Metabolic marker for auxiliary diagnosis of schizophrenia, construction method and application

Abstract

The invention provides a metabolic marker for auxiliary diagnosis of schizophrenia, a construction method and application. Wherein the metabolic marker comprises lysophosphatidylcholine (LPC) (17: 0 / 0: 0), oleamide, lysophosphatidyl ethanolamine LysoPE (18: 0 / 0: 0), palmitamide, lysophosphatidyl ethanolamine LysoPE (0: 0 / 20: 4), lysophosphatidyl ethanolamine LysoPE (22: 6 / 0: 0), p-methylphenol sulfate, lactic acid, 3-hydroxybutyric acid, lysophosphatidyl ethanolamine LysoPE (18: 2 / 0: 0), lysophosphatidyl ethanolamine LysoPE (16: 0 / 0: 0) and sphingosine 1-phosphate. By utilizing the metabolic marker, auxiliary diagnosis of schizophrenia can be rapidly and accurately completed in the early stage of disease attack.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of biomedical technology, and relates to the application of metabolic markers in disease diagnosis, in particular to metabolic markers for assisting in the diagnosis of schizophrenia, a construction method and application. BACKGROUND

[0002] Schizophrenia is a chronic and severe mental illness, with a lifetime risk of about 1%. Schizophrenia usually occurs in adolescence and early adulthood, and is mainly characterized by hallucinations, delusions, thought disorders and cognitive disorders. Schizophrenia patients are prone to self-injury and suicide, with a risk of 5% to 10% of suicide death in their lifetime. In addition, due to poor dietary structure and lifestyle, schizophrenia patients have a significantly increased risk of cardiovascular disease, type 2 diabetes and metabolic syndrome, resulting in a 15-year shorter average life expectancy than the general population.

[0003] The onset of schizophrenia is affected by genetic and environmental factors. Due to the heterogeneity of the etiology, the physiological and pathological mechanisms of the disease are not yet clear. The current diagnostic criteria for schizophrenia is still based on clinical symptoms, relying on psychiatrists to make judgments through interviews and observations of patients' behavior. The lack of objective and accurate diagnostic indicators often leads to missed diagnosis and misdiagnosis, which is not conducive to early screening.

[0004] Metabolome is the dynamic whole of endogenous metabolites in the body. By detecting changes in endogenous metabolites in the human body, the dynamic multi-parameter metabolic response of the disease can be monitored. Metabolomics uses techniques such as nuclear magnetic resonance and mass spectrometry to predict major diseases or conduct in-depth research on the mechanisms of major diseases, in order to provide scientific basis for the prevention and treatment of diseases. However, the field still faces challenges in metabolite sensitivity, data complexity and omics data integration.

[0005] Machine learning algorithms have unique advantages in processing high-dimensional data sets, can effectively mine biological information, and truly reflect biochemical processes under physiological and pathological conditions. It is a new type of technical means for efficient processing of high-throughput big data. In recent years, machine learning has been applied to data analysis and disease classification in metabolomics, effectively improving the accuracy of metabolite identification, the efficiency of data analysis and the accuracy of diagnostic methods.

[0006] Therefore, how to use metabolomics combined with machine learning to screen metabolic markers, effectively improve the accuracy and efficiency of the diagnostic method, and realize the auxiliary diagnosis of schizophrenia, is a technical problem that needs to be solved by those skilled in the art. SUMMARY

[0007] To solve the above problems, the application provides a set of metabolic markers for assisting in diagnosing schizophrenia.

[0008] The first object of the application provides a set of metabolic markers for assisting in diagnosing schizophrenia, which comprises lysophosphatidylcholine LPC (17:0 / 0:0), oleamide, lysophosphatidylethanolamine LysoPE (18:0 / 0:0), palmitamide, lysophosphatidylethanolamine LysoPE (0:0 / 20:4), lysophosphatidylethanolamine LysoPE (22:6 / 0:0), p-methylphenol sulfate, lactic acid, 3-hydroxybutyric acid, lysophosphatidylethanolamine LysoPE (18:2 / 0:0), lysophosphatidylethanolamine LysoPE (16:0 / 0:0) and sphingosine-1-phosphate.

