Microbial marker for detecting heart failure as well as screening method and application thereof

Abstract

The invention relates to a microbial marker for detecting heart failure as well as a screening method and application of the microbial marker. The microbial marker comprises a Phocaecola A858004 bacterium genus, an Acetafactor bacterium genus, an Enteroclock bacterium genus, an Alistipes A871400 bacterium genus, an Anerobutyricum bacterium genus, a BlautiaA141781 bacterium genus and a Fimenecus bacterium genus. The invention also relates to a method for screening the microbial marker and application of the microbial marker. According to the present invention, the change characteristics of the intestinal microorganisms of the heart failure patient are researched based on the 16S rRNA sequencing data to obtain the heart failure related microorganism marker, such that the important significance and the application value are provided for supplementing the heart failure biological marker.

Description

Technical Field

[0001] The present invention belongs to the technical field of biomedicine, and relates to a microbial biomarker for detecting heart failure, a screening method thereof and an application thereof. Background Art

[0002] Heart failure (HF) is a complex clinical syndrome. The "National Heart Failure Guidelines" points out its three major connotations: (1) Abnormalities in cardiac structure and function directly damage ventricular diastolic function and systolic function; (2) Manifestation of corresponding clinical symptoms and signs related to heart failure; (3) A significant increase in the level of natriuretic peptide, or cardiogenic pulmonary and systemic congestion revealed by imaging examinations, or an increase in ventricular filling pressure confirmed by hemodynamic examinations. The early clinical manifestations of heart failure are chest tightness, shortness of breath, fatigue, etc., and the late clinical manifestations are dyspnea, orthopnea, etc., which seriously affect the quality of life of patients and their families and also bring a heavy burden to the social economy.

[0003] In clinical practice, echocardiography, as the preferred imaging examination method, has the advantages of non-invasiveness and convenience. However, in some complex cases, its diagnostic accuracy may be limited, and it is difficult to comprehensively capture the subtle changes in cardiac function. Endomyocardial biopsy (EMB) and invasive hemodynamic monitoring, as more in-depth evaluation methods, can provide more detailed information, but due to their invasiveness, patient acceptance and safety become considerations. In addition, genetic testing and cardiac biomarker monitoring have added new perspectives to the diagnosis of heart failure, such as the widespread use of brain natriuretic peptide (BNP) and N-terminal pro-brain natriuretic peptide (NT-proBNP) in clinical practice. However, the existing biomarkers BNP and NT-proBNP, although their expressions are consistent with the severity of heart failure, are not specific diagnostic indicators, and blood samples of patients need to be collected, and the detection cost is relatively high.

[0004] The gut microbiota, as an indispensable part of the human body, jointly maintains the body's homeostasis by closely cooperating with the host. More and more evidence shows that the biodiversity and stability of the gut microbiota in heart failure patients have decreased significantly. With the rapid development of metagenomics technology, the method of constructing a disease diagnosis model using fecal sample microbiome sequencing data has become more and more mature, and gut microbiota has the potential to be used as a diagnostic and predictive biomarker for various diseases. Moreover, the diagnostic method using feces as the test sample has the significant advantage of non-invasiveness. Therefore, microorganisms are expected to become biomarkers for the diagnosis of heart failure, which can not only enrich the existing cardiac biomarker system, but also provide more diverse and accurate options for the construction of diagnosis models. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the purpose of the present invention is to provide a microbial biomarker for detecting heart failure, as well as a screening method and application thereof.

[0006] To achieve the purpose of this invention, the following technical solutions are adopted:

[0007] In the first aspect, the present invention provides a microbial biomarker for detecting heart failure, and the microbial biomarker includes the genus Phocaeicola_A_858004, the genus Acetatifactor, the genus Enterocloster, the genus Alistipes_A_871400, the genus Anaerobutyricum, the genus Blautia_A_141781, and the genus Fimenecus.

[0008] The present invention has discovered for the first time the biomarkers of gut microbiota in heart failure. By detecting the relative abundance of gut microbiota, the probability of heart failure occurrence can be predicted based on the differential characteristics of microbiota between patients with the disease and normal populations. The heart failure-related biomarkers are detected based on gut microbiota sequencing data, with accurate and safe results and a non-invasive sampling method.

