Axillary osmidrosis skin microbial marker combination, screening method and application thereof

Abstract

This invention discloses a combination of axillary bromhidrosis skin microbial biomarkers, a screening method, and their applications, belonging to the field of bromhidrosis diagnostic technology. By collecting axillary skin surface samples from bromhidrosis patients and healthy subjects, a combination of axillary bromhidrosis skin microbial biomarkers was screened using metagenomic shotgun sequencing. A random forest diagnostic model constructed based on this combination of axillary bromhidrosis skin microbial biomarkers achieved an area under the receiver operating characteristic (AUC) of 0.9222 in testing, demonstrating a very high ability to distinguish between bromhidrosis patients and healthy individuals. This indicates that the biomarker combination can provide objective, high-resolution, systematic, and quantifiable microbiome-based diagnosis, significantly improving the accuracy, consistency, and generalizability of diagnosis, and solving the problem that existing axillary bromhidrosis biomarkers cannot specifically distinguish between bromhidrosis patients and healthy individuals.

Description

Technical Field

[0001] This invention belongs to the field of bromhidrosis diagnosis technology, specifically relating to a combination of skin microbial markers for bromhidrosis in the armpit, a screening method, and their application. Background Technology

[0002] Body odor (bromhidrosis) is a common, genetically predisposed skin appendage disorder characterized by a distinctive, pungent odor emanating from the armpits. This significantly impacts patients' social activities and mental health, reducing their quality of life. Currently, clinical diagnosis of body odor relies primarily on the physician's subjective olfactory examination, lacking objective and quantifiable biological diagnostic criteria. This subjective diagnostic approach is easily influenced by the assessor's clinical experience, testing conditions, and the patient's subjective descriptions, resulting in poor diagnostic repeatability and low consistency (consistency in diagnoses between different assessors or by the same assessor at different time points is difficult to guarantee). This is especially problematic for mild or atypical cases, easily leading to misdiagnosis or missed diagnosis, failing to meet the needs of precise clinical diagnosis.

[0003] In recent years, with the development of microbiome research, studies have gradually recognized that the composition of the human skin microbiome, especially the microbiota in the armpit, is closely related to body odor, providing a new research direction for the objective diagnosis of bromhidrosis (body odor). The armpit is moist, warm, and has abundant sweat secretion, providing a suitable microenvironment for the colonization and growth of microorganisms (especially bacteria). These bacteria colonizing the armpit can metabolize odorless precursors (such as proteins and lipids) secreted by apocrine sweat glands, generating volatile short-chain fatty acids, sulfides, and other odorous metabolites, thus forming the characteristic pungent odor of bromhidrosis. Based on this pathogenesis, screening for axillary skin microbial markers to achieve the objective diagnosis of bromhidrosis has become a research hotspot in recent years. However, existing research on skin microbial biomarkers for axillary bromhidrosis still has many shortcomings: On the one hand, existing reports mostly focus on the abundance changes of single or a few genera, lacking a systematic and panoramic analysis of the bromhidrosis-related microbial community. The research results are scattered and lack systematicity, failing to clearly define a stable and synergistic combination of "beneficial bacteria" and "pathogenic bacteria" between healthy individuals and bromhidrosis patients. In other words, a complete microbial biomarker system with high diagnostic specificity has not yet been established. On the other hand, some studies have used targeted sequencing technologies such as 16S rRNA gene sequencing to observe certain genera (such as Staphylococcus) in the axillary microbiome of bromhidrosis patients. StaphylococcusThe abundance of these microorganisms varies significantly, and most of these studies suffer from insufficient depth of research. Most studies are limited to sequencing of the variant regions of bacterial 16S rRNA genes, which can only identify microorganisms at the genus level. The resolution is limited, making it difficult to accurately identify specific bacterial species with diagnostic value and screen potential microbial biomarkers. Furthermore, it is impossible to comprehensively obtain functional gene information related to odor metabolism and fully analyze the functional potential of the microbial community, thus limiting the application of microbial biomarkers in the clinical diagnosis of bromhidrosis. Summary of the Invention

[0004] In order to overcome the shortcomings of the prior art, the present invention aims to provide a combination of skin microbial markers for axillary bromhidrosis, a screening method and its application, in order to solve the technical problem that existing axillary bromhidrosis biomarkers cannot accurately and specifically distinguish bromhidrosis patients from healthy people.

