseline correction, spectral peak matching and logarithmic transformation and Min-Max normalization, to obtain the relative intensity (%) of 303 metabolite biomarkers in plasma, i.e., the relative concentration value, such as Figure 1 as shown in the heat map.

[0065] Specifically, a Savitzky-Golay (SG) filter can be used for spectral smoothing; then an adaptive iterative algorithm is used to identify baseline points, a polynomial fit is performed on the baseline points identified in the metabolite peak graph, the fitted baseline is subtracted from the original spectrum, and the corrected spectrum is checked to ensure that the baseline is flat and has no negative values; then the peak detection parameters are set: signal-to-noise ratio threshold = 3, mass accuracy tolerance 50ppm, continuous wavelet transform (CWT) is used to identify the peak position of possible metabolite markers, and further peak fitting is performed to obtain the corresponding peak position, peak height and peak area, and the metabolite marker level is relatively quantified according to the peak height; finally, logarithmic transformation and Min-Max normalization are performed to obtain the relative concentration values ​​of 303 markers.

[0066] The relative concentration value obtained here can be the peak intensity value after logarithmic transformation and min-max normalization, in %, which is used to measure the relative concentration of metabolites in plasma.

[0067] By comparing the HMDB (Human Metabolome Database), 303 plasma small molecule metabolic biomarkers can be identified.

[0068] In one embodiment, the feature selection module is used to perform feature selection based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​obtained from each plasma sample, so as to screen out the plasma small molecule metabolic biomarkers for early identification of metabolically unhealthy obesity as the plasma small molecule metabolic biomarkers after feature selection.

[0069] First, feature selection module Lasso feature selection: Lasso feature selection algorithm and five-fold cross validation were used on R (4.3.1) software to perform feature selection on all plasma small molecule metabolic biomarkers in the first set, such as 303 plasma small molecule metabolic biomarkers. The implementation of Lasso feature selection algorithm depends on glmnet package, where glmnet package generates the corresponding hyperparameter lambda through built-in program.

[0070] A sequence of 100 values ​​that controls the intensity of the L1 regularization penalty on the model coefficients and determines the 50%

[0071] The lambda.min with the lowest error rate in cross-validation was finally calculated to be lambda.min = 0.02. 27 metabolite biomarkers were screened out based on the standard of minimizing the error rate. The 27 metabolite biomarkers showed good results in identifying metabolically unhealthy obesity. Specifically, the error rate is the lowest when lambda = 0.02, and the lambda at this time is called lambda.min.

[0072] Table 1: Information about 27 metabolic biomarkers

[0073]

[0074] As shown in Table 1, the 27 metabolite biomarkers include mercaptoethylamine, 4-methylpentanal, 3-hydroxybutyric acid, aminophenol, pyruvate, threonine, succinic anhydride, threitol, glycerol, hypotaurine, creatinine, adenine, valine, cysteine, isonicotinic acid, taurine, glutamate, methionine, citrulline, glucose, arginine, ascorbic acid, lipoic acid, acetylcarnitine, glycerophosphocholine, stearic acid and sphingosine.

[0075] Specifically, R software is a language and environment for statistical computing and graphics. It supports cross-platform support for Windows, Mac, Linux, etc. It is a GNU project, similar to the S language and environment developed by John Chambers and his colleagues at Bell Labs (formerly AT&T, now Lucent Technologies). R can be considered a different implementation of S. There are some important differences, but many codes written for S remain unchanged when running under R. R provides a wide variety of statistics (linear and nonlinear modeling, classical statistical tests, time series analysis, classification, clustering, etc.) and graphics techniques, and is highly extensible. The S language is often the tool of choice for research in statistical methods, and the R language provides an open source way to participate in that activity.

[0076] In one embodiment, the feature selection module Lasso feature selection may include the steps described below.

