Application of dry syndrome markers in dry syndrome diagnostic product and method for constructing dry syndrome diagnostic model

Abstract

This invention discloses the application of Sjögren's syndrome biomarkers in products for diagnosing Sjögren's syndrome and a method for constructing a diagnostic model for Sjögren's syndrome, relating to the field of Sjögren's syndrome diagnostic technology. The Sjögren's syndrome biomarker includes the Val to Phe ratio. The diagnostic model established based on this single biomarker has an AUC value greater than 0.9 on its ROC curve, exhibiting extremely high diagnostic specificity, sensitivity, and accuracy. Combinations of this biomarker with other biomarkers also demonstrate extremely high diagnostic efficacy. Therefore, the Sjögren's syndrome diagnostic biomarkers provided by this invention have promising application prospects in the diagnosis of Sjögren's syndrome.

Description

Technical Field

[0001] This invention relates to the field of diagnostic technology for Sjögren's syndrome, and more specifically, to the application of Sjögren's syndrome biomarkers in products for diagnosing Sjögren's syndrome and methods for constructing diagnostic models for Sjögren's syndrome. Background Technology

[0002] Currently, Sjögren's syndrome (SS) is a chronic autoimmune disease that primarily affects glands that secrete bodily fluids, such as the lacrimal and salivary glands. Its diagnosis is mainly based on clinical symptoms of dryness, serological autoantibodies, and histopathological examination of glandular tissue. The current classification criteria for SS emphasize glandular immune infiltration and positive anti-SSA / Ro serology. Positive saliva biopsy or positive anti-SSA / Ro serology is mandatory. Other diagnostic criteria include: focal lymphocytic infiltration on labial gland biopsy, a Schirmer test (tear secretion test) ≤5 mm / 5 min in at least one eye, an OSS corneal staining score ≥5 points in at least one eye, and a spontaneous salivary flow rate ≤0.1 mL / min.

[0003] Studies show that approximately 70%-80% of patients with primary Sjögren's syndrome are positive for anti-SSA antibodies, while the remaining 20%-30% may be completely negative for antibodies. Patients with Sjögren's syndrome may not present with typical Sjögren's symptoms in the early stages, making early diagnosis crucial for effective treatment.

[0004] In view of this, the present invention is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide the application of Sjögren's syndrome biomarkers in products for diagnosing Sjögren's syndrome and a method for constructing a diagnostic model for Sjögren's syndrome, thereby enabling early diagnosis of Sjögren's syndrome and facilitating early detection and treatment.

[0006] This invention is implemented as follows: In a first aspect, the present invention provides the application of a reagent for detecting markers of Sjögren's syndrome in the preparation of products for diagnosing Sjögren's syndrome, wherein the markers of Sjögren's syndrome include the ratio of Val to Phe.

[0007] Secondly, the present invention also provides a method for constructing a diagnostic model for Sjögren's syndrome, which includes the following steps: i: Obtain feature data representing Sjögren's syndrome biomarkers in the training samples; the Sjögren's syndrome biomarkers are those mentioned above. ii: Utilize feature data to construct a diagnostic model for Sjögren's syndrome through machine learning models.

[0008] Thirdly, the present invention also provides a diagnostic device for Sjögren's syndrome, which includes: an input module, a control module and an output module; The input module is configured to: acquire feature data of biomarkers in the subject sample, wherein the biomarkers are selected from the above-mentioned Sjögren's syndrome biomarkers; The control module includes an assessment module configured to generate a risk prediction result for Sjögren's syndrome of the subject based on the characteristic data of the biomarkers of Sjögren's syndrome contained in the subject's sample. The output module is configured to output the subject's Sjögren's syndrome risk prediction results.

[0009] The present invention has the following beneficial effects: This invention integrates the expression levels of multiple amino acids and clinical indicators to screen biomarkers that show significant differences between patients with Sjögren's syndrome and healthy controls. Among these, various diagnostic models based on the valine / phenylalanine ratio exhibit extremely high diagnostic efficacy for Sjögren's syndrome. The diagnostic model based on this single biomarker has an AUC value greater than 0.9, demonstrating extremely high diagnostic specificity, sensitivity, and accuracy. Combinations of this biomarker with other biomarkers also exhibit extremely high diagnostic efficacy. Therefore, the diagnostic biomarkers for Sjögren's syndrome provided by this invention have promising applications in the diagnosis of Sjögren's syndrome.