[0009] The second aspect of the application also provides a prediction model for assisting in diagnosing schizophrenia, which comprises the above metabolic markers.

[0010] In some embodiments, the area under the curve of the prediction model for assisting in diagnosing schizophrenia is 0.976923.

[0011] The third aspect of the application also provides a kit for assisting in diagnosing schizophrenia, which comprises the above metabolic markers.

[0012] In some embodiments, the kit further comprises standard samples of ketoprofen and L-2-chlorophenylalanine.

[0013] The fourth aspect of the application also provides a screening method of metabolic markers for assisting in diagnosing schizophrenia, which comprises:

[0014] S1, collecting and processing plasma samples to obtain a data set;

[0015] S2, performing ultra-high performance liquid chromatography-quadrupole / orbitrap tandem high-resolution mass spectrometry analysis on the plasma samples;

[0016] S3, using Thermo Xcalibur software to perform data acquisition and processing, using Compound Discovery software to extract molecular formula, mass-to-charge ratio and secondary ion fragment information by using retention time information;

[0017] S4, introducing the information extracted in S3 into SIMCA for multivariate statistical analysis, establishing an orthogonal partial least squares discriminant analysis model, screening differential metabolites by variable importance projection index, including P value, vip value and differential fold FC; screening VIP>1, P<0.05 and differential metabolites meeting differential fold FC>1.25 or <0.8 as the candidate data set for selecting metabolic markers;

[0018] S5, screening biomarkers by integrated machine learning method from the candidate data set;

[0019] S6, splitting the data set into training set and test set, training 6 machine learning models one by one by grid search with cross-validation, and screening the final prediction model with optimal prediction ability from the differential metabolites.

[0020] In some embodiments of the screening method of metabolic markers for assisting the diagnosis of schizophrenia, the machine learning method includes extreme gradient boosting algorithm, random forest, least absolute shrinkage and selection operator algorithm, ridge regression and linear support vector machine.

[0021] In some embodiments of the screening method of metabolic markers for assisting the diagnosis of schizophrenia, the specific steps of screening biomarkers by integrated machine learning method are:

[0022] Each algorithm performs ten thousand independent random grouping experiments, and the feature factors ranked in the top ten in each experiment are counted; the algorithms include extreme gradient boosting algorithm, random forest, least absolute shrinkage, selection operator algorithm and ridge regression;

[0023] According to the contribution of each feature factor to the prediction accuracy, the importance ranking is performed; the voting strategy is adopted to vote the feature factors ranked in the top ten in each algorithm; the voting results of ten thousand experiments are accumulated, and the feature factors ranked in the top ten in each algorithm are selected;

[0024] Combined with linear support vector machine algorithm, the coefficients of irrelevant feature factors are reduced to 0 by regularization method;

[0025] The feature factors with coefficient 0 are deleted, and the feature factors screened in each experiment are counted;

[0026] The feature factors screened in each experiment are voted, and the voting results of ten thousand experiments are accumulated to select the feature factors ranked in the top ten;

[0027] The intersection of the voting results of the above algorithm is selected as the metabolic marker for the diagnosis of schizophrenia.

[0028] In some embodiments, the six machine learning models in the screening method of the metabolic marker for the auxiliary diagnosis of schizophrenia include decision tree, random forest, support vector machine, k-nearest neighbor, logistic regression and Gaussian naive Bayes.

[0029] In some embodiments, the screening method of the metabolic marker for the auxiliary diagnosis of schizophrenia, the machine learning model compares the prediction ability of each model by accuracy, sensitivity, specificity, precision, recall, Matthew correlation coefficient and area under ROC curve AUC, and selects the model with the optimal prediction ability as the final prediction model.

[0030] Beneficial effects:

[0031] The metabolic marker for the auxiliary diagnosis of schizophrenia, the construction method and the application provided by the application provide a group of metabolic markers, and the auxiliary diagnosis of schizophrenia can be quickly and accurately completed in the early stage of the disease by using the metabolic markers, the accuracy of the auxiliary diagnosis reaches 0.934783, and the AUC value reaches 0.976923. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 The receiver operating characteristic curve (ROC) analysis of the logistic regression model for the diagnosis of schizophrenia in the validation set.

