Systems, devices, or media for diagnosing or predicting type 2 diabetes based on a combination of gut microbial markers
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGDONG MEIGE GENE TECH CO LTD
- Filing Date
- 2026-03-18
- Publication Date
- 2026-06-19
AI Technical Summary
利用本申请的肠道微生物标志物组合的丰度数据，构建机器学习模型，能够用于诊断受试者是否患有二型糖尿病或预测受试者是否具有患二型糖尿病的风险，具有巨大的临床应用价值。
Smart Images
Figure CN122245712A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of gut microbiota biomarker technology, and more particularly to systems, devices or media for diagnosing or predicting type 2 diabetes based on combinations of gut microbiota biomarkers. Background Technology
[0002] The identification of gut microbiota relies on next-generation sequencing and bioinformatics techniques.
[0003] Next-generation sequencing (NGS), also known as next-generation sequencing or high-throughput sequencing, refers to a series of technologies capable of massively parallel sequencing of millions to billions of DNA molecules. Compared to first-generation sequencing, namely the Sanger sequencing method, the core breakthrough of NGS lies in its extremely high throughput, astonishing speed, and significantly reduced cost. Instead of sequencing a single DNA fragment, it fragments the DNA and simultaneously sequences tens of thousands of fragments, thus achieving "large-scale" genome analysis. After obtaining sequence data through NGS, FASTP is typically used for quality control, followed by assembly, clustering, and species annotation using Usearch / Qiime2 to obtain the gut microbiota information of the sample.
[0004] Once gut microbiota and disease information are obtained, disease prediction models can be constructed. These models typically use various machine learning algorithms such as random forests. Logistic regression is a machine learning model that combines linear regression and a logistic compression function. The compression function can compress an input ranging from negative infinity to positive infinity into an output of 0-1. This output can be used as the class prediction probability in a binary classification model, thus enabling the application of linear regression to classification problems. Linear regression can reduce the model's prediction error and improve performance by updating the weight coefficients of each feature through gradient descent. However, to avoid overfitting, a penalty term is usually added to linear regression. The L1 penalty term is the sum of the absolute values of each element of the weight w. When the penalty term is L1, the weight coefficients of some features can be set to 0, thereby achieving feature selection. Logistic regression is generally suitable for machine learning classification problems with small to medium amounts of data. Summary of the Invention
[0005] To solve at least one of the above-mentioned technical problems, the technical solution adopted in this application is as follows.
[0006] The first aspect of this application provides the use of an abundance detection reagent for a combination of gut microbiota markers in the preparation of a kit for diagnosing or predicting type 2 diabetes, the combination of gut microbiota markers comprising Leyella , Pseudomonas , Tyzzerella , Megasphaera , Pasteurellaceae, Pseudomonadales , Selenomonadaceae , Erysipelotrichales , Coprococcus and Romboutsia .
[0007] In some embodiments of this application, the abundance is obtained using qPCR, 16S RNA gene sequencing, or metagenomic sequencing.
[0008] In some preferred embodiments of this application, the abundance detection reagent refers to a high-throughput sequencing reagent, including nucleic acid extraction, amplification and / or purification reagents.
[0009] In some embodiments of this application, the abundance detection reagent includes primers and / or probes. Further, the probes can be fabricated into a chip.
[0010] In some specific embodiments of this application, it refers to obtaining it using 16S RNA gene sequencing, specifically including: Genomic DNA was extracted from the biological sample and 16S RNA gene sequencing was performed to obtain sequencing data. The raw reads of the sequencing data are preprocessed to filter out high-quality reads, which are then compared with the 16S RNA gene reference database. Chimeric sequences are removed. Finally, the filtered sequences are clustered according to a certain method (including but not limited to NanoCLUST) to obtain multiple sequence clustering operation taxonomic units (OTUs). Taxonomy annotation is performed on the OTUs to obtain the abundance data of each microorganism.
[0011] In some embodiments of this application, the abundance is relative abundance.
[0012] In this application, the biological samples include, but are not limited to, feces, intestinal lavage fluid, and anal swab samples, preferably feces samples.
[0013] A second aspect of this application provides a method for constructing a model for diagnosing or predicting type 2 diabetes, comprising the following steps: S1, Obtain abundance data of the combination of gut microbial markers described in the first aspect of this application in biological samples of a population, said population including patients with type 2 diabetes and subjects without type 2 diabetes; S2. Using the abundance data obtained in step S1, construct a machine learning model.
