Detection method for establishing brucella infected serum metabolic map based on laser dissociation mass spectrum

By using iron nanoparticle-assisted laser desorption/ionization mass spectrometry and machine learning algorithms, metabolic biomarkers for brucellosis were screened out, overcoming the limitations of existing diagnostic methods and achieving rapid and accurate diagnosis of brucellosis with the advantages of high sensitivity and low cost.

CN120927782APending Publication Date: 2025-11-11THE THIRD AFFILIATED HOSPITAL OF XINJIANG MEDICAL UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511024684.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-24
Publication Date
2025-11-11

AI Technical Summary

Technical Problem

Existing diagnostic methods for brucellosis have limitations in sensitivity and specificity, leading to misdiagnosis and mistreatment. There is an urgent need for a rapid and accurate diagnostic method.

Method used

By combining iron nanoparticle-assisted laser desorption/ionization mass spectrometry with machine learning algorithms, a serum metabolic profile of Brucella infection was constructed, and metabolic biomarkers such as glycerol, lactic acid, succinic acid, glycolic acid, and glycine were screened for use in the diagnosis of brucellosis.

Benefits of technology

It achieves accurate diagnosis of brucellosis with low detection cost, accurate results, high sensitivity, and can process a large number of samples in a short time. The coefficient of variation of the detection results is less than 10%, and the AUC of the diagnostic model reaches 0.946.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120927782A_ABST
    Figure CN120927782A_ABST
Patent Text Reader

Abstract

The invention relates to a detection method for establishing a brucella infected serum metabolic map based on laser dissociation mass spectrometry, which comprises the following steps: step 1, performing metabolism detection on serum samples of brucellosis patients and healthy volunteers by using an iron nanoparticle-assisted laser desorption ionization mass spectrometry technology to obtain a serum metabolism fingerprint map; step 2, carrying out various model training on a serum sample by adopting a machine learning algorithm on the serum metabolism fingerprint spectrogram obtained in the step 1, evaluating various models through a cross validation method, and screening a final diagnosis model according to a validation result; and step 3, diagnosing serum samples of brucellosis patients and healthy volunteers according to the diagnosis model in the step 2, and screening metabolic biomarkers in the serum samples according to a diagnosis result. According to the marker combination, the detection process is rapid and simple, the result is accurate and reliable, and the screened metabolic biomarker can be used for diagnosing brucellosis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of medical diagnostic technology, and specifically discloses a detection method for establishing a serum metabolic profile of Brucella infection based on laser dissociation mass spectrometry. Background Technology

[0002] Brucellosis, also known as undulant fever, is a common zoonotic infectious disease caused by Brucella bacteria. Humans can contract brucellosis through contact with the blood, urine, milk, or meat of infected animals. Globally, approximately 3.5 billion people are at ongoing risk of brucellosis infection. Currently, clinical diagnosis of brucellosis mainly relies on general laboratory tests, bacterial cultures, and serological tests. However, these methods still have limitations in sensitivity and specificity. Patients with brucellosis are often misdiagnosed and mistreated due to nonspecific clinical manifestations and limitations in testing methods, causing significant challenges to clinical diagnosis and treatment. New methods for rapid and accurate diagnosis of brucellosis are urgently needed.

[0003] Timely and accurate diagnosis and symptomatic treatment after the early clinical symptoms of brucellosis can reduce the recurrence rate and improve patient prognosis. The selection of biomarkers and detection methods is crucial for designing diagnostic assays. Molecular biomarkers at the genomic / proteomic / metabolic levels in bodily fluids are effective indicators for identifying changes in biological systems associated with pathological processes, and obtaining bodily fluids is less invasive. Genomic and proteomic biomarkers are located upstream of metabolic pathways and typically reflect indirect information about biological systems. In contrast, metabolic biomarkers, as end products of pathways, can directly characterize disease phenotypes and progression, offering unique advantages in early diagnosis. Therefore, exploring novel serum metabolic biomarkers is clinically significant for the diagnosis of brucellosis.

[0004] Mass spectrometry has become a primary tool for screening metabolic biomarkers. The development of matrix-assisted laser desorption / ionization mass spectrometry (MADS) has made it possible to directly detect metabolites in ultra-low volumes of bodily fluids without the need for enrichment or purification. This method offers the advantage of high-throughput screening and can be used for accurate, non-targeted metabolic analysis, thereby allowing the identification of specific disease biomarkers to reflect the true condition of patients.