[0033] Figure 2 The principal component analysis (PCA) score plot of the positive ion mode of the schizophrenia patient group (Schizophrenia, SP) vs. the healthy control group (Control, C) in the example;

[0034] Figure 3 The principal component analysis (PCA) score plot of the negative ion mode of the schizophrenia patient group (Schizophrenia, SP) vs. the healthy control group (Control, C) in the example;

[0035] Figure 4 The orthogonal partial least squares discriminant analysis (OPLS-DA) score plot of the positive ion mode of the schizophrenia patient group (Schizophrenia, SP) vs. the healthy control group (Control, C) in the example;

[0036] Figure 5 The orthogonal partial least squares discriminant analysis (OPLS-DA) score plot of the negative ion mode of the schizophrenia patient group (Schizophrenia, SP) vs. the healthy control group (Control, C) in the example;

[0037] Figure 6 Permutation plot for the OPLS-DA model in positive ion mode for Schizophrenia (SP) vs. Control (C) in the Example;

[0038] Figure 7 Permutation plot for the OPLS-DA model in negative ion mode for Schizophrenia (SP) vs. Control (C) in the Example;

[0039] Figure 8 Feature importance ranking for the Extreme Gradient Boosting (XGBoost) in the Example;

[0040] Figure 9 Feature importance ranking for the random forest (RF) in the Example;

[0041] Figure 10 Feature importance ranking for the least absolute shrinkage and selection operator (LASSO) in the Example;

[0042] Figure 11 Feature importance ranking for the Ridge Regression in the Example;

[0043] Figure 12 Receiver operating characteristic (ROC) analysis for schizophrenia diagnosis on the test set using six machine learning models: decision tree (DT), random forest (RF), support vector machine (SVM), k-nearest neighbor (kNN), logit regression (LR), and Gaussian Naive Bayesian (GNB) in the Example. DETAILED DESCRIPTION

[0044] Various exemplary embodiments of the present application will now be described in detail with reference to the drawings. The detailed description is not to be considered to limit the application to the specific aspects, features and embodiments described, but rather to be considered to be illustrative of certain aspects, features and embodiments of the application.

[0045] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. Additionally, for a range of values of, for example, concentrations, solvent amounts, and the like, there can be an intended distinction between the use of the term "comprises" or "comprising" and the term "consists" or "consisting." It is intended that the term "consists" or "consisting" be used to exclude any integer of the listed steps, elements, or components, but to otherwise permit the inclusion of any other element or component. It is further intended that the use of the term "comprises" or "comprising" be used to include the listed elements or components, but to also permit the inclusion of any other element or component.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present application, the preferred methods and materials are described. All publications mentioned herein are incorporated by reference to disclose and describe the methods and / or materials in connection with which the publications are cited. The citation of any reference is not an admission that it is prior art with respect to the present application.

[0047] With respect to the terms "comprising", "containing", "having", "including", and the like used herein, these terms are used in their open-ended, non-limiting sense to mean that the described feature is present, but not excluding the presence of other features. With respect to the term "and / or", it is used in its plain, ordinary sense to include any and all combinations of one or more of the associated listed items. Unless otherwise noted, % means percent by mass.

[0048] As shown in Figures 1 to 12 The first aspect of the present application provides a set of metabolic markers for the auxiliary diagnosis of schizophrenia, the metabolic markers comprising lysophosphatidylcholine LPC (17:0 / 0:0), oleamide, lysophosphatidylethanolamine LysoPE (18:0 / 0:0), palmitamide, lysophosphatidylethanolamine LysoPE (0:0 / 20:4), lysophosphatidylethanolamine LysoPE (22:6 / 0:0), p-cresol sulfate, lactic acid, 3-hydroxybutyric acid, lysophosphatidylethanolamine LysoPE (18:2 / 0:0), lysophosphatidylethanolamine LysoPE (16:0 / 0:0), and sphingosine-1 -phosphate.