[0077] First, the feature selection module divides the first set into 5 non-overlapping data, and each time uses one of the 5 non-overlapping data that has not been selected as the internal validation set, and each time uses the remaining 4 data as the training set; wherein the internal validation set and the training set both include non-overlapping subjects of metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO). Here, the internal validation set includes subjects of metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO); the training set includes subjects of metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO); the internal validation set and the training set do not overlap. For example, the subjects are divided into the first set (a total of 234 cases, including 167 MHO cases and 167 UHO cases) and the test set (a total of 78 cases, including 39 MHO cases and 39 UHO cases).

[0078] Secondly, the feature selection module uses the Lasso feature selection algorithm on the R (4.3.1) software based on each training set, and cyclically trains the feature selection model for 5 rounds to obtain the corresponding 5 feature selection models.

[0079] Finally, for each value in the sequence of 100 values ​​corresponding to the hyperparameter lambda, the feature selection module inputs a corresponding feature selection model from the five feature selection models in each corresponding internal validation set, and obtains a misclassification rate corresponding to each value on each internal validation set; from the five misclassification rates corresponding to each value, an average misclassification rate corresponding to each value is obtained; based on the minimum average misclassification rate among all values, lambda.min with the lowest misclassification rate is selected, where lambda is one of the hyperparameters; based on lambda.min with the lowest misclassification rate, the corresponding 27 metabolite markers are obtained.

[0080] Here, based on the fact that lambda.min is the best hyperparameter lambda, the corresponding features are extracted to screen out 27 metabolite markers. The relevant information of the 27 metabolic biomarkers is shown in Table 1.

[0081] Specifically, the lambda hyperparameters of the five feature selection models are the same. For each lambda value, the lambda value is determined, the features used are determined, and the corresponding metabolite markers can be mapped based on these features. Five feature selection models can be obtained through the training set. By looking at the average performance of the lambda value in the internal validation set of the five models, it is determined which lambda value has the best average performance on the internal validation set, that is, the lowest average misclassification rate.

[0082] In one embodiment, the training module is used to train a final prediction model for early identification of metabolically unhealthy obesity based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample.

[0083] Specifically, the relative concentrations (%) of 27 metabolic biomarkers in the MHO group and after normalization in the MHO group are shown in Table 2 .

[0084] Table 2: Relative concentrations (%) of 27 metabolic biomarkers in the MHO group and after normalization in the MHO group

[0085]

[0086] The final prediction model for early identification of metabolically unhealthy obesity is trained based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample, including: first, the training module trains the prediction model for early identification of metabolically unhealthy obesity based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample; second, the training module performs performance evaluation on the trained prediction model for early identification of metabolically unhealthy obesity, and uses the prediction model for early identification of metabolically unhealthy obesity whose performance evaluation meets preset requirements as the final prediction model.

[0087] The training module trains a predictive model for early identification of metabolically unhealthy obesity based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample, including:

[0088] The training module uses the ElasticNet algorithm on the R (4.3.1) software and based on the first set, to train a prediction model for early identification of metabolically unhealthy obesity. The implementation of the ElasticNet algorithm relies on the glmnet package, wherein the ElasticNet algorithm hyperparameter alpha for controlling the ratio of L1 regularization and L2 regularization is set to 0.5; s for controlling the penalty intensity of the parameter is set to 0.01.

[0089] The training module performs performance evaluation on the trained prediction model for early identification of metabolically unhealthy obesity, and uses the prediction model for early identification of metabolically unhealthy obesity that meets the preset requirements as the final prediction model, including: the training module obtains the trained prediction model for early identification of metabolically unhealthy obesity on the R (4.3.1) software, and the recognition performance on the first set and the recognition performance on the test set respectively; if the recognition performance of the trained prediction model for early identification of metabolically unhealthy obesity in the training set and the test set differs by less than a preset difference threshold, indicating that the recognition performance of the two are close, then the trained prediction model for early identification of metabolically unhealthy obesity is used as the final prediction model.

[0090] Here, the training module uses the trained prediction model for early identification of metabolically unhealthy obesity on the R (4.3.1) software to make predictions on the first set and the test set, respectively, and obtains the recognition performance (recognition performance) on the training set and the test set, respectively. The recognition performance includes: the AUC values ​​are 0.930 for the training set and 0.884 for the test set, respectively, showing high accuracy and sensitivity. The specific performance indicators are shown in Table 2. If the recognition performance curves of the training set and the test set are close to each other (Delong test P = 0.26>0.05), as Figure 2 As shown, it means that the recognition performance is close, which means that the prediction model trained for early identification of metabolically unhealthy obesity is a better model and can be used as the final prediction model.