[0010] The invention contributes to the early detection and treatment of Sjögren's syndrome, and facilitates timely treatment for patients with Sjögren's syndrome. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A graph showing the differences in the levels of eight biomarkers in the serum of SS and healthy individuals; Figure 2 The results of the Shapley Additive Explanation (SHAP) model for 8 markers are shown in the figure. Figure 3 Plot the ROC curves of eight V8 models built based on eight classifier algorithms using the training set (a) and the test set (b); Figure 4 ROC curve plotted for the model built using the test set based on the markers of Comparative Example 1; Figure 5 To plot the ROC curve of a single marker using a test set based on a random forest model; Figure 6To plot ROC curves for various combinations of markers based on a random forest model using a test set. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Where specific conditions are not specified in the embodiments, conventional conditions or conditions recommended by the manufacturer shall apply. Reagents or instruments whose manufacturers are not specified are all conventional products that can be purchased commercially.

[0014] Definition of noun The term "marker" broadly refers to any detectable compound or cell present in or derived from a sample, such as a protein, peptide, proteoglycan, glycoprotein, lipoprotein, cell, or any of the foregoing substances, that is differentiating molecule or differentiating fragment. For example, the detection of or binding to a specific antibody can indicate the presence of a specific antigen (e.g., a protein) in a sample. Here, a differentiating molecule or fragment is a molecule or fragment that, upon detection, indicates the presence or abundance of the aforementioned identified compound or cell. Markers can, for example, be isolated from the sample, measured directly in the sample, or detected or determined in the sample. Markers can, for example, be functional, partially functional, or non-functional. Markers may also be synonymous with "biomarker."

[0015] The term "sample" refers to a biological specimen obtained from or derived from an individual for a purpose. The source of the biological specimen can be a fresh, frozen, and / or preserved organ or tissue sample or solid tissue derived from a biopsy or primer; blood or any blood component. The term "sample" includes biological samples that have been manipulated in any way after acquisition, such as by reagent treatment, stabilization, enrichment for certain components (e.g., proteins or polynucleotides), or embedding in a semi-solid or solid matrix for sectioning purposes. In this invention, the sample is particularly a peripheral blood sample, a whole blood sample, or a serum sample.

[0016] The term "subject" as used in this article can be understood as anyone involved in the diagnosis of Sjögren's syndrome. A subject can be a patient in a clinical setting.

[0017] The term "predetermined threshold" refers to a parameter used to compare a marker or combination of markers in a subject's sample with a predetermined threshold when diagnosing disease risk, and outputs the subject's disease risk or disease course based on the comparison result.

[0018] In a first aspect, the present invention provides the application of a reagent for detecting markers of Sjögren's syndrome in the preparation of products for diagnosing Sjögren's syndrome, wherein the markers of Sjögren's syndrome include the ratio of Val to Phe.

[0019] The Val to Phe ratio refers to the ratio of the concentration of Val (valine) to the concentration of Phe (phenylalanine) in a sample. This ratio can be calculated by detecting the content of Val and Phe separately by LC-MS / MS.

[0020] Several different diagnostic models based on the valine / phenylalanine ratio have demonstrated extremely high diagnostic efficacy for Sjögren's syndrome. Diagnostic models based on this single biomarker exhibit an AUC value greater than 0.9 on their ROC curves, demonstrating exceptionally high diagnostic specificity, sensitivity, and accuracy. Biomarker combinations with other biomarkers also demonstrate extremely high diagnostic efficacy.

[0021] Compared to anti-SSA antibody detection, the Sjögren's syndrome biomarker provided by this invention has higher detection accuracy, specificity, and sensitivity.

[0022] In a preferred embodiment of the present invention, the markers for Sjögren's syndrome further include markers selected from at least one of the following: homocysteine, aspartic acid, cysteine, glutamic acid, valine, serum albumin to globulin ratio (A / G), and serum globulin concentration (G).