[0049] The second aspect of the present application provides a prediction model for the auxiliary diagnosis of schizophrenia, which comprises the above-mentioned metabolic markers.

[0050] The prediction model has an area under the curve of 0.976923.

[0051] The third aspect of the present application also provides a kit for the auxiliary diagnosis of schizophrenia, which comprises the above-mentioned metabolic markers.

[0052] In the scheme, the kit further comprises standard samples of ketoprofen and L-2-chlorophenylalanine

[0053] The fourth aspect of the present application also provides a screening method of metabolic markers applied to auxiliary diagnosis of schizophrenia, comprising:

[0054] S1, collecting and processing plasma samples to obtain a data set;

[0055] S2, performing ultra-high performance liquid chromatography-quadrupole / orbitrap tandem high-resolution mass spectrometry analysis on the plasma samples;

[0056] S3, using Thermo Xcalibur software to perform data acquisition and processing, using Compound Discovery software to extract molecular formula, mass-to-charge ratio, and secondary ion fragment information using retention time information;

[0057] S4, importing the information extracted in S3 into SIMCA for multivariate statistical analysis, establishing an orthogonal partial least squares discriminant analysis model, and screening differential metabolites using a variable importance projection index as a screening index, wherein the variable importance projection index includes P value, vip value, and difference fold FC; the differential metabolites screened with VIP>1, P<0.05, and meeting the difference fold FC>1.25 or <0.8 are used as a candidate data set for selecting metabolic markers;

[0058] S5, screening biomarkers using an integrated machine learning method, and screening differential metabolic markers from the candidate data set;

[0059] S6, splitting the data set into a training set and a test set, training six machine learning models one by one using a grid search with cross-validation, and screening a final prediction model with optimal prediction ability from the differential metabolic markers.

[0060] In the scheme, the screening method of metabolic markers applied to auxiliary diagnosis of schizophrenia, wherein the machine learning method includes extreme gradient boosting algorithm, random forest, least absolute shrinkage and selection operator algorithm, ridge regression, and linear support vector machine.

[0061] In the scheme, the screening method of metabolic markers applied to auxiliary diagnosis of schizophrenia, wherein the specific steps of screening biomarkers using an integrated machine learning method are as follows:

[0062] Each algorithm performs ten thousand independent random grouping experiments, and the feature factors ranked in the top ten in each experiment are counted; the algorithms include extreme gradient boosting algorithm, random forest, least absolute shrinkage, selection operator algorithm, and ridge regression.

[0063] According to the contribution of each characteristic factor to the prediction accuracy, the importance is ranked; a voting strategy is adopted to vote the characteristic factors ranked in the top ten in each algorithm; the voting results of ten thousand experiments are accumulated to select the characteristic factors ranked in the top ten in each algorithm;

[0064] In combination with the linear support vector machine algorithm, the coefficients of irrelevant characteristic factors are reduced to 0 through the regularization method;

[0065] The characteristic factors with the coefficients of 0 are deleted, and the characteristic factors obtained in each experiment are counted;

[0066] The characteristic factors obtained in each experiment are voted, and the voting results of ten thousand experiments are accumulated to select the characteristic factors ranked in the top ten;

[0067] The intersection of the voting results of the above algorithms is selected as the metabolic marker for the diagnosis of schizophrenia.

[0068] In the above scheme, the six machine learning models include decision tree, random forest, support vector machine, k-nearest neighbor, logistic regression and Gaussian naive Bayes.

[0069] In the above scheme, the machine learning model compares the prediction ability of each model in terms of accuracy, sensitivity, specificity, precision, recall, Matthew correlation coefficient and area under ROC curve AUC, and selects the model with the optimal prediction ability as the final prediction model.

[0070] Embodiment

[0071] In order to verify the metabolic marker described in the application, which can quickly and accurately assist in the early diagnosis of schizophrenia, the following experiments are carried out.

[0072] I. Experimental instruments and reagents

[0073] Ultimate 3000 ultra-high performance liquid chromatography system (Dionex Company, USA) connected with Q-Exactive quadrupole-electric field orbitrap high-resolution mass spectrometer (Thermo Fisher Scientific Company, USA); the chromatographic column adopts ACQUITY HSS T3 1.8 pm (2.1 x 100 mm) (Waters, Milford, MA, USA); Heraeus Fresco 17 centrifuge (Thermo Fisher Scientific).