[0014] In some embodiments of this application, the abundance is relative abundance.
[0015] In some embodiments of this application, the machine learning model is selected from any of the following: Logistic Regression Model, Support Vector Machine Model, Decision Tree Model, Random Forest Model, Neural Network Model, XGBoost Model, Linear Discriminant Analysis Model, GBDT Model, ADABoost Model, Naive Bayes Model, CatBoost Model, LightGBM Model, MLP Model, and ETC Model.
[0016] In some embodiments of this application, the machine learning algorithm is a logistic regression algorithm, and the machine learning model diagnoses whether a subject has type 2 diabetes or predicts whether a subject is at risk of developing type 2 diabetes based on the score obtained by the logistic regression algorithm.
[0017] A third aspect of this application provides a system for diagnosing or predicting type 2 diabetes, comprising the following modules: The data input module is used to input the abundance data of the gut microbiota biomarker combination in the obtained subject biological samples, wherein the gut microbiota biomarker combination includes Leyella , Pseudomonas , Tyzzerella , Megasphaera , Pasteurellaceae , Pseudomonadales , Selenomonadaceae , Erysipelotrichales , Coprococcus and Romboutsia ; A database storage module for storing abundance data of the gut microbiota marker combinations in biological samples of a population, including patients with type 2 diabetes and subjects without type 2 diabetes. The disease prediction module is connected to the data input module and the database storage module, respectively. It is used to construct a machine learning model using the abundance data of the gut microbiota marker combination in the biological samples of the population, and to diagnose whether the subject has type 2 diabetes or predict whether the subject is at risk of developing type 2 diabetes based on the abundance data of the gut microbiota marker combination in the biological samples of the subject obtained from the data input module.
[0018] In some embodiments of this application, the abundance is relative abundance.
[0019] In some embodiments of this application, the machine learning algorithm is a logistic regression algorithm, and the machine learning model diagnoses whether a subject has type 2 diabetes or predicts whether a subject is at risk of developing type 2 diabetes based on the score obtained by the logistic regression algorithm.
[0020] In some preferred embodiments of this application, the regression coefficients of each gut microbiome marker in the prediction model are as follows: ; In some preferred embodiments of this application, when the score is less than 0.5, the subject does not have type 2 diabetes or has a low risk of developing type 2 diabetes; when the score is greater than 0.5, the subject has type 2 diabetes or has a high risk of developing type 2 diabetes; otherwise, the subject has a moderate risk of developing type 2 diabetes.
[0021] A fourth aspect of this application provides a computer device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of any of the methods described in any of the second aspects of this application.
[0022] The fifth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of any of the methods described in the second aspect of this application.
[0023] Compared with the prior art, the invention title of this application has the following beneficial effects: Using the abundance data of the gut microbiota biomarker combination in this application, a machine learning model can be constructed that can be used to diagnose whether a subject has type 2 diabetes or predict whether a subject is at risk of developing type 2 diabetes, which has great clinical application value.
[0024] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description
[0025] The above and other objects, features, and advantages of exemplary embodiments of this application will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of this application are illustrated in the drawings by way of example and not limitation, in which: Figure 1 The ROC curve of the logistic regression model in the test set in Embodiment 1 of this application is shown; Figure 2 The score distribution of the logistic regression model on the independent validation set in Embodiment 2 of this application is shown; Figure 3 The ROC curve of the logistic regression model in the independent validation set in Embodiment 2 of this application is shown. Detailed Implementation
[0026] Unless otherwise stated, implied from the context, or as is customary in the art, all parts and percentages in this application are based on weight, and all testing and characterization methods used are concurrent with the filing date of this application. Where applicable, any patent, patent application, or disclosure relating to this application is incorporated herein by reference in its entirety, and its equivalent patent families are also incorporated herein by reference, particularly the definitions of relevant terms in the art disclosed in such documents. If any definition of a specific term disclosed in the prior art is inconsistent with any definition provided in this application, the definition provided in this application shall prevail.
[0027] To make the technical problems, technical solutions and beneficial effects solved by this application clearer, the following detailed description is provided in conjunction with embodiments.