[0005] Therefore, those skilled in the art are dedicated to developing a method for detecting serum metabolomics of Brucella infection based on laser dissociation mass spectrometry for the diagnosis of brucellosis. Summary of the Invention

[0006] This application provides a detection method for establishing a serum metabolic profile of Brucella infection based on laser dissociation mass spectrometry, in order to solve the problems mentioned in the background art.

[0007] This application discloses a method for detecting Brucella infection by establishing a serum metabolic profile based on laser dissociation mass spectrometry, the method comprising the following steps:

[0008] Step 1: Using iron nanoparticle-assisted laser desorption / ionization mass spectrometry, metabolic detection was performed on serum samples from brucellosis patients and healthy volunteers to obtain serum metabolic fingerprint profiles.

[0009] Step 2: Using the serum metabolic fingerprint obtained in Step 1, train multiple models on the serum samples using machine learning algorithms, evaluate the multiple models through cross-validation, and select the final diagnostic model.

[0010] Step 3: Diagnose serum samples from brucellosis patients and healthy volunteers based on the diagnostic model from Step 2. Based on the diagnostic results, screen metabolic biomarkers in the serum samples.

[0011] Furthermore, step 1 specifically includes:

[0012] Step 1.2, Instrument and Reagent Preparation: Matrix-Assisted Laser Desorption / Ionization Mass Spectrometry, serum sample, deionized water, matrix;

[0013] Step 1.3: Dilute the serum sample 10 times with deionized water;

[0014] Step 1.4: Prepare a matrix solution of 1 mg / mL using deionized water with iron nanoparticles;

[0015] Step 1.5: Prepare samples on the mass spectrometry target plate. Spot 1.5 μL of each diluted serum sample and dry at room temperature.

[0016] Step 1.6: Prepare the matrix on the mass spectrometry target plate, spotting 1.5 μL of each matrix solution and drying at room temperature;

[0017] Step 1.7: Collect a large number of serum sample metabolomics data using laser desorption / ionization mass spectrometry;

[0018] Step 1.8: Preprocess the serum sample metabolomics data to obtain features under positive ion mode.

[0019] Furthermore, the matrix in step 1.2 is iron nanoparticles.

[0020] Furthermore, the data acquisition in step 1.7 is specifically positive ion mode acquisition.

[0021] Furthermore, the serum samples in step 1.7 include serum samples from the brucellosis group and serum samples from the healthy control group.

[0022] Furthermore, the preprocessing in step 1.8 includes data resampling, spectral smoothing, baseline correction, feature extraction, peak alignment, and missing value imputation.

[0023] Furthermore, step 2 specifically includes:

[0024] Step 2.1, Machine Learning Environment Preparation: Python 3.7;

[0025] Step 2.2: Divide the serum samples into a training set and a test set;

[0026] Step 2.3: Train four models on the training set using machine learning algorithms; evaluate the multiple models using methods such as cross-validation; visualize the prediction results on the test set and select the final prediction model.

[0027] Furthermore, the four model training algorithms in step 2.3 include Algorithm 1 Ridge Regression, Algorithm 2 Decision Tree, Algorithm 3 Adaptive Boosting, and Algorithm 4 Random Forest.

[0028] Furthermore, step 3 specifically includes:

[0029] Step 3.1: Perform intensity analysis on the serum fingerprint profiles of the brucellosis group and the healthy control group, compare the average intensity of the serum fingerprint profile features, and screen out the features with an average intensity greater than or equal to 1000.

[0030] Step 3.2: Perform Student's t-test analysis on the serum fingerprint profiles of the brucellosis group and the healthy control group to compare whether there are significant differences in metabolite levels between the brucellosis group and the healthy control group, and screen for features with P less than 0.05.

[0031] Step 3.3: Use the Ridge algorithm to select features from 113 m / z signals on the training set for the serum fingerprint spectrum, calculate the contribution of each feature to the model prediction results, and select features with a contribution greater than or equal to 0.2.

[0032] Step 3.4: Perform cross-matching on the features selected by mean intensity, Student's t-test, and significant contribution score to finally screen differential metabolites and identify metabolic biomarkers.