[0023] Those skilled in the art can combine the Val to Phe ratio with at least one of the above seven biomarkers as a biomarker for Sjögren's syndrome.

[0024] For example, markers for Sjögren's syndrome are selected from: Val / Phe and homocysteine; Val / Phe and aspartic acid; Val / Phe and cysteine; Val / Phe and glutamate; Val / Phe and valine; Val / Phe and A / G; Val / Phe and G; Val / Phe, homocysteine ​​and aspartic acid; Val / Phe, homocysteine ​​and cysteine; Val / Phe, homocysteine ​​and glutamate; Val / Phe, homocysteine ​​and valine; Val / Phe, homocysteine ​​and A / G; Val / Phe, homocysteine ​​and G; Val / Phe, aspartic acid, cysteine ​​and glutamate; Val / Phe, cysteine, glutamate, valine and A / G.

[0025] Amino acid levels can be obtained by LC-MS / MS, and the ratio of serum albumin to globulin and serum globulin concentration can be obtained by routine biochemical assays.

[0026] Globulin (GLOB) concentration can be calculated by subtracting albumin concentration from total protein concentration. Total protein concentration can be quantified using the biuret method, while albumin can be quantified using the bromocresol green (BCG) method.

[0027] In a preferred embodiment of the present invention, the biomarker for Sjögren's syndrome is selected from at least one of the following biomarker combinations: (1) The ratio of Val to Phe and homocysteine; (2) The ratio of Val to Phe, homocysteine ​​and glutamate; (3) The ratio of Val to Phe, homocysteine, glutamate and serum globulin concentration; (4) Val to Phe ratio, homocysteine, glutamate, serum globulin concentration and aspartic acid; (5) Val to Phe ratio, homocysteine, glutamate, serum globulin concentration, aspartic acid and cysteine; (6) Val to Phe ratio, homocysteine, glutamate, serum globulin concentration, aspartic acid, cysteine ​​and valine; (7) The ratio of Val to Phe, homocysteine, glutamic acid, serum globulin concentration, aspartic acid, cysteine, valine and the ratio of serum albumin to globulin.

[0028] The AUC value of the ROC curve of the diagnostic model constructed based on the above biomarker combination is significantly improved compared with that of a single Val / Phe biomarker, resulting in higher diagnostic accuracy. The above biomarker combination (5)-(7) has higher diagnostic efficacy.

[0029] In a preferred embodiment of the present invention, the product for diagnosing Sjögren's syndrome is selected from a kit or a detection device.

[0030] In a preferred embodiment of the present invention, the reagent for detecting markers of Sjögren's syndrome is selected from at least one of LC-MS / MS detection reagents and biochemical detection reagents.

[0031] In a preferred embodiment of the present invention, the application is to detect Sjögren's syndrome diagnostic markers in whole blood or serum samples of the subject.

[0032] Secondly, the present invention also provides a method for constructing a diagnostic model for Sjögren's syndrome, which includes the following steps: i: Obtain feature data representing Sjögren's syndrome biomarkers in the training samples; the Sjögren's syndrome biomarkers are those mentioned above. ii: Utilize feature data to construct a diagnostic model for Sjögren's syndrome through machine learning models.

[0033] In a preferred embodiment of the present invention, the machine learning model is selected from at least one of the following: support vector machine, random forest, logistic regression, K-nearest neighbors, Gaussian Bayes, Naive Bayes, AdaBoost, XGBoost, and DT decision tree.

[0034] After multi-model validation, the random forest, decision tree, logistic regression, Gaussian Bayes, support vector machine and AdaBoost models constructed based on the Sjögren's syndrome biomarkers provided in this invention all have extremely high diagnostic efficacy, with extremely high diagnostic accuracy, specificity and sensitivity.

[0035] Characteristic data of biomarkers for Sjögren's syndrome in representative training samples were obtained by performing liquid chromatography-tandem mass spectrometry analysis, or LC-MS / MS and biochemical detection on representative training samples, to obtain the concentration and / or concentration ratio of biomarkers for Sjögren's syndrome in representative training samples.