[0005] To achieve the above objectives, the present invention employs the following technical solution: The first aspect of this invention discloses a combination of skin microbial markers for axillary bromhidrosis, including Prevotella faecalis (… Faecalibacterium prausnitzii Clostridium bacteria ( Clostridiales bacteria ), *Cryptospira marines* ( Golden seal ), Parabacterium dilatatum ( Parabacteroides distasonis ), human fecal parasartella ( Parasutterella excrementihominis Corynebacterium BCW_4722 ( Corynebacterium sp. BCW_4722), Propionibacterium acnes ( Cutibacterium acnes Staphylococcus epidermidis ( Staphylococcus epidermidis ) and Sphingosomalidone spp. 3F27F9 ( Sphingomonas sp. 3F27F9).

[0006] In a second aspect, the present invention discloses the application of the above-mentioned combination of skin microbial markers for axillary bromhidrosis in the preparation of products or devices for diagnosing or assisting in the diagnosis of axillary bromhidrosis.

[0007] Preferably, *Prevotella faecalis* ( Faecalibacterium prausnitzii Clostridium bacteria ( Clostridiales bacteria ), *Cryptospira marines* ( Golden seal ), Parabacterium dilatatum ( Parabacteroides distasonis ) and human fecal parasartella ( Parasutterella human excrement The relative abundance of Corynebacterium BCW_4722 was significantly reduced in patients with bromhidrosis. Corynebacterium sp. BCW_4722), Propionibacterium acnes ( Cutibacterium acnes Staphylococcus epidermidis ( Staphylococcus epidermidis ) and Sphingosomalidone spp. 3F27F9 ( Sphingomonas sp.The relative abundance of 3F27F9 was significantly increased in patients with bromhidrosis; the statistical criteria for significance were a false discovery rate of <0.05 after Benjamini-Hochberg correction and a linear discriminant analysis score of >4.0 in LEfSe analysis.

[0008] A third aspect of the present invention discloses a kit for diagnosing or assisting in the diagnosis of axillary bromhidrosis, the kit containing reagents for detecting the aforementioned combination of skin microbial markers for axillary bromhidrosis.

[0009] A fourth aspect of the present invention discloses a product for diagnosing or assisting in the diagnosis of axillary bromhidrosis, the product comprising primers, probes, antibodies, aptamers or chips that are specific to the aforementioned combination of skin microbial markers for axillary bromhidrosis.

[0010] A fifth aspect of the present invention discloses a method for screening a combination of skin microbial markers for axillary bromhidrosis, comprising the following steps: 1) Collect axillary skin surface samples from patients with bromhidrosis and healthy subjects, extract total microbial DNA, and perform shotgun metagenomic sequencing; 2) Perform quality control and species analysis on metagenomic sequencing data to obtain a table of relative abundance of microorganisms; 3) Based on the obtained microbial relative abundance table, the microbial species that showed significant differences between the two groups of patients with bromhidrosis and healthy subjects were selected as the combination of microbial markers for axillary bromhidrosis skin.

[0011] In a sixth aspect, the present invention discloses a method for constructing a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis, wherein a machine learning model is constructed using the relative abundance of the aforementioned combination of skin microbial markers for axillary bromhidrosis as a feature, the model is trained and evaluated, and a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis is obtained.

[0012] Preferably, the machine learning model is a random forest classification model, a support vector machine model, a gradient boosting decision tree model, a logistic regression model, or a neural network model.

[0013] Preferably, the machine learning model is a random forest classification model, using the relative abundance of the combination of skin microbial markers for axillary bromhidrosis as described in claim 1 as a feature, and employing 10-fold cross-validation to train and optimize the random forest classification model, calculating the area under the receiver operating characteristic curve and performing performance evaluation, to obtain a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis.