[0009] The NCBI Taxonomy (https: / / www.ncbi.nlm.nih.gov / taxonomy) database is a curated and standardized biological classification and naming system that covers the taxonomic information of all organisms in public sequence databases. It provides a unique scientific name and taxonomic hierarchical information for each species, including kingdom (k), phylum (p), class (c), order (o), family (f), and genus (g). The present invention has obtained the specific information of the above 7 biomarkers therefrom.

[0010] Phocaeicola_A_858004 is a Gram-negative anaerobic bacillus that widely exists in the human gut (https: / / www.ncbi.nlm.nih.gov / Taxonomy / Browser / wwwtax.cgi?mode=Info&id=909656&lvl=3&lin=f&keep=1&srchmode=1&unlock), and its detailed information is k__Bacteria|p__Bacteroidota|c__Bacteroidia|o__Bacteroidales|f__Bacteroidaceae|g__Phocaeicola_A_858004.

[0011] Acetatifactor is a bacterium belonging to the phylum Firmicutes, class Clostridia, family Lachnospiraceae (https: / / www.ncbi.nlm.nih.gov / Taxonomy / Browser / wwwtax.cgi?mode=Info&id=1427378&lvl=3&lin=f&keep=1&srchmode=1&unlock), with the detailed information of k__Bacteria|p__Firmicutes_A|c__Clostridia_258483|o__Lachnospirales|f__Lachnospiraceae|g__Acetatifactor.

[0012] Enterocloster is a Gram-positive anaerobic bacillus belonging to the phylum Firmicutes, class Clostridia (https: / / www.ncbi.nlm.nih.gov / Taxonomy / Browser / wwwtax.cgi?mode=Info&id=2719313&lvl=3&lin=f&keep=1&srchmode=1&unlock), with the detailed information of k__Bacteria|p__Firmicutes_A|c__Clostridia_258483|o__Lachnospirales|f__Lachnospiraceae|g__Enterocloster.

[0013] Alistipes_A_871400 is a Gram-negative bacterium of the phylum Bacteroidota (https: / / www.ncbi.nlm.nih.gov / Taxonomy / Browser / wwwtax.cgi?mode=Info&id=239759&lvl=3&lin=f&keep=1&srchmode=1&unlock), with the detailed information of k__Bacteria|p__Bacteroidota|c__Bacteroidia|o__Bacteroidales|f__Rikenellaceae|g__Alistipes_A_871400.

[0014] Anaerobutyricum is a common intestinal bacterium, belonging to the phylum Firmicutes, class Clostridia, family Lachnospiraceae (https: / / www.ncbi.nlm.nih.gov / Taxonomy / Browser / wwwtax.cgi?mode=Info&id=2569097&lvl=3&lin=f&keep=1&srchmode=1&unlock), with detailed information of k__Bacteria|p__Firmicutes_A|c__Clostridia_258483|o__Lachnospirales|f__Lachnospiraceae|g__Anaerobutyricum.

[0015] Blautia_A_141781 is a class of strict anaerobic bacteria with probiotic properties, widely present in the feces and intestines of mammals (https: / / www.ncbi.nlm.nih.gov / Taxonomy / Browser / wwwtax.cgi?mode=Info&id=572511&lvl=3&lin=f&keep=1&srchmode=1&unlock), with detailed information of k__Bacteria|p__Firmicutes_A|c__Clostridia_258483|o__Lachnospirales|f__Lachnospiraceae|g__Blautia_A_141781.

[0016] Fimenecus is a Gram-negative anaerobic bacterium, belonging to the phylum Firmicutes, class Clostridia, family Oscillospiraceae (https: / / www.ncbi.nlm.nih.gov / Taxonomy / Browser / wwwtax.cgi?mode=Info&id=2840564&lvl=3&lin=f&keep=1&srchmode=1&unlock), with detailed information of k__Bacteria|p__Firmicutes_A|c__Clostridia_258483|o__Oscillospirales|f__Acutalibacteraceae|g__Fimenecus.

[0017] In a second aspect, the present invention provides an application of the microbial marker for detecting heart failure according to the first aspect in the preparation of a product for detecting heart failure.

[0018] In a third aspect, the present invention provides a kit for detecting heart failure, and the kit includes reagents for detecting the microbial markers for detecting heart failure according to the first aspect.

[0019] In a fourth aspect, the present invention provides a method for screening microbial markers of heart failure according to the first aspect, and the method includes:

[0020] (1) Obtain microbial sequencing data of a disease group and a normal group, and perform preprocessing;

[0021] (2) Quantify and annotate the relative abundances of microorganisms in the preprocessed microbial sequencing data;

[0022] (3) Perform differential analysis on the relative abundances of microorganisms in the disease group and the normal group to obtain microorganisms with significant differences;

[0023] (4) Screen the microorganisms with significant differences, and that's it.