[0036] Thirdly, the present invention also provides a diagnostic device for Sjögren's syndrome, which includes: an input module, a control module and an output module; The input module is configured to: acquire feature data of biomarkers in the subject sample, wherein the biomarkers are selected from the above-mentioned Sjögren's syndrome biomarkers; The control module includes an assessment module configured to generate a risk prediction result for Sjögren's syndrome of the subject based on the characteristic data of the biomarkers of Sjögren's syndrome contained in the subject's sample. The output module is configured to output the subject's Sjögren's syndrome risk prediction results.

[0037] In a preferred embodiment of the present invention, the evaluation module is configured to: input the characteristic data of the Sjögren's syndrome markers contained in the subject's sample into a pre-established Sjögren's syndrome diagnostic model to obtain a Sjögren's syndrome risk prediction value; compare the Sjögren's syndrome risk prediction value with a predetermined threshold; and generate the subject's Sjögren's syndrome risk prediction result based on the comparison result.

[0038] For example, a logistic regression model can be established to obtain the cut-off value of the model, such as 0.50 (i.e., a predetermined threshold). When the predicted risk value for Sjögren's syndrome in the tested sample is >0.50, it is diagnosed as Sjögren's syndrome; when it is <0.50, it is diagnosed as non-Sjögren's syndrome (healthy control). The predicted risk value for Sjögren's syndrome can be calculated from the dependent variable of the logistic regression model, or obtained from the ROC curve.

[0039] The predetermined thresholds include, but are not limited to, positive judgment values, predicted probability thresholds, dependent variable thresholds, and other critical values ​​that can be used to divide results and clarify judgment boundaries. The setting of these thresholds should be based on the performance requirements of the target detection scenario (such as sensitivity, specificity, accuracy, etc.) and determined through clinical data validation, statistical model analysis (such as ROC curve analysis combined with Youden index calculation, etc.) or industry standard calibration, so as to ensure their validity and reliability in the detection method or diagnostic model.

[0040] Fourthly, the present invention provides a diagnostic device for Sjögren's syndrome, the device comprising a processor and a memory, the memory storing a set of executable program instructions, the set of executable program instructions being loaded and executed by the processor to implement a method for diagnosing Sjögren's syndrome, the method for diagnosing Sjögren's syndrome comprising: The characteristic data of biomarkers in the subject's sample were obtained, and the biomarkers were selected from the above-mentioned Sjögren's syndrome biomarkers; based on the characteristic data of the Sjögren's syndrome biomarkers contained in the subject's sample, the risk prediction result of Sjögren's syndrome for the subject was generated.

[0041] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0042] Example 1 This embodiment involves the screening of diagnostic biomarkers.

[0043] 1. Materials and Methods: (1) Participants Patients included in this study were treated at Xijing Hospital (Xi'an, China) between June 1, 2022 and November 2, 2023, and were divided into two groups: patients with Sjögren's syndrome (SS) (n=108) and a control group (n=104). All SS patients met the 2016 ACR / EULAR criteria. SS patients included those who had not received any steroid, immunosuppressive, or antibiotic treatment in the past three months. Participants in the control group were healthy and disease-free. Exclusion criteria were: (1) receiving or having a history of immunosuppressive therapy, and (2) receiving or having a history of hormone therapy.

[0044] (2) Blood preparation After fasting overnight, venous blood is collected by an experienced nurse and processed according to the requirements of biochemical, immunological and routine blood tests.

[0045] (3) Data collection All tests were performed in the clinical laboratory of Xijing Hospital, and the data was uploaded and stored in the LIS system. Patient basic information was collected and entered by trained personnel. All test results underwent quality control before, during, and after the tests to ensure quality and accuracy.