[0028] The following examples are used to illustrate preferred embodiments of this application. Those skilled in the art will understand that the techniques disclosed in the examples represent technologies discovered by the inventors that can be used to implement this application, and therefore can be considered preferred embodiments of this application. However, those skilled in the art should understand from this specification that many modifications can be made to the specific embodiments disclosed herein, still yielding the same or similar results, without departing from the spirit or scope of this application.
[0029] 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 pertains, and all materials cited herein and referenced by them are incorporated herein by reference.
[0030] Those skilled in the art will recognize, or can learn through routine experimentation, many equivalents of the specific embodiments of the invention described herein. These equivalents will be included in the claims.
[0031] Unless otherwise specified, the experimental methods used in the following examples are conventional methods. Unless otherwise specified, the instruments and equipment used in the following examples are all conventional laboratory instruments and equipment; unless otherwise specified, the experimental materials used in the following examples were all purchased from conventional biochemical reagent stores.
[0032] Example 1: Model Construction 1. Data Acquisition In this embodiment, the data used to build the model comes from https: / / www.nature.com / articles / s41591-018-0164-x. This dataset contains data from 2008 healthy individuals and data from a varying number of patients with multiple diseases, including 585 patients with type 2 diabetes. The data includes information such as gut microbiota, where the gut microbiota information includes the names of microorganisms at the phylum, class, order, family, and genus levels, along with their corresponding relative abundances.
[0033] 2. Dataset Construction and Preprocessing The training set and the test set were obtained by splitting healthy people and type 2 diabetes patients in an 8:2 ratio. Since there was a large difference between the number of patients and healthy people in the training set, the inventors randomly sampled (random downsampling) the healthy people in the training set to make the ratio of healthy people to patients in the new training set 1:1.
[0034] 3. Model training and hyperparameter optimization On the rebalanced training set, a logistic regression model with L1 regularization (Lasso) is used. A set of candidate regularization strength parameters C values is set: [0.001, 0.003, 0.01, 0.03, 0.1, 0.3, 1, 3, 5, 7] (the larger the C value, the stronger the penalty and the simpler the model).
[0035] Evaluation was performed using 10-fold cross-validation: the training set was divided into 10 equal parts, and 9 parts were used for training and 1 part for validation in turn, for a cycle of 10 times.
[0036] The optimal parameters were selected by comparing the average performance (balanced_accuracy) of the model in 10 validations under different C values, as shown in Table 1.
[0037] Table 1: Average performance of the model in 10 validations under different C values ; As shown in Table 1, the regularization strength of C=5 achieves the best bias-variance tradeoff on this dataset.
[0038] Using the optimal parameters (C=5) and L1 penalty term determined above, the final logistic regression model is retrained on the entire balanced training set.
[0039] 4. Model Validation Validate the trained model on the test set, such as Figure 1As shown, the model's AUC (area under the curve) reaches 0.8122, indicating that the model has a good overall ability to distinguish between type 2 diabetes patients and healthy individuals. Sensitivity and specificity are 0.7265 and 0.7692, respectively, indicating that the model can correctly identify approximately 72.65% of actual patients and approximately 76.92% of actual healthy individuals. These indicators collectively demonstrate that the model has robust and balanced predictive ability.
[0040] 5. Feature Analysis and Extraction A key characteristic of L1 regularization is its tendency to compress unimportant feature coefficients to zero, thereby enabling embedded feature selection. Based on the final trained model, the coefficients corresponding to each feature are examined and further sorted in descending order of their absolute values. The absolute value of the coefficient directly measures the relative importance of that feature in predicting the risk of type 2 diabetes (the larger the absolute value, the greater the impact; the positive or negative sign indicates the direction of the impact).
[0041] From the ranking list, the top 10 features are extracted. These features represent the set of biomarkers or clinical indicators that the current logistic regression model considers most important for predicting the risk of type 2 diabetes. After removing features with a coefficient of 0, the feature list shown in Table 2 is obtained.
[0042] Table 2: Key features for predicting the risk of type 2 diabetes ; The analysis results indicate that the intestinal microorganism is *Rhizobium* spp. Leyella Pseudomonas spp. Pseudomonas Pseudomonas spp. Tyzzerella genus *Gastromycium* Megasphaera Pasteuraceae Pasteurellaceae Pseudomonas Pseudomonadales Lunaecidae Selenomonadaceae , Erysipelothrixales Erysipelotrichales spp. Coprococcus and Rombutzella genus Romboutsia It is significantly associated with the risk of type 2 diabetes and can serve as a gut microbiota biomarker for the diagnosis or prediction of type 2 diabetes.