[0024] Preferably, the preprocessing includes:

[0025] (1) After removing the PCR primer sequences from the sequencing data using Cutadapt, remove the sliding windows with an average base quality lower than 20 and the fragments with a reads fragment length less than 100;

[0026] (2) Denoise and remove chimeras from the sequencing data processed in step (1) using the DADA2 tool.

[0027] Preferably, the quantification and annotation of the relative abundances of microorganisms include:

[0028] (1) Perform ASV clustering to generate a table of relative abundances of microorganisms;

[0029] (2) Align the ASV table with the Greengenes2 database to determine the species.

[0030] For the alignment with the Greengenes2 database in step (2), the specific file used is gg_2022_10_backbone_full_length.nb.qza.

[0031] Preferably, the screening of the differential microorganisms in step (4) includes: constructing a machine learning classification model for the differentially significant microorganisms obtained in step (3), and retaining the differential microorganisms with a single feature AUC greater than 0.5.

[0032] In a fifth aspect, the present invention provides an application of the microbial markers for detecting heart failure according to the first aspect in constructing a heart failure diagnosis model and / or preparing a heart failure diagnosis device.

[0033] In a sixth aspect, the present invention provides a heart failure diagnosis model, which is a neural network model. The input samples include the relative abundances of the microbial markers described in the first aspect. The activation function of the hidden layer is ReLU, the activation function of the output layer is Sigmoid, the optimizer is Adam, the learning rate is 0.005, the neuron structure of the hidden layer is 100×100, the number of neuron layers is 4, the neuron structure is fully connected, the proportion of randomly inactivated neurons in each hidden layer is 0.2, and the output sample is the probability of heart failure;

[0034] The judgment criterion for having heart failure is that the probability value is greater than 0.5.

[0035] In a seventh aspect, the present invention provides a heart failure diagnosis device, which includes a detection unit and an evaluation unit;

[0036] The detection unit is used to perform the following:

[0037] Detect the relative abundances of the microbial markers of heart failure described in the first aspect in the sample to be tested;

[0038] The evaluation unit is used to perform the following:

[0039] Input the relative abundances of the microbial markers detected by the detection unit into the heart failure diagnosis model described in the sixth aspect to output a predicted probability;

[0040] The judgment criterion for having heart failure is that the probability value is greater than 0.5.

[0041] Preferably, the sample to be tested includes feces.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention discovers for the first time the biomarkers of gut microbiota in heart failure. By detecting the relative abundances of gut microbiota and predicting the probability of heart failure according to the differential characteristics of microbiota between patients with the disease and normal people. The present invention has been proven by a large number of experiments that the biomarkers screened in this application have high accuracy for the diagnosis of heart failure.

[0044] 2. The heart failure-related biomarkers of the present invention are detected based on gut microbiota sequencing data, with accurate results, safety, and a non-invasive sampling method.

[0045] 3. The present invention proposes a new method for screening heart failure biomarkers. By simply extracting the relative abundances of microbiota, and then through strict data screening, noise reduction processing, and experimental verification, high-efficiency heart failure biomarkers can finally be screened out.

[0046] 4. The present invention further provides a method for constructing a heart failure diagnosis model. Based on the biomarkers screened by the present invention, a model with higher specificity, better screening efficiency and accuracy can be constructed through the heart failure model construction method, so as to more effectively perform non-invasive diagnosis of heart failure.

[0047] 5. The heart failure-related biomarkers of the present invention can be used to prepare heart failure diagnostic reagents or kits, which can comprehensively and comprehensively obtain the changes in biological functions under disease states and be used for the diagnosis of heart failure patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is the workflow diagram of the method for screening heart failure biomarkers of the present invention.

[0049] Figure 2 is the 10-fold cross-validation result diagram of the optimal microorganism combination in Example 1 of the present invention.

[0050] Figure 3 is the validation result diagram of the optimal microorganism combination in the test set in Example 2 of the present invention.

[0051] Figure 4 is the result diagram of the disease specificity evaluation experiment in Example 3 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] The technical solutions of the present invention will be further described below through specific embodiments. Those skilled in the art should understand that the embodiments are only for helping to understand the present invention and should not be regarded as specific limitations on the present invention.