[0046] (4) LC-MS / MS materials and equipment HPLC-MS / MS was used to analyze 21 amino acids [alanine (Ala), arginine (Arg), aspartic acid (Asp), citrulline (Cit), cysteine ​​(Cys), glutamine (Gln), glutamic acid (Glu), glycine (gly), homocysteine ​​(Hcy), histidine (His), leucine (Leu), lysine (Lys), methionine (Met), ornithine (Orn), phenylalanine (Phe), proline (Pro), serine (Ser), threonine (Thr), tryptophan (Trp), tyrosine (Tyr), and valine (Val)]. Standard AA was labeled (Group A, Cambridge Isotope Laboratory) and compared with 15N13C Gly, d4 Ala, d4C13 Arg, d3Asp, d2 Cit (d2 citrulline), d3 Glu, d3 Leu, d3 Met, d2 Orn (d2 ornithine), 13C6-Phe, 13C6-Tyr and d8-Val (Cambridge Isotope Laboratory). All analyses were performed using LC-MS / MS.

[0047] (5) LC-MS / MS sample pretreatment To detect 21 amino acids, 50 μL of serum was dropped onto blank blood collection filter paper and allowed to fully saturate. A 3.5 mm strip of the dried serum filter paper was removed with a stick and then extracted with 100 μL of a known concentration of isotope standard extract for 15 minutes. The extract was dried under nitrogen. A 3 mol / L solution of n-butanol hydrochloride was prepared using n-butanol and acetyl chloride at a volume ratio of 9:1. Then, 60 μL of n-butanol hydrochloride was added to the dried sample, and a derivatization reaction was carried out at 65 °C for 20 minutes. The derivatized solution was freeze-dried in N2, and then 100 μL of ACN solution was added as the detection solution. The mass spectra of the 21 substances tested on the serum slide were analyzed using a series of mass spectrometers, and their concentrations were calculated.

[0048] (6) Mass spectrometry analysis To detect 21 amino acids (AAs), 100% ACN was used as the HPLC eluent, and the optimized HPLC parameters are listed in Table 1. The injection volume was 20 μL. Two experiments were performed in positive mode within one mass spectrometry cycle, and the optimized MS / MS parameters are listed in Table 2. In Experiment 1, the scan type was as follows: neutral loss, loss of 102.00 Da, starting at 140.00 Da and stopping at 280.00 Da; CEP started at 13.07 and stopped at 17.46. In Experiment 2, the scan type was multiple reaction monitoring (MRM), and the optimized parameters are shown in Table 3.

[0049] Table 1: Gradient conditions of mobile phase for liquid chromatography detection of 21 amino acids

[0050] Table 2: Mass spectrometry parameters for detecting 21 amino acids

[0051] Table 3: Mass spectrometry parameters for multiple reaction monitoring (MRM) mode detection of 21 amino acids

[0052] (7) Statistical analysis Mass spectrometry data were analyzed using Analyst (version 1.6.2) and ChemView software (version 1.6.1; ABSciex, Darmstadt, Germany). Quantitative data were analyzed using SPSS (version 23.0; IBM, Armonk, NY, USA), with independent t-tests and p-values ​​(p<0.05) used for comparisons. Quantitative data are expressed as mean ± standard deviation. Data visualization was performed using GraphPad Prism (version 5; GraphPad Software, San Diego, CA, USA) and R software (version 3.6.2; R Statistical Computing Project, p<0.05).

[0053] 2. Establishment of the diagnostic model To build an effective diagnostic model, Spearman, VT, and MIC methods were used to screen for the 30 indicators with the highest contribution values, and common indicators were considered as potential indicators for model building. To enhance and validate the effectiveness of the SS diagnostic model, eight machine learning classification algorithms (random forest, decision tree, logistic regression, Gaussian Bayes, support vector machine, and AdaBoost) were used to construct the model. The sensitivity, specificity, accuracy, positive predictive value, negative predictive value, area under the curve (AUC), and ROC curve of these models were evaluated. To ensure the stability of all algorithms, 5-fold cross-validation was performed on the data from step 1, with 5 runs used for training and 5 runs used for testing.

[0054] Patient characteristics results show: By analyzing participants' basic information, including height, weight, blood pressure, and lifestyle habits, as well as the expression levels and differences of all indicators, there were no significant differences in basic information between the SS group and the control group. Specifically, the SS group and the control group were similar in age, sex, and BMI, with no significant differences; there were also no significant differences in blood type, smoking, and alcohol consumption.