[0014] A seventh aspect of the present invention discloses an apparatus for diagnosing or assisting in the diagnosis of axillary bromhidrosis, comprising: The detection unit is configured to perform the following: detecting the relative abundance value of each single bacterial species in a sample of the axillary skin surface to be tested, wherein the single bacterial species includes the above-mentioned combination of axillary bromhidrosis skin microbial markers; An evaluation unit is used to perform the following: inputting the abundance detected by the detection unit into a model obtained by the above-described method for constructing a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis, and outputting the probability of having bromhidrosis.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This invention provides a combination of skin microbial markers for axillary bromhidrosis, consisting of 5 types of health-related beneficial bacteria (Prevotella coli). Faecalibacterium prausnitzii Clostridium bacteria ( Clostridiales bacteria ), *Cryptospira marines* ( Golden seal ), Parabacterium dilatatum ( Parabacteroides distasonis ) and human fecal parasartella ( Parasutterella excrementihominis )) and 4 potential pathogens associated with body odor (Corynebacterium BCW_4722) Corynebacterium sp. BCW_4722), Propionibacterium acnes ( Cutibacterium acne Staphylococcus epidermidis ( Staphylococcus epidermidis ) and Sphingosomalidone spp. 3F27F9 ( Sphingomonas sp. The study comprises 3F27F9. This combination exhibits: 1) high diagnostic accuracy: a random forest diagnostic model constructed based on the axillary bromhidrosis skin microbial biomarker combination achieved an area under the receiver operating characteristic (AUC) of 0.9222 during testing, demonstrating a high ability to distinguish bromhidrosis patients from healthy individuals; 2) it provides objective diagnostic criteria: this axillary bromhidrosis skin microbial biomarker combination can serve as a standardized tool, eliminating reliance on subjective olfactory diagnosis by clinicians. The relative abundance of the selected axillary bromhidrosis skin microbial biomarker combination is used for assessment, with repeatable and quantifiable results. Furthermore, functional analysis preliminarily reveals changes in metabolic pathways behind the alterations in the biomarker microbiota. The axillary microbiota of bromhidrosis patients shows increased activity in metabolic pathways that more readily produce volatile substances, such as the degradation of valine, leucine, and isoleucine, while activity is reduced in pathways such as amino acid biosynthesis. This functionally explains the potential causes of odor production, providing a basis for understanding the microecological mechanisms of bromhidrosis and future intervention targets. Therefore, this axillary bromhidrosis skin microbial marker can be used for objective, high-resolution, systematic and quantifiable microbiome diagnosis, significantly improving the accuracy, consistency and generalizability of diagnosis, solving the problem that existing axillary bromhidrosis biomarkers cannot specifically distinguish bromhidrosis patients from healthy people, and changing the current situation where bromhidrosis diagnosis relies on subjective experience.

[0016] This invention provides a method for screening a combination of microbial biomarkers for axillary bromhidrosis (body odor) skin. 1) The method involves direct sampling of the skin surface, which is simple and non-invasive; 2) Screening for axillary bromhidrosis skin microbial biomarkers is based on metagenomic shotgun sequencing. Metagenomic shotgun sequencing can unbiasedly obtain the genetic information of all microorganisms in the sample, thereby achieving high-precision identification at the species level and comprehensive analysis of functional pathways; 3) Through systematic bioinformatics analysis (Wilcoxon test, LEfSe analysis), a combination of biomarkers containing stable and synergistically changing characteristics of 5 health-related beneficial bacteria and 4 bromhidrosis-related potential pathogens is screened, rather than using a single indicator. This method overcomes the problems of low resolution and limited field of view in existing research methods, accurately screening a group of biologically significant microbial biomarkers (including missing protective bacteria and over-proliferating potential pathogens) that can be used to distinguish bromhidrosis patients from healthy armpits.

[0017] This invention provides a method for constructing a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis. Based on the skin microbial markers of axillary bromhidrosis screened by metagenomic shotgun sequencing, and combined with machine learning algorithms, a high-precision, objective, and quantifiable model for diagnosing or assisting in the diagnosis of axillary bromhidrosis (AUC=0.9222) is constructed. It can achieve quantitative output from "microbial characteristics" to "disease risk probability", overcoming the shortcomings of traditional subjective diagnosis. Attached Figure Description