[0033] Step 3.5: Based on the differential metabolites screened in Step 3.4, their corresponding molecular formulas were determined and identified in the human metabolome database as glycerol, lactic acid, succinic acid, glycolic acid, and glycine.

[0034] The method for detecting Brucella infection serum metabolomics based on laser dissociation mass spectrometry proposed in this application has the following beneficial effects:

[0035] 1. First, the metabolic fingerprint profiles in the serum of brucellosis patients and healthy subjects were determined by iron particle-assisted laser desorption / ionization mass spectrometry. Then, the combination of biomarkers was screened out by machine learning. A biomarker combination for brucellosis diagnosis was constructed, comprising glycerol, lactic acid, succinic acid, glycolic acid, and glycine. This approach addresses the limitations of traditional methods and holds promise for clinical application in brucellosis diagnosis. The serum biomarker combination achieved accurate diagnosis of brucellosis (AUC = 0.946), demonstrating the low cost of this method. Utilizing iron nanoparticle-assisted laser desorption / ionization mass spectrometry, 300 samples can be processed per hour, with a detection time of approximately [time value missing] per sample. The method exhibits high repeatability, a coefficient of variation of less than 10%, and accurate results. Four machine learning algorithms (Ridge Regression, Decision Tree, Adaptive Augmentation, and Random Forest) were used on the training set for model evaluation, resulting in a more accurate final diagnostic model. Five differentially expressed metabolites were selected using average signal intensity, T-test, and Ridge score, ensuring the reliability of the selected metabolite results and their applicability to brucellosis diagnosis.

[0036] 2. Recording serum metabolic fingerprints using iron nanoparticle-enhanced laser desorption / ionization (NPELDI) mass spectrometry. Iron nanoparticles enhance the thermally driven desorption and cationization of serum metabolites, promoting highly sensitive recording of serum metabolite fingerprints in mass spectrometry. Rapid detection was achieved with simple pretreatment (10-fold dilution) and minimal biological sample consumption (1.5 μL), requiring only approximately 10 minutes. Attached Figure Description

[0037] The accompanying drawings are provided to further explain this application and form part of the specification. They are used together with the embodiments of this application to explain the application and do not constitute a limitation thereof. In the drawings:

[0038] Figure 1 These are representative serum metabolic profiles of the brucellosis group and the healthy control group in Example 1;

[0039] Figure 2 This is a heatmap of 113 m / z signals corresponding to the preprocessed metabolic fingerprint spectra of serum samples from 131 brucellosis patients and healthy volunteers in Example 1.

[0040] Figure 3 This is the diagnostic performance of the machine learning model based on a combination of 5 metabolic biomarkers in Example 1 on the training set;

[0041] Figure 4 This is the diagnostic performance of the machine learning model based on a combination of 5 metabolic biomarkers in Example 1 on the validation set. Detailed Implementation

[0042] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application.

[0043] Unless otherwise specified, the examples and comparative examples are parallel experiments with the same components, component content, preparation steps, and preparation parameters. The experimental methods in the following examples are conventional methods unless otherwise specified. Unless otherwise specified, the experimental materials used in the following examples are all purchased from commercial channels.

[0044] Unless otherwise specified, all reagents should be used as is without further purification.

[0045] The iron nanoparticles, with a particle size of 200-300 nm, were synthesized via a solvothermal method. The matrix-assisted laser desorption / ionization mass spectrometry (MAMS) detection conditions were as follows: metabolic fingerprint detection was performed in positive ion mode, with a detection range of 100-1000 Da, a laser wavelength of 355 nm, and a laser frequency of 2000 Hz; the delay time was 150 ns; the accelerating voltage was 20 kV; the repetition rate of the delayed extraction was 1000 Hz; and each analysis accumulated 2000 laser irradiations.

[0046] Example 1:

[0047] Serum metabolic fingerprint data were acquired using iron nanoparticle-enhanced laser desorption / ionization time-of-flight mass spectrometry.

[0048] Step 1: Instrument and reagent preparation: matrix-assisted laser desorption / ionization mass spectrometry, serum sample, deionized water, matrix (iron nanoparticles);

[0049] Step 2: Dilute the serum sample 10 times with deionized water;

[0050] Step 3: Prepare a matrix solution of 1 mg / mL using deionized water with iron nanoparticles;

[0051] Step 4: Sample preparation is performed on the mass spectrometry target plate. 1.5 μL of each diluted serum sample is spotted and dried at room temperature.