[0055] Performance evaluation of candidate metrics based on classification algorithms: Based on the differences in the expression levels of various amino acids and clinical indicators, three methods (Spearman, VT, and MIC) were used to cross-screen the top 30 indicators with the greatest differences, and potential biomarkers for SS and control groups were extracted for modeling.

[0056] Eight biomarkers showing the greatest reproducibility across the Spearman, VT, and MIC methods [aspartate (Asp), cysteine ​​(Cys), glutamate (Glu), homocysteine ​​(Hcy), valine (Val), valine / phenylalanine (Val / Phe), albumin / globulin (A / G), serum globulin concentration (G)] were selected as candidate biomarkers. The levels of these eight biomarkers in SS and healthy individuals were used as references. Figure 1 As shown in the figure. The results showed that there were significant differences in these eight biomarkers between SS and healthy individuals (p<0.0001). Albumin / globulin (A / G) and serum globulin concentration (G) were obtained by routine biochemical assays.

[0057] Eight-indicator Shapley Additive Explanation (SHAP) model reference Figure 2 As shown, Hcy contributes the most to the model, followed by Val / Phe, Glu, G, Asp, Cys, Val, and A / G.

[0058] Furthermore, eight classifier algorithms were used to build a joint diagnostic model (V8) for eight biomarkers, and the efficacy of the diagnostic models was tested. Five-fold cross-validation was performed on the data, with five runs used for training and five runs for testing, all randomly assigned. The sample consisted of 54 SS cases and 52 normal control cases. The ROC, sensitivity, specificity, and accuracy of the V8 model obtained in the training set were >0.969, 81.91%, 96.19%, and 91.90%, respectively (Table 4). Figure 3 (Figure a in the text).

[0059] Table 4. Statistics on ROC, sensitivity, specificity, and accuracy of the V8 model.

[0060] ROC curves were plotted for eight V8 models built based on eight different classifier algorithms using the test set. AUC values, sensitivity, specificity, and accuracy were calculated. Figure 3 Figure b shows that the AUC values ​​in the test set are all >0.896, the sensitivity is all >78.19%, the specificity is all >90.35%, and the accuracy is all >87.98%.

[0061] In other words, the diagnostic model established by these eight biomarkers has extremely high diagnostic efficacy.

[0062] Example 2 This embodiment provides a diagnostic method for Sjögren's syndrome, which uses a combination of eight biomarkers (V8) to construct a logistic regression model for Sjögren's syndrome.

[0063] (1) Mass spectrometry was performed on 44 serum samples according to the method in Example 1 to obtain the results of Hcy, Val / Phe, Glu, G, Asp, Cys, Val and A / G respectively.

[0064] (2) Then, substitute the mass spectrometry parameters from step (1) into the logistic regression model of the eight biomarker combinations and output the prediction results.

[0065] Sjögren's syndrome is diagnosed according to the following criteria.

[0066] The cut-off value of the diagnostic model of the combination of 8 biomarkers for Sjögren's syndrome was 0.50. A value >0.50 was used to diagnose Sjögren's syndrome (SS), and a value <0.50 was used to diagnose non-SS (healthy control).

[0067] The results in Table 4 show that the combination of the seven biomarkers has extremely high diagnostic accuracy and consistency.

[0068] Table 4. Statistical table of diagnostic results for 7 biomarker combinations

[0069] Comparative Example 1 ROC curves of anti-SSA (immunoblotting, OUMEN Hangzhou Medical Laboratory Diagnostics Co., Ltd., CD241121AB) models built based on six classifier algorithms were plotted using the test set of Example 1.

[0070] Results reference Figure 4 As shown, the results indicate that the AUC value of the ROC curve plotted by the anti-SSA model established by the RF classifier algorithm is only 0.797, which is far lower than the effect of the combined diagnosis of 8 biomarkers in Example 1. The sensitivity, specificity, and accuracy are 59.98%, 100%, and 85.92%, respectively.

[0071] Based on the previously established RF model of 8 biomarker combinations and the anti-SSA RF model of Comparative Example 1, the diagnostic efficacy was tested using test set samples. The results showed that, under the same type of RF model, the 8 biomarker combinations provided by this invention have significantly better diagnostic efficacy for Sjögren's syndrome.