[0018] Figure 1 This is a stacked diagram of the Top 20 dominant species obtained through species composition analysis according to the present invention; where A represents the genus level and B represents the species level. Figure 2 This is a graph showing the Alpha diversity correlation index at different classification levels in this invention; where A represents the genus level, B represents the species level, 1 represents the Shannon index, and 2 represents the Simpson index; ***P<0.0001; Figure 3 The figures show the Beta diversity (PCoA) and NMDS at different taxonomic levels according to the present invention; where A represents PCoA, B represents NMDS, 1 represents the genus level, and 2 represents the species level. Figure 4 The Top 30 bacteria with differential relative abundance at the species level in this invention (A) and LEfSe analysis (B) are shown. Figure 5 The ROC curve and Mean Decrease Accuracy plot of the random forest of this invention; Figure 6 These are the KEGG pathway bar chart and GRSA significant enrichment chart of the present invention. Detailed Implementation

[0019] To enable those skilled in the art to understand the features and effects of the present invention, the following description and definitions are only general descriptions of the terms and expressions mentioned in the specification. Unless otherwise specified, all technical and scientific terms used herein have the ordinary meaning understood by those skilled in the art regarding the present invention, and in case of conflict, the definitions in this specification shall prevail.

[0020] The theories or mechanisms described and disclosed herein, whether right or wrong, should not in any way limit the scope of the invention, that is, the contents of the invention can be implemented without being limited by any particular theory or mechanism.

[0021] In this article, unless otherwise specified, “contains,” “includes,” “containing,” “has,” or similar terms cover the meanings of “composed of” and “mainly composed of,” for example, “A contains a” covers the meanings of “A contains a and others” and “A contains only a.”

[0022] For the sake of brevity, not all possible combinations of the technical features in each implementation scheme or embodiment are described herein. Therefore, as long as there is no contradiction in the combination of these technical features, the technical features in each implementation scheme or embodiment can be combined arbitrarily, and all possible combinations should be considered within the scope of this specification.

[0023] In this article, all microbial biomarkers are species known from public databases, including Corynebacterium BCW_4722 ( Corynebacterium sp. The NCBI classification identification number (Taxonomy ID) for BCW_4722 is 1972128; Sphingomonas spp. 3F27F9 ( Sphingomonas sp. The NCBI classification identification number (Taxonomy ID) for 3F27F9 is 2502209.

[0024] This invention provides a method for screening a combination of skin microbial markers for axillary bromhidrosis, comprising the following steps: 1. Sample collection and sequencing Axillary skin surface samples were collected from subjects in both the bromhidrosis patient group and the healthy control group. Subjects were required to avoid using underarm care products for at least 48 hours prior to sampling. After standardized collection using sterile swabs, total microbial DNA was extracted and subjected to shotgun metagenomic sequencing.

[0025] 2. Data Analysis and Biomarker Screening Sequencing data underwent quality control to remove host sequences. Species annotation was performed using tools such as MetaPhlAn4 to obtain a relative abundance table of microorganisms. Wilcoxon rank-sum test (FDR < 0.05) and LEfSe analysis (LDA > 4.0) were used to screen for microbial species that showed significant differences between the bromhidrosis patient group and the healthy control group, resulting in a combination of microbial biomarkers for axillary bromhidrosis skin. The aforementioned assemblage of axillary bromhidrosis skin microbial markers includes the following two categories of microorganisms that exhibit stable and significantly different abundances in the axillary skin of bromhidrosis patients and healthy individuals: 1) Microorganisms with significantly reduced relative abundance (health-related biomarkers) in patients with bromhidrosis: including Prevotella faecalis ( Faecalibacterium prausnitzii Clostridium bacteria ( Clostridiales bacteria ), *Cryptospira marines* ( Golden seal ), Parabacterium dilatatum ( Parabacteroides distasonis ) and human fecal parasartella ( Parasutterella excrementihominis ); 2) Microorganisms with significantly increased relative abundance in patients with bromhidrosis (bromhidrosis-related markers): including Corynebacterium BCW_4722 ( Corynebacterium sp. BCW_4722), Propionibacterium acnes ( Cutibacterium acnes Staphylococcus epidermidis ( Staphylococcus epidermidis ) and Sphingomonas spp. 3F27F9 ( Sphingomyelomonas sp. 3F27F9); The statistical criteria for "significance" are a false discovery rate (FDR) after Benjamini-Hochberg correction < 0.05 and a linear discriminant analysis (LDA) score > 4.0 in the LEfSe analysis.