[0074] HPLC grade methanol, acetonitrile, formic acid were purchased from Thermo Fisher Scientific (USA); L-2-chloro-phenylalanine, ketoprofen standard were purchased from J&K Chemical (Beijing, China); Experimental water was Wahaha purified water;

[0075] II. Experimental Methods

[0076] 1. Experimental samples

[0077] After obtaining the consent of the patients, patients who met the diagnostic criteria for schizophrenia in the Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition, DSM-5) were included, a total of 116 patients with schizophrenia admitted to the inpatient department of Zhumadian City Second People's Hospital from May 2023 to August 2023 were collected. In addition, 110 healthy volunteers were collected by the physical examination department as the normal control group. All blood samples were collected in the morning on an empty stomach, 3 mL of venous blood was collected from each research object and placed in an EDTA anticoagulant blood collection tube, and after collection, it was centrifuged at 3000 r / min for 10 min at 4°C, the supernatant (plasma) was aspirated, and after aliquoting, it was immediately frozen and stored in a -80°C refrigerator.

[0078] 2. Preparation of plasma samples

[0079] After the plasma sample was taken out and placed on ice for thawing, 50 pL of plasma sample was aspirated into a 1.5 mL centrifuge tube after vortexing, 150 pL of methanol solution containing internal standard (containing ketoprofen 500 ng / mL, L-2-chloro-phenylalanine 50 ng / mL) was added, vortexed for 30 seconds, centrifuged at 13000 rpm for 10 min (4°C), and the supernatant was aspirated into a sample vial, and the sample was obtained. From all plasma samples, 10 pL was vortexed and mixed, and QC sample pretreatment was performed according to the same method as above, and the sample was obtained. In order to ensure the reliability of the data, QC samples were analyzed during the whole process of metabolomics data collection of all samples, 5 QC samples were detected before sample analysis, and after the instrument was stabilized, the sample analysis began, and QC solution was detected every 5 samples.

[0080] 3. Detection method of metabolites in plasma

[0081] Chromatographic separation was performed on an ultra-performance liquid chromatography (UPLC, Waters Ultimate 3000, USA) with an ACQUITY UPLC BEH C18 column (50 x 2.1 mm, 1.7 μm) at 40 °C. The mobile phase consisted of A phase of acetonitrile and B phase of 0.1% formic acid in water. The gradient elution conditions were as follows: 0-0.5 min, 5% A; 0.5-1.0 min, 5%-60% A; 0.5-1.0 min, 5%-60% A; 1.0-7.0 min, 60%-80% A; 7.0-9.0 min, 80%-100% A; 9.0-11.0 min, 100% A; 11.0-13.0 min, 5% A; the flow rate was 0.2 mL / min, and the effluent was directly introduced into the mass spectrometer without splitting for detection. Mass spectrometric analysis was performed on a Q Exactive quadrupole-Orbitrap mass spectrometer with a heated electrospray ion source (HESI). The auxiliary gas temperature was 300 °C, the ion source temperature was 350 °C, the capillary temperature was 320 °C, the auxiliary gas flow rate was 10 uL / min, the mass spectrometric analysis scan mode was full scan / ddms2, the first mass spectrometric resolution was 70000, the second mass spectrometric resolution was 17500, the mass-to-charge window width was set to 8, the collision energy gradient was 20, 40 and 60 eV, and the scanning range was 80.00-1200.00 m / z in positive and negative ion modes, the spray voltage and the sheath gas flow rate were 3.50 kV, 40 μL / min and 2.80 kV, 38 μL / min, respectively. All samples were randomly injected.