[0091] The metabolically unhealthy obesity recognition model provided by the present invention has the characteristics of low sample consumption and high reproducibility. By comparing the performance of the training set and the performance of the external test set, it is judged whether the model for metabolically unhealthy obesity recognition has the risk of overfitting. If the performance is close, it means that the model has universality. If it is not close, it means that the generalization ability of the model is relatively poor and a better model needs to be trained. The performance of machine learning is shown in Table 3.

[0092] Table 3. Performance of machine learning obtained on an independent test set of metabolic biomarkers

[0093]

[0094] The model prediction results are represented by a 2x2 confusion matrix, where rows represent actual labels and columns represent predicted labels. The confusion matrix divides the samples into four categories:

[0095] TP (True Positive): The number of samples that are actually positive examples and predicted to be positive examples.

[0096] TN (True Negative) True Negative Examples: The number of samples that are actually negative examples but predicted to be negative examples.

[0097] FP (False Positive): False positive examples: The number of samples that are actually negative examples but predicted to be positive examples.

[0098] FN (False Negative) False negative examples: The number of samples that are actually positive examples but predicted to be negative examples.

[0099] Accuracy: The proportion of samples predicted correctly in the total samples, calculated as (TP+FP) / (FN+TN+TP+TN).

[0100] F1 value: It is a comprehensive indicator of the proportion of samples predicted as positive examples among the actual positive examples and the proportion of correct predictions among the positive examples. The calculation formula is 2*P*R / (P+R), where P=TP / (TP+FP) and R=TP / (TP+FN).

[0101] Sensitivity: The proportion of samples predicted to be positive among samples that are actually positive. The calculation formula is TP / (TP+FN).

[0102] Specificity: The proportion of samples predicted to be negative among samples that are actually negative. The calculation formula is TN / (TN+FP).

[0103] PPV: Positive predictive value, that is, the proportion of people who are truly positive among those who test positive. The calculation formula is TP / (TP+FP).

[0104] NPV: Negative predictive value, that is, the proportion of people who test negative who are actually negative. The calculation formula is TN / (TN+FN).

[0105] AUC: The area under the ROC curve. The ROC curve is a curve composed of the true positive rate and the false positive rate under different thresholds. The AUC is not affected by the ratio of positive and negative samples. It reflects the overall performance of the model under different thresholds.

[0106] The range is 0 to 1. The larger the value, the better the overall performance of the model. It can be calculated by the following formula:

[0107]

[0108] Among them, TPR=TP / (TP+FN), FPR=FP / (TN+FP).

[0109] In one embodiment, the identification module is used to obtain the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the person to be tested, and input the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the person to be tested into the final prediction model to obtain a prediction result of whether the person to be tested belongs to metabolically unhealthy obesity.

[0110] Here, the method for obtaining the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested can refer to the method for obtaining in step 1.

[0111] In addition, experimental verification can also be performed to evaluate the ability of the model to reflect future cardiovascular risks: in the test set, for metabolically unhealthy obese (MUO) individuals identified by the final prediction model based on the aforementioned 27 metabolite biomarkers and metabolically healthy obese (MHO) individuals identified by the model, China-PAR estimated the 10-year cardiovascular event rate based on risk factors. Here, China-PAR is a tool recommended by the Joint Committee on Cardiovascular Disease Risk Assessment and Management Guidelines in the Chinese Cardiovascular Disease Risk Assessment and Management Guidelines (2019) for estimating the 10-year cardiovascular event rate, where cardiovascular events include fatal or non-fatal atherosclerotic cardiovascular disease and stroke, and the calculation is performed under the instructions of the China-PAR online tool (www.cvdrisk.com.cn). In the test set, the mean 10-year cardiovascular event rate of MHO predicted by the metabolomics of the final prediction model of the present invention was 5.91, and the standard deviation was 1.66; the mean 10-year cardiovascular event rate of MUO predicted by the metabolomics of the final prediction model of the present invention was 8.11, and the standard deviation was 1.57, and the difference between the two groups was significant (Mann-Whitney test P<0.001). Figure 3a and 3b As shown. In the test set, the 10-year cardiovascular event risk of MUO patients identified by the final prediction model of the present invention was estimated according to China-PAR, and the MUO patients identified by the model were compared with the MHO patients identified by the model. The analysis showed that the final prediction model of the present invention identified that the 10-year cardiovascular event risk of MUO patients was significantly increased.