[0026] This invention also provides a method for constructing a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis. The method uses the relative abundance of the combination of axillary bromhidrosis skin microbial markers obtained by the above method as a feature to construct a random forest classification model. Ten-fold cross-validation is used to train and optimize the parameters of the random forest classification model. The area under the receiver operating characteristic curve is calculated and its performance is evaluated to obtain a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis.

[0027] Regarding the technical solution of this invention, without departing from the core concept of "constructing a diagnostic model by analyzing differences in the axillary microbiome," those skilled in the art can foresee reasonable alternatives in the following technical aspects: Species analysis tool alternatives: In addition to MetaPhlAn4, other metagenomic species annotation workflows can be used, such as those based on Kraken2 / Bracken or sequence assembly and alignment.

[0028] Alternative methods for differential analysis: In addition to the Wilcoxon test combined with LEfSe analysis, other statistical methods or tools specifically designed for microbiome data (such as DESeq2, ANCOM-BC, etc.) can be used to screen for differentially expressed species.

[0029] Alternative classification models: In addition to random forests, other machine learning models (such as support vector machines, gradient boosting decision trees, logistic regression, or neural networks) can also be used to build classifiers based on the same combination of markers.

[0030] Functional analysis tool alternatives: In addition to HUMANN3 and KEGG, other mainstream functional databases (such as GO, COG, CAZy, etc.) and corresponding analysis workflows can also be used for functional annotation and pathway enrichment analysis.

[0031] The core premise is that the implementation of any of the above-mentioned alternative technical means must rely on the specific combination of microbial markers disclosed in this invention (i.e., the group of bacteria that show stable differences between healthy and bromhidrotic armpits) as a common basis, and their purpose is to achieve the detection, analysis or modeling of this core biological characteristic.

[0032] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading this description, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined in this application.

[0033] The following examples use instruments and equipment conventional in the art. Experimental methods in the following examples, unless otherwise specified, are generally performed under standard conditions or as recommended by the manufacturer. All raw materials used in the following examples are conventional commercially available products with specifications in the art, unless otherwise stated.

[0034] Example 1: Screening of skin microbial marker combinations for axillary bromhidrosis I. Sample Collection Strictly following the protocol requirements, a total of 85 subjects were included, including 39 cases in the bromhidrosis group (Case) and 46 healthy controls (Con). Simultaneously, a total sample size of less than 20 cases and a total sample size of more than 200 cases were used as control experiments.

[0035] All subjects signed informed consent forms and adhered to the 48-hour prohibition of medication before sampling. Sterile swabs were soaked in PBS buffer and used to standardize the scraping of skin from the mid-axillary region.

[0036] II. Metagenomic Sequencing Using Magbeads Fast DNA ® The kit extracts total DNA from the samples collected in step one. After passing quality testing, a paired-end library with an average insert fragment size of approximately 400 bp is constructed using shotgun sequencing. High-throughput sequencing is then performed on the Illumina platform, with an average sequencing data volume of no less than 6 Gb per sample. A control sample with a sequencing data volume of less than 5 Gb per sample is used.

[0037] III. Data Analysis and Biomarker Screening 1. Use fastp (v0.23.0) for quality control of the raw data and Bowtie2 (v2.3.5.1) to remove human-derived sequences.

[0038] 2. Species analysis: MetaPhlAn4 was used for species annotation to obtain a relative abundance table at the species and genus levels.

[0039] 3. Difference analysis: The Wilcoxon rank-sum test (after Benjamini-Hochberg correction, FDR<0.05) was used to compare the differences in microbial abundance between the two groups; LEfSe analysis (with LDA Score>4.0) was used to screen for marker species with significant differences between groups.

[0040] Species composition analysis results as follows Figure 1 As shown, there are fundamental differences in the axillary microbial community structure between the bromhidrosis group and the healthy group. At the genus level: in the bromhidrosis group, Corynebacterium genus ( Corynebacterium ), Propionibacterium acnes ( Cutibacterium Staphylococcus spp. Staphylococcus ) and Sphingosomalmonella spp. ( Sphingomyelomonas The relative abundance of ) was significantly increased; the abundance of genus *Faecalibacterium* ( Faecalibacterium ) and Bacteroides ( Bacteroides The relative abundance of ) decreased significantly ( Figure 1 In the middle A); at the species level: in the bromhidrosis group, Corynebacterium BCW_4722 ( Corynebacterium sp. BCW_4722), Propionibacterium acnes ( Cutibacterium acnes Staphylococcus epidermidis ( Staphylococcus epidermidis ) and Sphingosomalidone spp. 3F27F9 ( Sphingomonas sp. The relative abundance of *3F27F9* was significantly increased; *Prevotella* ( Faecalibacterium prausnitzii The relative abundance of ) decreased significantly ( Figure 1 (B)