[0053] Example 1

[0054] 1.1. Data collection

[0055] According to the subject inclusion criteria and relevant human ethics regulations, subject enrollment and sample collection were carried out at Anzhen Hospital Affiliated to Capital Medical University. A total of 200 samples were included, including 150 heart failure samples and 50 healthy controls, and the heart failure patients and healthy subjects were matched in age and gender. The collected fecal samples were subjected to 16S rRNA sequencing.

[0056] The specific subject inclusion criteria are as follows:

[0057] 1) Age ≥ 45 years old;

[0058] 2) In the heart failure patient group, patients with a previous definite diagnosis of myocardial infarction and / or angina pectoris, NYHA heart function grade ≥ III and / or LVEF < 40%

[0059] 3) Electrocardiogram of healthy subjects in the past three months showed normal, coronary negative, and no organic heart disease.

[0060] The case exclusion criteria are as follows:

[0061] 1) Have undergone major intestinal surgery within 5 years;

[0062] 2) Inflammatory bowel disease;

[0063] 3) Acute gastroenteritis;

[0064] 4) Clostridium difficile infection (recurrence) or Helicobacter pylori infection;

[0065] 5) Persistent or chronic diarrhea;

[0066] 6) Chronic constipation;

[0067] 7) Peptic ulcer;

[0068] 8) Gastric or intestinal polyps;

[0069] 9) Digestive tract tumors;

[0070] 10) Irritable bowel syndrome;

[0071] 11) Acute and chronic cholecystitis, hepatitis;

[0072] 12) Have taken or injected antibiotics, probiotic drugs in the past three months;

[0073] 13) Terminal diseases;

[0074] 14) Have been included in other ongoing clinical trials.

[0075] 1.2 Data preprocessing

[0076] Use Cutadapt (https: / / github.com / marcelm / cutadapt / ) to remove adapter primers from the sequencing data. The specific parameters are "cutadapt -g ADAPTER_FWD -G ADAPTER_REV --minimum-length 100 --quality-cutoff 20" (remove the primer sequences at the 5' end and 3' end of the reads; remove the sliding window with an average base quality lower than the threshold (20); the minimum length of the retained reads is 100).

[0077] Then, the qiime tools import command of QIIME 2 (https: / / qiime2.org / ) software was used to import the sequencing data into a data format that can be analyzed by QIIME 2. Immediately afterwards, the qiime dada2 denoise-paired command was used for noise reduction and chimera removal, with specific parameters being "--p-trim-left-f 0 --p-trunc-len-f 241 --p-trim-left-r 0 --p-trunc-len-r 251" (trimming and truncating the forward read sequences, retaining sequences with a length of at least 241 base pairs; trimming and truncating the reverse read sequences, retaining sequences with a length of at least 251 base pairs).

[0078] 1.3. Quantitative analysis and annotation of microbial relative abundance

[0079] First, the qiime tools export command of QIIME 2 was used to cluster the ASVs of microorganisms. Subsequently, the biom convert command of BIOM Format (https: / / pypi.org / project / biom-format / ) was used to convert the ASV table into a TSV file that is convenient to read. Finally, QIIME 2 was used for species classification annotation of microbial sequences, with specific parameters being "qiime feature-classifier classify-sklearn --i-classifier Greengenes2" (species annotation based on machine learning algorithms in the scikit-learn library; based on the sequences and classification information in the Greengenes2 database).

[0080] 1.4. Screening for differential gut microorganisms

[0081] We used the R package LEfSe (https: / / huttenhower.sph.harvard.edu / lefse / ) to select differential gut microorganisms, and a total of 54 differential microbial genera were obtained. The LDA values of the specific differential microorganisms are shown in Table 1.

[0082] Table 1

[0083]

[0084]

[0085] 1.5. Screening for gut microbial diagnostic markers for heart failure

[0086] Screening of diagnostic markers for all differential microorganisms using the TensorFlow machine learning framework and SHAP. First, a feedforward neural network model was constructed based on the relative abundances of differential gut microorganisms using TensorFlow; then, SHAP was used to interpret the output of the machine learning model and evaluate the importance of calculating microbial features; finally, 7 microbial genera with a single feature AUC greater than 0.5 were retained as the optimal diagnostic marker combination, and the average relative abundances of the 7 optimal microbial genera are shown in Table 2.