[0043] Example 2 Independent Validation of the Model To verify the performance of the logistic regression model constructed in Example 1, the inventors further collected data from 23 independent patients with type 2 diabetes and 77 healthy individuals, and obtained the abundance levels of the gut microbiota markers screened in Example 1, as shown in Table 2.
[0044] Table 2: Abundance levels of gut microbial biomarkers in independent validation clusters ; The scores of each subject were calculated using the model constructed in Example 1, such as... Figure 2 As shown. The independent validation set is further classified using a threshold, and the results are as follows. Figure 3 As shown. From Figure 3 As can be seen, the AUC, sensitivity, and specificity of the model constructed in Example 1 in distinguishing between type 2 diabetes patients and healthy individuals in the independent validation set are 0.7820, 0.7391, and 0.7922, respectively, indicating that the model also has good discriminative ability in the independent validation set.
[0045] All references to this application are incorporated herein by reference as if each reference were individually incorporated herein by reference. Furthermore, it should be understood that after reading the foregoing teachings of this application, those skilled in the art can make various alterations or modifications to this application, and these equivalent forms also fall within the scope defined by the appended claims.
Claims
1. Use of an assay for the abundance of a combination of markers of gut microorganisms in the manufacture of a kit for the diagnosis or prognosis of type 2 diabetes, characterized in that, The combination of gut microbial markers comprises Leyella , Pseudomonas , Tyzzerella , Megasphaera , Pasteurellaceae , Pseudomonadales , Selenomonadaceae , Erysipelotrichales , Coprococcus and Romboutsia .
2. Use according to claim 1, characterized in that, The abundance is obtained by qPCR, 16S RNA gene sequencing or metagenomic sequencing.
3. A method of constructing a model for diagnosing or predicting type 2 diabetes, characterized by, The method comprises the following steps: S1, obtaining the abundance data of the intestinal microbial marker combination in the biological sample of the population, wherein the population comprises type 2 diabetes patients and non-type 2 diabetes subjects; S2, constructing a machine learning model by using the abundance data obtained in step S1.
4. The method of claim 3, wherein, The abundance is a relative abundance.
5. The method according to claims 3 and 4, characterized in that, The machine learning model is selected from any one of the following: a logistic regression model, a support vector machine model, a decision tree model, a random forest model, a neural network model, an XGBoost model, a linear discriminant analysis model, a GBDT model, an ADABoost model, a naive Bayes model, a CatBoost model, a LightGBM model, an MLP model and an ETC model.
6. The method of claim 5, wherein, The machine learning algorithm is a logistic regression algorithm, and the machine learning model is used to diagnose whether the subject has type 2 diabetes or predict whether the subject has the risk of type 2 diabetes according to the score obtained by the logistic regression algorithm.
7. A system for diagnosing or predicting type 2 diabetes, characterized by, The method comprises the following modules: a data input module for inputting obtained abundance data of a combination of gut microbial markers in a biological sample of a subject, the combination of gut microbial markers comprising Leyella , Pseudomonas , Tyzzerella , Megasphaera , Pasteurellaceae , Pseudomonadales , Selenomonadaceae , Erysipelotrichales , Coprococcus and Romboutsia ; A database storage module is configured to store the abundance data of the intestinal microbial marker combination in the biological sample of the population, wherein the population comprises type 2 diabetes patients and non-type 2 diabetes subjects; A disease prediction module is connected to the data input module and the database storage module respectively, and is configured to construct a machine learning model by using the abundance data of the intestinal microbial marker combination in the biological sample of the population, and diagnose whether the subject has type 2 diabetes or predict whether the subject has the risk of type 2 diabetes based on the abundance data of the intestinal microbial marker combination in the biological sample of the subject obtained from the data input module.
8. The system of claim 7, wherein, The abundance is a relative abundance.
9. A computer device, comprising: The method comprises the following steps: a memory configured to store a computer program; a processor configured to execute the computer program to implement the steps of the method according to any one of claims 3-6.
10. A computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and the computer program is executed by a processor to implement the steps of the method according to any one of claims 3-6.