[0041] The results of the diversity analysis are as follows Figure 2 and Figure 3As shown, the alpha diversity (Shannon index, Simpson index) in the bromhidrosis group was significantly lower than that in the healthy group (P<0.001). Figure 2 Beta diversity (PCoA, NMDS) showed significant separation between the two community structures (PERMANOVA test, species level R). 2 =0.118, P=0.0001, Figure 3 ).

[0042] Based on differential species analysis ( Figure 4 (A) and LEfSe analysis ( Figure 4 Key differentially expressed bacterial species were jointly screened by the Chinese and B groups. Among them, the bacterial species that were significantly enriched in the healthy group included: *Prevotella coli* (…). Faecalibacterium prausnitzii Clostridium bacteria ( Clostridiales bacterium ), *Cryptospira marines* ( Phocaeicola dorei ), Parabacterium dilatatum ( Parabacteroides distasonis ) and human fecal parasartella ( Parasutterella excrementihominis ); the bacterial species significantly enriched in the bromhidrosis group include: Corynebacterium BCW4722 ( Corynebacterium sp. BCW_4722), Propionibacterium acnes ( Cutibacterium acnes Staphylococcus epidermidis ( Staphylococcus epidermidis ) and Sphingomonas strain 3F27F9 ( Sphingomonas sp. 3F27F9). These species collectively form the core of the microbial biomarker ensemble of axillary bromhidrosis.

[0043] For the sequencing sample size, the total sample size is 85 cases (39 cases in the bromhidrosis group and 46 cases in the healthy group), which falls within the preferred range of 30-100 cases. At this sample size, the species difference analysis has sufficient statistical power, and the model based on it performs relatively stably in independent validation. Controls (total sample size less than 20 cases) suffer from insufficient sample representativeness, weak statistical power, and difficulty in reliably screening stable differentially expressed microbial biomarkers. The model based on it is prone to overfitting, and the AUC value drops significantly in independent validation, indicating poor generalization ability. Controls (extremely large sample size but not necessary), although exceeding 200 cases, offer the advantage of improved statistical precision, but significantly increase research costs and time, while the marginal improvement in model performance is limited. From a cost-effectiveness perspective for diagnostic applications, the preferred range of this invention achieves optimal efficiency while ensuring high performance.

[0044] For sequencing depth, shotgun metagenomic sequencing was used, with a target data volume of at least 6 Gb / sample in the experimental group. This depth is sufficient to cover low-abundance species, ensuring the completeness of species annotation and functional analysis, and laying a data foundation for high-precision models. The control group had less than 5 Gb / sample, resulting in insufficient sequencing depth. This led to the ineffective detection or inaccurate quantification of low-abundance but critical differentially expressed species (such as certain protective anaerobes), thus affecting the completeness of biomarker combinations. Models built with this incomplete data will suffer compromised diagnostic performance (AUC).

[0045] Example 2: Construction of a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis I. Feature and Label Preparation The relative abundance of the key differentially expressed species (species level) finally screened in Example 1 was used as the feature variable (X). The clinical diagnosis result of the sample (body odor = 1, healthy = 0) was used as the label (Y).

[0046] II. Model Training and Optimization 1. A random forest classifier was constructed using the randomForest package in R software (v4.3.1). All 85 samples were randomly divided into a training set (70%) and a test set (30%), with the random seed fixed to ensure repeatability. 10-fold cross-validation was used for model training and feature selection on the training set; simultaneously, logistic regression and support vector machine models were used as controls. 2. Rank all species variables by importance and add them to the model sequentially, then plot the average cross-validation error curve. Select the smallest feature set corresponding to the minimum average error plus one standard deviation as the optimal feature set.

[0047] 3. The final model is retrained on the training set using the optimal feature set. After cross-validation and feature selection, the optimal feature subset containing key differential species is determined to build the final model, resulting in a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis.