[0087] Table 2

[0088] Genus name of microorganism Average relative abundance of normal control group Average relative abundance of disease group Phocaeicola_A_858004 0.2620 0.1030 Acetatifactor 0.0033 0.0105 Enterocloster 0.0159 0.0049 Alistipes_A_871400 0.0077 0.0216 Anaerobutyricum 0.0040 0.0085 Blautia_A_141781 0.0207 0.0348 Fimenecus 0.0141 0.0225

[0089] 1.6. Construction and evaluation of the diagnostic model

[0090] For the optimal gut microbial markers, we first optimized the hyperparameters of the neural network model, mainly including: the number of neural network layers, the number of neurons, the learning rate, etc. The average AUC, sensitivity, and specificity of the 10-fold cross-validation of the diagnostic model for a single microbial feature are shown in Table 3. According to the order of the microbial feature AUC, the number of features was increased in turn, and the average AUC, sensitivity, and specificity of the 10-fold cross-validation of the diagnostic model for multi-microbial features obtained are shown in Table 4.

[0091] Table 3

[0092] Genus name of microorganism AUC Sensitivity Specificity Phocaeicola_A_858004 0.8427 0.6833 0.8250 Acetatifactor 0.7747 0.7167 0.6583 Enterocloster 0.7691 0.7250 0.7000 Alistipes_A_871400 0.7615 0.5500 0.7833 Anaerobutyricum 0.7490 0.6750 0.5167 Blautia_A_141781 0.7264 0.7500 0.5917 Fimenecus 0.7000 0.5917 0.7167

[0093] When the number of microbial features reached 7, the model performance reached stability. At this time, the minimum feature combination for constructing the heart failure diagnostic model was obtained, namely the combination of 7 microbial genera: Phocaeicola_A_858004, Acetatifactor, Enterocloster, Alistipes_A_871400, Anaerobutyricum, Blautia_A_141781, Fimenecus. The highest average 10-fold cross-validation AUC of the combined microbial feature model can reach 0.87. The results of the 10-fold cross-validation of the combined microbial feature model within different queues are as Figure 2 described, and the AUC, sensitivity, and specificity of each fold are shown in Table 5. It can be seen that the combined microbial feature model is superior to the single microbial feature model. Therefore, the combination of 7 microbial genera, namely Phocaeicola_A_858004, Acetatifactor, Enterocloster, Alistipes_A_871400, Anaerobutyricum, Blautia_A_141781, Fimenecus, was determined as the optimal microbial marker combination scheme.

[0094] Table 4

[0095] Number of microbial features Average AUC Sensitivity Specificity 1 0.76 0.89 0.24 2 0.76 0.30 0.94 3 0.83 0.42 0.93 4 0.85 0.38 0.88 5 0.84 0.38 0.98 6 0.88 0.68 0.91 7 0.87 0.73 0.90

[0096] Table 5

[0097]

[0098]

[0099] Example 2

[0100] Test set verification

[0101] For microbial sequencing data, we randomly divided the dataset into a training set and a test set at a ratio of 8:2, with 20% of the data as the test set. Based on the optimal microorganisms we confirmed (Phocaeicola_A_858004, Acetatifactor, Enterocloster, Alistipes_A_871400, Anaerobutyricum, Blautia_A_141781, Fimenecus, a total of 7 microbial genera), we used the test set data to verify the microbial characteristics. The AUC of the characteristics of the combination of 7 microbial genera in the test set reached 0.87, and the results are as Figure 3 shown. The above results indicate that the diagnostic markers and non-invasive diagnostic models in this example have high accuracy and high clinical value.

[0102] Example 3

[0103] Specificity verification

[0104] Experimental materials: We collected the intestinal microbial sequencing data of other diseases except heart failure in the database for specificity verification, including Coronary Artery Disease (CAD; cohort GSE242047, the number of disease samples is 52, and the number of healthy control samples is 52), Autism Spectrum Disorder (ASD; cohort PRJNA624252, the number of disease samples is 30, and the number of healthy control samples is 20), and Crohn's Disease (CD; cohort PRJEB38969, the number of disease samples is 23, and the number of healthy control samples is 20).

[0105] Experimental method: For sequencing data of different diseases, based on the optimal combination of gut microbial markers we confirmed, models were constructed for each disease respectively, and the results of 10-fold cross-validation were obtained. That is, the data of each disease were randomly and evenly divided into 10 folds internally. Each fold was used as the test set in turn, and the remaining 9 folds were used as the training set for model construction to obtain the average AUC of 10 folds.