[0052] Step 5: Prepare the matrix on the mass spectrometry target plate, spotting 1.5 μL of each matrix solution and drying at room temperature;

[0053] Step 6: Acquire serum metabolomics data using laser desorption / ionization mass spectrometry (positive ion mode acquisition);

[0054] Step 7: Following Step 6, metabolomics data were collected from 131 serum samples (66 brucellosis patients and 65 healthy controls). A comparison of typical mass spectrometry peaks between serum samples from brucellosis patients and healthy volunteers is shown below. Figure 1 ;

[0055] Step 8: Preprocessing of metabolomics data from 131 serum samples included data resampling, spectral smoothing, baseline correction, feature extraction, peak alignment, and missing value imputation, resulting in 113 metabolic features in positive ion mode. The positive ion mode metabolic fingerprints of the 131 serum samples are shown below. Figure 2 .

[0056] Example 2

[0057] Machine learning is used to apply serum metabolomics data to achieve accurate prediction of brucellosis:

[0058] Step 1: Machine Learning Environment Preparation: Python 3.7;

[0059] Step 2: Divide the 131 serum samples into a training set (90 serum samples, brucellosis / healthy control = 50 / 40) and a test set (41 serum samples, brucellosis / healthy control = 16 / 25);

[0060] Step 3: Evaluate the model using four machine learning algorithms (Algorithm 1 Ridge Regression, Algorithm 2 Decision Tree, Algorithm 3 Adaptive Augmentation, Algorithm 4 Random Forest) on the training set and select the final prediction model; the algorithm diagnostic performance is shown in Table 1.

[0061] Table 1

[0062] Predictive Model Area under the curve Sensitivity Specificity accuracy Algorithm 1 0.9 0.88 0.9 0.889 Algorithm 2 0.795 0.8 0.775 0.789 Algorithm 3 0.787 0.8 0.75 0.79 Algorithm 4 0.887 0.86 0.9 0.867

[0063] As shown in Table 1, Algorithm 1 (Ridge Regression) has the highest area under the curve and sensitivity; therefore, Algorithm 1 is selected as the final prediction model. The ROC curve evaluation results of the machine learning models are shown in [Table 1]. Figure 3 The model's visual prediction results on the test set can be found in [link to test set]. Figure 4 .

[0064] Example 3

[0065] Screening and identification of metabolic biomarkers in serum:

[0066] Step 1: Perform intensity analysis on the serum fingerprint profiles of the brucellosis group and the healthy control group, compare the average intensity of the serum fingerprint profile features, and screen out the features with an average intensity greater than or equal to 1000.

[0067] Step 2: Student's t-test was performed on the serum fingerprint profiles of the brucellosis group and the healthy control group to compare whether there were significant differences in metabolite levels between the brucellosis group and the healthy control group, and features with P less than 0.05 were screened.

[0068] Step 3: Use the Ridge algorithm to select features from 113 m / z signals on the training set for the serum fingerprint spectrum, calculate the contribution of each feature to the model prediction results, and select features with a contribution greater than or equal to 0.2.

[0069] Step 4: Cross-match the features selected by the mean intensity, Student's t-test, and Ridge important feature score to finally select 5 differential metabolites.

[0070] Step 5: Based on the differential metabolites screened in Step 4, their corresponding molecular formulas were determined, and the metabolites were preliminarily identified as glycerol, lactic acid, succinic acid, glycolic acid, and glycine by searching the human metabolome database. The test results are shown in Table 1.

[0071] Step 6: To verify the necessity of the metabolite combination, diagnostic models were established using single metabolites (glycerol, lactic acid, succinic acid, glycolic acid, glycine) and combinations (glycerol + lactic acid + succinic acid + glycolic acid + glycine). The results are shown in Tables 2 and 3.

[0072] Table 2

[0073]

[0074]

[0075] Table 3

[0076] AUC glycerin 0.849 lactic acid 0.769 Succinic acid 0.844 glycolic acid 0.854 glycine 0.795 Glycerin + Lactic acid + Succinic acid + Glycolic acid + Glycine 0.946

[0077] The results in Tables 2 and 3 show that the diagnostic efficacy was optimal only when all five metabolites were used in combination. The combined diagnostic efficacy of glycerol, lactate, succinic acid, glycolic acid and glycine (AUC = 0.946) was significantly higher than that of a single metabolite (AUC range 0.769-0.849).