[0082] 4. Data processing and analysis

[0083] All data were collected and processed by Thermo Xcalibur software (version 3.0, Thermo Scientific, USA), and then Compound Discovery software (version 3.0, Thermo Scientific, USA) was used for peak correction, peak matching and information extraction of peak alignment. The specific parameters are as follows: the retention time (RT) width is set to 0.1 min, the mass width is 5ppm, the intensity threshold for filtering ion peaks is 1,000,000 in both positive and negative ion modes. The data set was imported into the multivariate statistical analysis software SIMCA (version 14.1, Umetrics, Sweden) for orthogonal partial least squares discrimination analysis (OPLS-DA), and 200 permutation tests were performed to assess the risk of model overfitting. By establishing the OPLS-DA model, the variable importance ranking (VIP value) was obtained, and the metabolites with VIP>1 were subjected to independent sample t test to screen out the differential metabolites with statistical significance (P<0.05), and the information such as the first accurate m / z, the secondary fragment ion spectrum, the retention time (RT) collected by mass spectrometry was compared and identified with the database, including ChemSpider, MassList with first mass spectrum information, mzCloud, laboratory self-built database mzVault, Human Metabolome database (HMBD, http: / / hmdb.ca / ), MassBank (https: / / massbank.eu / MassBank / ), PubChem compound database (https: / / pubchem.ncbi.nlm.nih.gov / ), and finally the compound structure was confirmed by comparison with the standard.

[0084] VIP > 1, P < 0.05 and meet FC > 1.25 or < 0.8 as a data set for selecting metabolic markers, and the biomarker screening is carried out by using an integrated method, which makes full use of the advantages of individual technologies, overcomes its limitations, and improves the reliability. Combined with five algorithms, including: Extreme Gradient Boosting (XGBoost), random forest (RF), least absolute shrinkage and selection operator (LASSO), Ridge Regression and Linear Support Vector Classification (LinearSVC), for feature selection and dimensionality reduction. Further, based on the feature factors obtained by each algorithm, a voting strategy is applied for final prediction. The voting process relies on the combined performance of the five algorithms, thereby avoiding large errors or false classifications that may be caused by a single model. Finally, according to the voting results, the metabolic markers for diagnosing schizophrenia are screened.

[0085] The schizophrenia diagnosis metabolic markers of the above step are characterized, after obtaining the consent of the patient, the inpatients meeting the Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5) schizophrenia diagnostic criteria are included, and 30 cases of schizophrenia patients in the inpatient department of Zhumadian City Second People's Hospital from May 2023 to August 2023 are collected, and 30 cases of healthy volunteers collected by the physical examination department are collected as normal population as a validation set. The constructed logistic regression diagnostic model is used for performance test, and through accuracy, sensitivity, specificity, precision, recall rate, Matthews correlation coefficient (MCC) and AUC (as shown in Table 1), it can be found that the accuracy of the diagnostic model in the validation set reaches 0.966667, and the AUC value reaches 0.994444, as shown in Figure 1 The verification result shows that the model has good prediction ability, which indicates that the model can be used for early diagnosis of schizophrenia.

[0086] Table 1 Performance index of validation set logistic regression (logit regression, LR) learning model

[0087]

[0088] III. Experimental results

[0089] Principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) (e.g.) Figures 2 to 7 , Figures 2 to 7 (A represents the positive ion mode, and B represents the negative ion mode) It can be seen that the schizophrenia group (SP) and the normal control group (Control, C) composed of healthy individuals are clearly separated and cluster separately, indicating a significant difference in endogenous metabolites between schizophrenia patients and healthy individuals. The model evaluation parameters of OPLS-DA show that the model has good explanatory and predictive capabilities. The Response Ranking Test (RPT) method was used for permutation validation after 200 modeling iterations, confirming that the model is stable and reliable, and there is no overfitting.