[0106] Experimental results: As Figure 4 shown, the upper edge of each box plot of each disease in the figure is the highest AUC among the 10 folds, the lower edge is the lowest AUC among the 10 folds, the upper and lower edges of the box body are the two quartiles of the 10-fold AUC respectively, and the line in the middle of the box body is the median of the 10-fold AUC. The results of statistical tests show that the AUC of heart failure is significantly higher than that of other intestinal diseases, with significant differences, indicating that specific verification confirms that the microbial markers and diagnostic models have high specificity for heart failure, which can avoid the occurrence of false positives in clinical applications and can also assist in differential diagnosis.

[0107] The applicant declares that the present invention uses the above embodiments to illustrate a microbial marker for detecting heart failure, its screening method and application, but the present invention is not limited to the above embodiments, that is, it does not mean that the present invention must rely on the above embodiments to be implemented. Those skilled in the art should understand that any improvement of the present invention, the equivalent replacement of each raw material of the products of the present invention, the addition of auxiliary components, the selection of specific methods, etc. all fall within the protection scope and disclosure scope of the present invention.

[0108] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the technical concept scope of the present invention, various simple modifications can be made to the technical solutions of the present invention, and these simple modifications all belong to the protection scope of the present invention.

[0109] In addition, it should be noted that in the various specific technical features described in the above specific embodiments, they can be combined in any appropriate manner without conflict. To avoid unnecessary repetition, the present invention will not separately describe various possible combination methods.

Claims

1. A microbial biomarker for detecting heart failure, characterized in that, The microbial markers include the genera Phocaeicola_A_858004, Acetatifactor, Enterocloster, Alistipes_A_871400, Anaerobutyricum, Blautia_A_141781, and Fimenecus.

2. Use of the microbial marker for detecting heart failure according to claim 1 in the preparation of a product for detecting heart failure.

3. A kit for detecting heart failure, characterized in that, The kit includes reagents for detecting the microbial marker for detecting heart failure according to claim 1.

4. A screening method for heart failure microbial markers according to claim 1, characterized in that, The method includes: (1) Obtain microbial sequencing data of the disease group and the normal group, and perform preprocessing; (2) Quantify and annotate the relative abundances of microorganisms in the preprocessed microbial sequencing data; (3) Perform differential analysis on the relative abundances of microorganisms in the disease group and the normal group to obtain significantly different microorganisms; (4) Screen the significantly different microorganisms to obtain the desired ones.

5. The screening method according to claim 4, wherein The preprocessing includes: (1) After removing the PCR primer sequences from the sequencing data using Cutadapt, remove the sliding windows with an average base quality lower than 20 and the fragments with a read length less than 100; (2) Denoise and remove chimeras from the sequencing data processed in step (1) using the DADA2 tool.

6. The screening method according to claim 4 or 5, characterized in that The quantification and annotation of the relative abundances of microorganisms include: (1) Perform ASV clustering to generate a table of relative abundances of microorganisms; (2) Compare the ASV table with the Greengenes2 database to determine the species.

7. The screening method according to any one of claims 4-6, characterized in that, The screening of the differential microorganisms in step (4) includes: constructing a machine learning classification model for the significantly different microorganisms obtained in step (3), and retaining the differential microorganisms with a single feature AUC greater than 0.

5.

8. Use of the microbial marker for detecting heart failure according to claim 1 in constructing a heart failure diagnosis model and / or preparing a heart failure diagnosis device.

9. A heart failure diagnosis model, characterized in that, The heart failure diagnosis model is a neural network model. The input sample includes the relative abundances of the microbial markers according to claim 1. The activation function of the hidden layer is ReLU, the activation function of the output layer is Sigmoid, the optimizer is Adam, the learning rate is 0.005, the neuron structure of the hidden layer is 100×100, the number of neuron layers is 4, the neuron structure is fully connected, the proportion of randomly inactivated neurons in each hidden layer is 0.2, and the output sample is the probability of heart failure; The judgment criterion for having heart failure is that the probability value is greater than 0.

5.

10. A heart failure diagnosis device, characterized in that, The diagnosis device includes a detection unit and an evaluation unit; The detection unit is used to perform the following: Detect the relative abundances of the microbial markers for heart failure described in claim 1 in a test sample; The evaluation unit is used to perform the following: Input the relative abundances of the microbial markers detected by the detection unit into the heart failure diagnosis model according to claim 9, and output the predicted probability; The judgment criterion for having heart failure is that the probability value is greater than 0.5.

Images