[0112] In addition, the model can also be verified by evaluating the ability of potential metabolically unhealthy obesity: among the individuals with clinical evaluation of MHO in the test set, the metabolically unhealthy obesity (MUO) individuals and metabolically healthy obesity (MHO) individuals identified by the final prediction model based on the aforementioned 27 metabolite biomarkers, the 10-year cardiovascular event rates were estimated according to China-PAR. Among the individuals with clinical evaluation of MHO in the test set, the estimated 10-year cardiovascular event rate of MHO predicted by metabolomics was a mean of 5.60, with a standard deviation of 1.56; the estimated 10-year cardiovascular event rate of MUO predicted by metabolomics of the final prediction model of the present invention was a mean of 7.58, with a standard deviation of 1.48, and the difference between the two groups was significant (Mann-Whitney test P<0.001). MHO patients who are clinically asymptomatic may express a metabolomics pattern similar to that of MUO patients, have a higher 10-year cardiovascular event risk, and are thus predicted as potential MUO individuals by the final prediction model of the present invention. Figure 3a and 3c As shown. Through the analysis of increased cardiovascular risk, data analysis showed that among the MHO population without obvious clinical symptoms, the population predicted as potential MUO by the final model of the present invention had a significantly increased risk of cardiovascular events estimated by China-PAR in 10 years, suggesting the value of early identification based on the metabolomics model.

[0113] Figures 3a to 3c is a 10-year cardiovascular event risk comparison based on an independent test set in one embodiment of the present invention, Figure 3a The middle is the number of MHO and MUO in the independent test set based on clinical symptoms / metabolomics prediction; Figure 3b Middle is the comparison of 10-year cardiovascular event risk between MHO and MUO patients predicted by metabolomics in an independent test set; Figure 3c The middle ones are individuals who were judged as MHO based on clinical symptoms in the independent test set. Some of them were predicted to be MUO by the metabolomics model and had a higher 10-year cardiovascular risk, so they should be considered as potential MUO patients.

[0114] The present invention also provides a computer-readable storage medium having computer executable instructions stored thereon, wherein when the computer executable instructions are executed by a processor, the processor is caused to execute the following steps: step S1, a collection module collects plasma samples of each subject with metabolically healthy obesity and metabolically unhealthy obesity, and obtains plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample; step S2, a feature selection module performs feature selection based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​obtained from each plasma sample, so as to screen and obtain plasma small molecule metabolic biomarkers for early identification of metabolically unhealthy obesity; Step S3: the training module trains a final prediction model for early identification of metabolically unhealthy obesity based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample; Step S4: the recognition module obtains the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested, and inputs the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested into the final prediction model to obtain a prediction result of whether the subject to be tested belongs to metabolically unhealthy obesity. The plasma small molecule metabolic biomarkers after feature selection included mercaptoethylamine, 4-methylvaleraldehyde, 3-hydroxybutyrate, aminophenol, pyruvate, threonine, succinic anhydride, threitol, glycerol, hypotaurine, creatinine, adenine, valine, cysteine, isonicotinic acid, taurine, glutamate, methionine, citrulline, glucose, arginine, ascorbic acid, lipoic acid, acetylcarnitine, glycerophosphocholine, stearic acid and sphingosine.