[0072] Experimental Example 1 The diagnostic efficacy of a single biomarker was evaluated using the test set samples from Example 1. RF models were constructed separately, and the diagnostic results are shown below. Figure 5 The cutoff value for the diagnostic models is 0.5.

[0073] Figure 5 In the study, the AUC value of the ROC curve of the Hcy model was only 0.870, with sensitivity, specificity and accuracy of 61.89%, 84.70% and 76.74%, respectively.

[0074] The AUC value of the Val / Phe model's ROC curve was 0.909, with sensitivity, specificity, and accuracy of 79.90%, 88.10%, and 85.22%, respectively.

[0075] The AUC value of the ROC curve of the Glu model was 0.825, and the sensitivity, specificity and accuracy were 69.98%, 83.69% and 78.83%, respectively.

[0076] The AUC value of the ROC curve of the G model was 0.802, with sensitivity, specificity, and accuracy of 67.77%, 81.68%, and 76.85%, respectively.

[0077] The AUC value of the ROC curve of the ASP model was 0.883, and the sensitivity, specificity and accuracy were 79.90%, 87.99% and 85.20%, respectively.

[0078] The Cys model had an AUC of 0.886, with sensitivity, specificity, and accuracy of 84.07%, 85.95%, and 85.20%, respectively.

[0079] The AUC value of the Val model's ROC curve was 0.836, with sensitivity, specificity, and accuracy of 65.93%, 81.43%, and 76.05%, respectively.

[0080] The AUC value of the ROC curve of the A / G model was 0.790, and the sensitivity, specificity and accuracy were 59.93%, 92.44% and 81.04%, respectively.

[0081] Experimental Example 2 The diagnostic efficacy of the two biomarker combinations was evaluated using the test set samples from Example 1. An RF model was constructed, and the diagnostic efficacy was referenced... Figure 6 As shown.

[0082] The AUC values ​​of the ROC curves for the Hcy and Val / Phe models were 0.965, with sensitivity, specificity, and accuracy of 95.96%, 87.89%, and 90.84%, respectively.

[0083] Experimental Example 3 The diagnostic efficacy of the three biomarker combinations was evaluated based on the test set samples from Example 1. An RF model was constructed, and the diagnostic efficacy was referenced... Figure 6 As shown.

[0084] The AUC value of the ROC curves for the Hcy, Val / Phe, and Glu models was 0.977, with sensitivity, specificity, and accuracy of 98.04%, 95.66%, and 96.50%, respectively.

[0085] Experiment Example 4 The diagnostic efficacy of four biomarker combinations was evaluated based on the test set samples from Example 1. An RF model was constructed, and the diagnostic efficacy was referenced... Figure 6 As shown.

[0086] The AUC values ​​of the ROC curves for the Hcy, Val / Phe, Glu, and G models were 0.985, with sensitivity, specificity, and accuracy of 96.08%, 92.44%, and 93.68%, respectively.

[0087] Experimental Example 5 The diagnostic efficacy of five biomarker combinations was evaluated based on the test set samples from Example 1. An RF model was constructed, and the diagnostic efficacy was referenced... Figure 6 As shown.

[0088] The AUC values ​​of the ROC curves for the Hcy, Val / Phe, Glu, G, and Asp models were 0.990, with sensitivity, specificity, and accuracy of 96.08%, 94.59%, and 95.08%, respectively.

[0089] Experimental Example 6 The diagnostic efficacy of six biomarker combinations was evaluated based on the test set samples from Example 1. An RF model was constructed, and the diagnostic efficacy was referenced... Figure 6 As shown.

[0090] The AUC values ​​of the ROC curves for the Hcy, Val / Phe, Glu, G, Asp, and Cys models were 0.991, with sensitivity, specificity, and accuracy of 96.08%, 95.59%, and 95.77%, respectively.

[0091] Example 7 The diagnostic efficacy of seven biomarker combinations was evaluated based on the test set samples from Example 1. An RF model was constructed, and the diagnostic efficacy was referenced... Figure 6 As shown.