[0090] Differential metabolites with VIP > 1, P < 0.05, and FC > 1.25 or < 0.8 were used as the dataset for selecting metabolic biomarkers. Four machine learning methods—XGBoost, RF, LASSO, and Ridge Regression—were used to select the top ten predictors based on feature importance scores (see [link to dataset]). Figures 8 to 11 Simultaneously, the LinearSVC algorithm was combined with different feature selection strategies. Regularization was used to minimize or eliminate the coefficients of irrelevant feature factors. Then, based on the feature factors selected by these different algorithms, a voting strategy was applied for final prediction. The voting process relies on the combined performance of these five algorithms, thus avoiding large errors or misclassifications that might occur with a single model. Finally, based on the voting results, metabolic biomarkers for schizophrenia diagnosis were selected, and the results are shown in Table 2. Compared with the normal control group, lysophosphatidylcholine (LPC) (17:0 / 0:0), oleamide, lysophosphatidylethanolamine (LysoPE) (18:0 / 0:0), palmitamide, lysophosphatidylethanolamine (LysoPE) (0:0 / 20:4), lysophosphatidylethanolamine (LysoPE) (22:6 / 0:0), p-methylphenol sulfate, lactic acid, 3-hydroxybutyric acid, lysophosphatidylethanolamine (LysoPE) (18:2 / 0:0), lysophosphatidylethanolamine (LysoPE) (16:0 / 0:0), and sphingosine 1-phosphate can serve as potential diagnostic biomarkers for schizophrenia.

[0091] Table 2 Plasma metabolic markers in patients with schizophrenia

[0092]

[0093] The above step metabolic marker group was characterized, and a diagnostic model was established with disease conditions as labels. The performance indicators of six machine learning models were calculated (see Table 3). By comparing the accuracy, sensitivity, specificity, precision, recall, Matthews coefficient (MCC), and AUC, it can be found that the prediction of schizophrenia by logistic regression (Ridge Regression, LR) reached the optimal performance, with an accuracy of 0.934783 and an AUC value of 0.976923 (see Figure 12 ).

[0094] Table 3 Performance indicators of six machine learning models

[0095] Table 3 Performance indicators of six machine learning models

[0096]

[0097] In summary, lysophosphatidylcholine LPC (17:0 / 0:0), oleamide, lysophosphatidylethanolamine LysoPE (18:0 / 0:0), palmitamide, lysophosphatidylethanolamine LysoPE (0:0 / 20:4), lysophosphatidylethanolamine LysoPE (22:6 / 0:0), p-methylphenyl sulfate, lactic acid, 3-hydroxybutyric acid, lysophosphatidylethanolamine LysoPE (18:2 / 0:0), lysophosphatidylethanolamine LysoPE (16:0 / 0:0) and sphingosine-1 -phosphate can be used in combination for the diagnosis of schizophrenia, with high diagnostic accuracy. Those skilled in the art can develop them into schizophrenia diagnostic kits, which contain standards of the 12 compounds, lysophosphatidylcholine LPC (17:0 / 0:0), oleamide, lysophosphatidylethanolamine LysoPE (18:0 / 0:0), palmitamide, lysophosphatidylethanolamine LysoPE (0:0 / 20:4), lysophosphatidylethanolamine LysoPE (22:6 / 0:0), p-methylphenyl sulfate, lactic acid, 3-hydroxybutyric acid, lysophosphatidylethanolamine LysoPE (18:2 / 0:0), lysophosphatidylethanolamine LysoPE (16:0 / 0:0) and sphingosine-1 -phosphate, for detecting lysophosphatidylcholine LPC (17:0 / 0:0), oleamide, lysophosphatidylethanolamine LysoPE (18:0 / 0:0), palmitamide, lysophosphatidylethanolamine LysoPE (0:0 / 20:4), lysophosphatidylethanolamine LysoPE (22:6 / 0:0), p-methylphenyl sulfate, lactic acid, 3-hydroxybutyric acid, lysophosphatidylethanolamine LysoPE (18:2 / 0:0), lysophosphatidylethanolamine LysoPE (16:0 / 0:0) and sphingosine-1 -phosphate in the sample to be tested. The kit can also use ketoprofen and L-2-chlorophenylalanine standards as detection internal standards.

[0098] Many modifications and variations of the present application described in the specification are possible without departing from the scope or spirit of the application. Other implementations of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. The specification and examples given are exemplary only.