[0115] The present invention also provides a detection kit for early identification of metabolically unhealthy obesity, wherein the detection kit protects metabolite biomarkers, and the metabolite biomarkers include at least the following substances: mercaptoethylamine, 4-methylvaleraldehyde, 3-hydroxybutyric acid, aminophenol, pyruvic acid, threonine, succinic anhydride, threitol, glycerol, hypotaurine, creatinine, adenine, valine, cysteine, isonicotinic acid, taurine, glutamic acid, methionine, citrulline, glucose, arginine, ascorbic acid, lipoic acid, acetylcarnitine, glycerophosphocholine, stearic acid and sphingosine.

[0116] In summary, the identification model construction system for metabolically unhealthy obesity disclosed in the present invention uses high-throughput metabolite detection technology and machine learning algorithms to provide an innovative and effective solution for the early diagnosis of metabolically unhealthy obesity (MUO). Metabolomics is an advanced omics technology that can perform global analysis of metabolite small molecules in biological samples with high throughput. Before the disease phenotype appears, metabolomics can systematically detect metabolic disorders in the body, thus showing broad prospects in the diagnosis of diseases. Plasma, as an easily accessible biological fluid, contains rich metabolite information and can reflect the dynamic changes of various metabolic pathways in the body. Plasma metabolites are considered to be early markers of diseases such as cardiovascular diseases and lipid metabolism disorders. The present invention systematically analyzes plasma metabolites and selects specific metabolic markers related to metabolically unhealthy obesity, thereby establishing a set of effective tools for early identification of MUO.

[0117] The present invention exhibits excellent risk assessment capabilities. Through machine learning of plasma metabolite biomarkers, it is possible to identify metabolically unhealthy obese patients early before clinical symptoms appear, helping doctors and patients take intervention measures before cardiovascular risks increase. The identification model construction system for metabolically unhealthy obesity provided by the present invention is based on a small amount of sampling of plasma metabolites, which is simpler than the traditional physical examination process, reduces the burden of patient participation, and is suitable for wide clinical promotion. Through the combination of machine learning models and high-throughput mass spectrometry technology, high sensitivity (87.18%) and specificity (71.79%) identification of MUO can be achieved, ensuring its reliability and accuracy in clinical applications. The identification model construction system for metabolically unhealthy obesity provided by the present invention provides a scientific basis for individualized health management, helps to customize personalized intervention measures, and is of great significance, especially in the prevention of cardiovascular diseases.

[0118] The metabolically unhealthy obesity identification model construction system of the present invention is used for the detection and prevention of metabolically unhealthy obesity (MUO). By identifying MUO patients, their potential cardiovascular disease risk can be accurately judged even when obvious clinical symptoms (such as hypertension, elevated blood sugar or dyslipidemia) have not yet appeared. This system has excellent sensitivity, specificity and accuracy, and the detection process is simple and rapid. It is suitable for large-scale clinical screening and intervention, and brings great benefits to the prevention and early intervention of cardiovascular diseases. The present invention can efficiently distinguish metabolically healthy obesity (MHO) from metabolically unhealthy obesity (MUO) through metabolite data in plasma, and provide a scientific basis for individualized health management, especially in the early prevention of cardiovascular events. It has important application value.

[0119] In addition, the present invention discloses a metabolite composition for early identification of metabolically unhealthy obesity (MUO), and these metabolites can be used to develop a kit to provide a convenient and rapid detection tool to provide support for clinical diagnosis and preventive measures.

[0120] The detailed contents of the embodiments of each device of the present invention can be found in the corresponding parts of each system embodiment, which will not be described in detail here.

[0121] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

[0122] It should be noted that the present invention can be implemented in software and / or a combination of software and hardware, for example, it can be implemented using an application-specific integrated circuit (ASIC), a general-purpose computer or any other similar hardware device. In one embodiment, the software program of the present invention can be executed by a processor to implement the steps or functions described above. Similarly, the software program of the present invention (including related data structures) can be stored in a computer-readable recording medium, for example, a RAM memory, a magnetic or optical drive or a floppy disk and the like. In addition, some steps or functions of the present invention can be implemented using hardware, for example, as a circuit that cooperates with a processor to perform various steps or functions.