[0092] The AUC values ​​of the ROC curves for the Hcy, Val / Phe, Glu, G, Asp, Cys, and Val models were 0.992, with sensitivity, specificity, and accuracy of 94.00%, 97.81%, and 96.48%, respectively.

[0093] In summary, the biomarker combination provided by this invention has extremely high diagnostic efficacy for Sjögren's syndrome, which is beneficial for the early detection and treatment of Sjögren's syndrome.

[0094] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. The application of reagents for detecting markers of Sjögren's syndrome in the preparation of products for diagnosing Sjögren's syndrome, characterized in that, The markers for Sjögren's syndrome include the ratio of Val to Phe.

2. The application according to claim 1, characterized in that, The markers for Sjögren's syndrome also include markers selected from at least one of the following: homocysteine, aspartic acid, cysteine, glutamic acid, valine, the ratio of serum albumin to globulin, and serum globulin concentration.

3. The application according to claim 2, characterized in that, The markers for Sjögren's syndrome are selected from at least one of the following combinations of markers: (1) The ratio of Val to Phe and homocysteine; (2) The ratio of Val to Phe, homocysteine ​​and glutamate; (3) The ratio of Val to Phe, homocysteine, glutamate and serum globulin concentration; (4) Val to Phe ratio, homocysteine, glutamate, serum globulin concentration and aspartic acid; (5) Val to Phe ratio, homocysteine, glutamate, serum globulin concentration, aspartic acid and cysteine; (6) Val to Phe ratio, homocysteine, glutamate, serum globulin concentration, aspartic acid, cysteine ​​and valine; (7) The ratio of Val to Phe, homocysteine, glutamic acid, serum globulin concentration, aspartic acid, cysteine, valine and the ratio of serum albumin to globulin.

4. The application according to claim 1, characterized in that, The diagnostic product for Sjögren's syndrome is selected from kits or testing devices.

5. The application according to claim 1, characterized in that, The reagents for detecting markers of Sjögren's syndrome are selected from at least one of LC-MS / MS detection reagents and biochemical detection reagents.

6. The application according to claim 5, characterized in that, The application is to detect diagnostic markers for Sjögren's syndrome in whole blood or serum samples of subjects.

7. A method for constructing a diagnostic model for Sjögren's syndrome, characterized in that, It includes the following steps: i: Obtain feature data representing Sjögren's syndrome biomarkers in the training samples; the Sjögren's syndrome biomarkers are any one of claims 1-6; ii: Using the aforementioned feature data, a diagnostic model for Sjögren's syndrome is constructed through a machine learning model.

8. The method for constructing a diagnostic model for Sjögren's syndrome according to claim 7, characterized in that, The machine learning model is selected from at least one of the following: Support Vector Machine, Random Forest, Logistic Regression, K Nearest Neighbors, Gaussian Bayes, Naive Bayes, AdaBoost, XGBoost, and DT Decision Tree; The characteristic data of the Sjögren's syndrome biomarkers in the representative training samples were obtained by performing liquid chromatography-tandem mass spectrometry analysis, or LC-MS / MS and biochemical detection on the representative training samples, to obtain the concentration and / or concentration ratio of the Sjögren's syndrome biomarkers in the representative training samples.

9. A diagnostic device for Sjögren's syndrome, characterized in that, It includes: Input module, control module, and output module; The input module is configured to: acquire feature data of biomarkers in subject samples, wherein the biomarkers are selected from any one of claims 1-6 for Sjögren's syndrome biomarkers; The control module includes an evaluation module configured to generate a risk prediction result for Sjögren's syndrome of the subject based on the characteristic data of the biomarkers of Sjögren's syndrome contained in the subject's sample. The output module is configured to output the subject's Sjögren's syndrome risk prediction results.

10. The diagnostic device for Sjögren's syndrome according to claim 9, characterized in that, The assessment module is configured to: input the characteristic data of Sjögren's syndrome markers contained in the subject's sample into a pre-established Sjögren's syndrome diagnostic model to obtain a Sjögren's syndrome risk prediction value; compare the Sjögren's syndrome risk prediction value with a predetermined threshold; and generate the subject's Sjögren's syndrome risk prediction result based on the comparison result.

Images