[0078] The specific embodiments of this application have been described in detail above, but they are merely examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications and substitutions to this application are also within the scope of this application. Therefore, all equivalent changes and modifications made without departing from the spirit and scope of this application should be covered within the scope of this application.

Claims

1. A method for detecting Brucella infection by establishing a serum metabolic profile based on laser dissociation mass spectrometry, characterized in that, Includes the following steps: Step 1: Use iron nanoparticle-assisted laser desorption / ionization mass spectrometry to perform metabolic detection on serum samples and obtain serum metabolic fingerprint spectrum; Step 2: After preprocessing the metabolic fingerprint spectrum obtained in Step 1, a diagnostic model is constructed using machine learning algorithms, and the optimal model is selected through cross-validation. Step 3: Use the model from Step 2 to perform sample diagnosis and screen differentially expressed metabolites as biomarkers.

2. The detection method according to claim 1, characterized in that, Step 1 further includes: The serum sample was diluted 10 times with deionized water and then spotted onto the target plate with an equal volume of 1 mg / mL iron nanoparticle matrix solution. Mass spectrometry data acquisition was performed in positive ion mode, with each sample consuming 1.5 μL of serum.

3. The detection method according to claim 1, characterized in that, The data preprocessing includes: data resampling, spectral smoothing, baseline correction, feature extraction, peak alignment, and missing value filling; after preprocessing, 113 characteristic peaks are obtained in the positive ion mode.

4. The detection method according to claim 1, characterized in that, The machine learning algorithm includes: The samples are divided into training and test sets; A diagnostic model is constructed using the ridge regression algorithm.

5. The detection method according to claim 1, characterized in that, The screening of differentially expressed metabolites as biomarkers includes: Select characteristic peaks with an average signal strength ≥1000; Differential metabolites with p < 0.05 were screened using the Student's-test. Metabolites with a Ridge importance score ≥0.2 were analyzed; Finally, cross-validation yielded a combination of glycerol, lactic acid, succinic acid, glycolic acid, and glycine.

6. The detection method according to claim 1, characterized in that, The serum samples included brucellosis serum samples and serum samples from healthy control groups.

7. The detection method according to claim 1, characterized in that, The metabolic detection parameters include: mass spectrometry detection of metabolic fingerprints uses positive ion mode, detection range is set to 100-1000 Da, laser wavelength is 355 nm, laser frequency is 2000 Hz; delay time is 150 ns; acceleration voltage is 20 kV, delay extraction repeatability is 1000 Hz; and each analysis accumulates 2000 laser irradiations.

8. The detection method as described in claim 1, characterized in that, Step 2 also includes: Step 2.1, Machine Learning Environment Preparation: Python 3.7; Step 2.2: Divide the serum samples into a training set and a test set; Step 2.3: Train four models on the training set using machine learning algorithms; evaluate the models using cross-validation; visualize the prediction results on the test set and select the final prediction model.

9. The detection method according to claim 8, characterized in that, The four model training algorithms in step 2.3 include Algorithm 1 Ridge Regression, Algorithm 2 Decision Tree, Algorithm 3 Adaptive Boosting, and Algorithm 4 Random Forest.

10. The detection method according to claim 1, characterized in that, Step 3 specifically includes: Step 3.1: Perform intensity analysis on the serum fingerprint spectrum, compare the average intensity of the serum fingerprint spectrum features, and screen out the features with an average intensity greater than or equal to 1000. Step 3.2: Perform Student's t-test analysis on the serum fingerprint profiles of the brucellosis group and the healthy control group to compare whether there are significant differences in metabolite levels and screen for features with P less than 0.

05. Step 3.3: Use the Ridge algorithm to select features from 113 m / z signals on the training set for the serum fingerprint spectrum, calculate the contribution of each feature to the model prediction results, and select features with a contribution greater than or equal to 0.

2. Step 3.4: Perform cross-matching on the metabolites selected by mean intensity, Student's t-test, and Ridge important feature score to finally screen differential metabolites and identify metabolic biomarkers; Step 3.5: Based on the differential metabolites screened in Step 3.4, they were identified as glycerol, lactic acid, succinic acid, glycolic acid, and glycine.