[0123] In addition, a part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. The program instruction for calling the method of the present invention may be stored in a fixed or removable recording medium, and / or transmitted through a data stream in a broadcast or other signal-bearing medium, and / or stored in a working memory of a computer device that runs according to the program instruction. Here, according to an embodiment of the present invention, a device is included, the device including a memory for storing computer program instructions and a processor for executing program instructions, wherein, when the computer program instruction is executed by the processor, the device is triggered to run the method and / or technical solution based on the aforementioned multiple embodiments of the present invention.

[0124] It is obvious to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the device claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

Claims

1. A system for building a recognition model of metabolically unhealthy obesity, characterized in that: include: A collection module, used for collecting plasma samples from subjects with metabolically healthy obesity and metabolically unhealthy obesity, and obtaining plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample; A feature selection module is used to obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample, perform feature selection, and screen out plasma small molecule metabolic biomarkers for early identification of metabolically unhealthy obesity as plasma small molecule metabolic biomarkers after feature selection; A training module, used to train a final prediction model for early identification of metabolically unhealthy obesity based on plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample; An identification module is used to obtain the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested, and input the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested into the final prediction model to obtain a prediction result of whether the subject to be tested belongs to metabolically unhealthy obesity; The plasma small molecule metabolic biomarkers after feature selection include: Mercaptoethylamine, 4-methylvaleraldehyde, 3-hydroxybutyric acid, aminophenol, pyruvic acid, threonine, succinic anhydride, threitol, glycerol, hypotaurine, creatinine, adenine, valine, cysteine, isonicotinic acid, taurine, glutamic acid, methionine, citrulline, glucose, arginine, ascorbic acid, lipoic acid, acetylcarnitine, glycerophosphocholine, stearic acid, and sphingosine.

2. The metabolically unhealthy obesity identification model construction system according to claim 1, characterized in that: The acquisition module is used to obtain a first subject group and a second subject group that do not overlap each other, wherein the first subject group and the second subject group both include subjects with metabolically healthy obesity and metabolically unhealthy obesity that do not overlap each other; The collection module is used to collect first plasma samples from the first subject group, and obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each first plasma sample as a first set; The collection module is used to collect second plasma samples from the second subject group, and obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each second plasma sample as a test set.

3. The identification model construction system of metabolically unhealthy obesity according to claim 2, characterized in that: The feature selection module is used to perform feature selection on all plasma small molecule metabolic biomarkers in the first set using the Lasso feature selection algorithm and five-fold cross validation on the R software. The implementation of the Lasso feature selection algorithm depends on the glmnet package, wherein the glmnet package generates a sequence of 100 values ​​corresponding to the hyperparameter lambda through a built-in program, which is used to control the intensity of the L1 regularization penalty on the model coefficients and determine the lambda.min when the error rate is the lowest in the five-fold cross validation. The lambda.min is finally calculated, and the metabolite biomarkers are screened out based on the lambda.min standard.

4. The identification model construction system of metabolically unhealthy obesity according to claim 3, characterized in that: The feature selection module is used to divide the first set into 5 non-overlapping data, each time using one of the 5 non-overlapping data that has not been selected as an internal validation set, and each time using the remaining 4 data as training sets; wherein the internal validation set and the training set both include non-overlapping subjects with metabolically healthy obesity and metabolically unhealthy obesity; The feature selection module is used to use the Lasso feature selection algorithm on the R software based on each training set to cyclically train the feature selection model for 5 rounds to obtain corresponding 5 feature selection models; The feature selection module is used for inputting a corresponding feature selection model of the five feature selection models into each internal validation set corresponding to each value in a sequence of 100 values ​​corresponding to the hyperparameter lambda, and obtaining a misclassification rate corresponding to each value on each internal validation set; obtaining an average misclassification rate corresponding to each value from the five misclassification rates corresponding to each value; selecting lambda.min with the lowest corresponding misclassification rate based on the minimum average misclassification rate among all values, wherein lambda is one of the hyperparameters; obtaining a corresponding metabolite marker based on lambda.min with the lowest misclassification rate as a plasma small molecule metabolic biomarker after feature selection.