Claims

1. A set of metabolic markers for use in the auxiliary diagnosis of the diagnosis of schizophrenia, characterized in that, The metabolic markers include lysophosphatidylcholine LPC (17:0 / 0:0), oleamide, lysophosphatidylethanolamine LysoPE (18:0 / 0:0), palmitamide, lysophosphatidylethanolamine LysoPE (0:0 / 20:4), lysophosphatidylethanolamine LysoPE (22:6 / 0:0), p-methylphenol sulfate, lactic acid, 3-hydroxybutyric acid, lysophosphatidylethanolamine LysoPE (18:2 / 0:0), lysophosphatidylethanolamine LysoPE (16:0 / 0:0), and sphingosine-1-phosphate.

2. A predictive model for aiding diagnosis of schizophrenia, characterized in that, The metabolic markers include the metabolic markers of claim 1.

3. The prediction model for assisting in the diagnosis of schizophrenia according to claim 2, wherein, The area under the curve of the prediction model is 0.976923.

4. A kit for assisting diagnosis of schizophrenia, characterized by comprising the antibody of claim 1 or 2. The metabolic markers include the metabolic markers of claim 1.

5. The kit of claim 3, wherein The kit further comprises standard samples of ketoprofen and L-2-chlorophenylalanine.

6. A screening method for metabolic markers for use in the auxiliary diagnosis of schizophrenia, characterized by, The kit further comprises standard samples of ketoprofen and L-2-chlorophenylalanine. S1, collecting and processing plasma samples to obtain a data set; S2, performing ultra-high performance liquid chromatography-quadrupole / orbitrap high-resolution mass spectrometry analysis on the plasma samples; S3, using Thermo Xcalibur software for data acquisition and processing, using Compound Discovery software retention time information, extracting molecular formula, mass-to-charge ratio, and secondary ion fragment information; S4, importing the information extracted in S3 into SIMCA for multivariate statistical analysis, establishing an orthogonal partial least squares discriminant analysis model, using variable importance projection indicators as screening indicators to screen differential metabolites, the variable importance projection indicators including P value, vip value, and difference fold FC; screening differential metabolites with VIP>1, P<0.05, and meeting the difference fold FC>1.25 or <0.8 as the candidate data set for selecting metabolic markers; S5, screening biomarkers using an integrated machine learning method, and screening differential metabolic markers from the candidate data set; S6, splitting the data set into a training set and a test set, training six machine learning models one by one using cross-validation grid search, and screening a final prediction model with optimal prediction ability from the differential metabolic markers.

7. The method for screening of metabolic markers for aiding diagnosis of schizophrenia according to claim 6, wherein, The machine learning method includes extreme gradient boosting algorithm, random forest, least absolute shrinkage and selection operator algorithm, ridge regression, and linear support vector machine.

8. The method for screening of metabolic markers for aiding diagnosis of schizophrenia according to claim 7, wherein, The specific steps of screening biomarkers using an integrated machine learning method are as follows: Each algorithm performs ten thousand independent random grouping experiments, and the feature factors with importance ranking in the top ten in each experiment are counted; the algorithms include extreme gradient boosting algorithm, random forest, least absolute shrinkage, selection operator algorithm, and ridge regression; According to the contribution of each feature factor to the prediction accuracy, the importance ranking is performed; Using a voting strategy, the feature factors with importance ranking in the top ten in each algorithm are voted; The voting results of ten thousand experiments are accumulated, and the feature factors with top ten votes in each algorithm are selected; Combining linear support vector machine algorithm, the coefficients of irrelevant characteristic factors are reduced to 0 by regularization method; The characteristic factors with coefficients of 0 are deleted, and the characteristic factors obtained in each experiment are counted; The characteristic factors obtained in each experiment are voted, the voting results of ten thousand experiments are accumulated, and the top ten characteristic factors are selected; The intersection of the voting results of the above algorithm is selected as the metabolic marker for schizophrenia diagnosis.

9. The method for screening of metabolic markers for aiding diagnosis of schizophrenia according to claim 6, wherein, The six machine learning models include decision tree, random forest, support vector machine, k-nearest neighbor, logistic regression and Gaussian naive Bayes.

10. The screening method for metabolic markers for aiding diagnosis of schizophrenia according to claim 8, wherein The machine learning models are compared in terms of accuracy, sensitivity, specificity, precision, recall, Matthew correlation coefficient and area under ROC curve (AUC) to select the model with the best prediction ability as the final prediction model.

Images