5. The metabolically unhealthy obesity identification model construction system according to claim 2, characterized in that: The training module is used to train a prediction model for early identification of metabolically unhealthy obesity based on plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample; the trained prediction model for early identification of metabolically unhealthy obesity is performance evaluated, and the prediction model for early identification of metabolically unhealthy obesity whose performance evaluation meets preset requirements is used as the final prediction model.

6. The metabolically unhealthy obesity identification model construction system according to claim 5, characterized in that: The training module is used to train a prediction model for early identification of metabolically unhealthy obesity using an elastic network algorithm on R software and based on the first set. The implementation of the elastic network algorithm relies on the glmnet package, wherein the ElasticNet algorithm hyperparameter alpha for controlling the ratio of L1 regularization and L2 regularization is set to 0.5; and s for controlling the penalty intensity of the parameter is set to 0.

01.

7. The metabolically unhealthy obesity identification model construction system according to claim 6, characterized in that: The training module is used to obtain the recognition performance of the trained prediction model for early identification of metabolically unhealthy obesity on the first set and the recognition performance on the test set on the R software; if the recognition performance of the trained prediction model for early identification of metabolically unhealthy obesity differs between the training set and the test set by less than a preset difference threshold, the trained prediction model for early identification of metabolically unhealthy obesity is used as the final prediction model.

8. A computer-readable storage medium having computer-executable instructions stored thereon, wherein: When the computer executable instructions are executed by a processor, the processor performs the following steps: The collection module collects plasma samples from subjects with metabolically healthy obesity and metabolically unhealthy obesity, and obtains plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample; The feature selection module obtains plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample, performs feature selection, and screens out plasma small molecule metabolic biomarkers for early identification of metabolically unhealthy obesity as plasma small molecule metabolic biomarkers after feature selection; The training module trains the final prediction model for early identification of metabolically unhealthy obesity based on the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample; The recognition module obtains the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested, and inputs the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested into the final prediction model to obtain a prediction result of whether the subject to be tested belongs to metabolically unhealthy obesity; Plasma small molecule metabolic biomarkers after feature selection, including: Mercaptoethylamine, 4-methylvaleraldehyde, 3-hydroxybutyric acid, aminophenol, pyruvic acid, threonine, succinic anhydride, threitol, glycerol, hypotaurine, creatinine, adenine, valine, cysteine, isonicotinic acid, taurine, glutamic acid, methionine, citrulline, glucose, arginine, ascorbic acid, lipoic acid, acetylcarnitine, glycerophosphocholine, stearic acid, and sphingosine.

9. A detection kit for early identification of metabolically unhealthy obesity, characterized in that: The detection kit contains metabolite biomarkers, which at least include the following substances: Mercaptoethylamine, 4-methylvaleraldehyde, 3-hydroxybutyric acid, aminophenol, pyruvic acid, threonine, succinic anhydride, threitol, glycerol, hypotaurine, creatinine, adenine, valine, cysteine, isonicotinic acid, taurine, glutamic acid, methionine, citrulline, glucose, arginine, ascorbic acid, lipoic acid, acetylcarnitine, glycerophosphocholine, stearic acid, and sphingosine; The metabolite biomarkers are feature selected and identified through the following modules: A collection module, used for collecting plasma samples from subjects with metabolically healthy obesity and metabolically unhealthy obesity, and obtaining plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample; A feature selection module is used to obtain plasma small molecule metabolic biomarkers and their relative concentration values ​​from each plasma sample, perform feature selection, and screen out plasma small molecule metabolic biomarkers for early identification of metabolically unhealthy obesity as plasma small molecule metabolic biomarkers after feature selection; A training module, used to train a final prediction model for early identification of metabolically unhealthy obesity based on plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection in each plasma sample; The identification module is used to obtain the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested, and input the plasma small molecule metabolic biomarkers and their relative concentration values ​​after feature selection of the subject to be tested into the final prediction model to obtain a prediction result of whether the subject to be tested belongs to metabolically unhealthy obesity.

Images