[0048] III. Model Evaluation The performance of the obtained models for diagnosing or assisting in the diagnosis of axillary bromhidrosis was evaluated on independent test sets. Receiver operating characteristic (ROC) curves were plotted, and metrics such as area under the curve (AUC), sensitivity, and specificity were calculated. Mean Decrease Accuracy was also acquired and visualized.

[0049] Evaluation results show that the random forest model exhibits excellent diagnostic performance on the test set, as indicated by the ROC curve ( Figure 5As can be seen from Figure A), its AUC value reaches 0.9222, indicating that the model has extremely high discriminative power. Feature importance ranking (Mean Decrease Accuracy) Figure 5 (B) Confirmed that the combination of axillary bromhidrosis skin microbial markers screened in Example 1 made the greatest contribution to the classification of the model.

[0050] Comparing the performance of different classification algorithms on the dataset of this invention, the Random Forest algorithm effectively handles the sparsity, high dimensionality, and nonlinear relationships of microbial abundance data, achieving an excellent AUC of 0.9222 on this dataset. The control algorithm (logistic regression), as a linear model, struggles to capture the complex interactions between microorganisms, resulting in a lower AUC value (approximately 0.65~0.70). While the control algorithm (support vector machine) can handle nonlinearity, it is sensitive to parameter and kernel function selection, has high training costs on high-dimensional data, and its final performance is slightly lower than that of Random Forest. These results indicate that, given the microbiome data characteristics involved in this invention, Random Forest is an effective and preferred algorithm for achieving high-precision classification. However, other algorithms can also achieve diagnostic purposes as alternatives, although their performance may differ.

[0051] Example 3: Microbial functional pathway analysis reveals the potential metabolic mechanism of bromhidrosis I. Functional Spectrum Analysis The metagenomic sequencing data obtained in Example 1 were functionally annotated using HUMAnN3 software to quantify the abundance of functional genes and metabolic pathways in the microbial community.

[0052] II. Differential Pathway Analysis Based on the pathway abundance table, the Wilcoxon test was used to identify metabolic pathways with significant differences between the bromhidrosis group and the healthy group (FDR<0.05).

[0053] III. Pathway Enrichment Analysis Further GRAS (or similar) pathway enrichment analysis was performed using the ReporterScore package to identify functional modules that underwent systematic changes between the two groups.

[0054] The results of the KEGG pathway enrichment analysis are as follows: Figure 6 As shown in Figure A, the group with bromhidrosis showed significant enrichment in pathways such as valine, leucine, and isoleucine degradation, aromatic compound degradation, and tyrosine metabolism. In contrast, the healthy control group showed greater activity in pathways such as amino acid biosynthesis and sphingolipid metabolism. The GRSA analysis results are as follows: Figure 6As shown in Figure B, the results further highlight the significance of these metabolic changes, and specifically point out that the "Staphylococcus aureus infection" pathway is significantly enriched in the bromhidrosis group, functionally linking specific pathogens to the disease phenotype. These results, from the perspective of microbial community function, corroborate the biological rationale for the combination of axillary bromhidrosis skin microbial markers screened in Example 1, and suggest that the production of odor substances may be related to the activation of specific bacterial metabolic pathways.

[0055] Example 4: A kit for diagnosing or assisting in the diagnosis of axillary bromhidrosis (body odor) Example 4 of this invention is an example of a kit for diagnosing or assisting in the diagnosis of axillary bromhidrosis provided by this invention. This kit is based on a combination of skin microbial markers for axillary bromhidrosis and corresponding detection methods. It can quickly and accurately detect the relative abundance of target bacteria in the sample to be tested, providing a reliable basis for the diagnosis or auxiliary diagnosis of axillary bromhidrosis. It is also simple to operate, highly applicable, and can be used for clinical testing or screening in primary healthcare institutions.

[0056] The kit includes reagents for detecting each individual bacterial species in the axillary bromhidrosis skin microbial marker combination obtained in Example 1, and the specific composition is as follows (prepared per person / set): Sample processing reagents include 10 mL of sterile PBS buffer (pH 7.4), 2 sterile swabs, 5 mL of sample lysis buffer (containing lysozyme and proteinase K, used to break microbial cell walls and release DNA), 1 DNA purification column and 2 mL of elution buffer, used for collecting, lysing and extracting and purifying total DNA from the axillary skin surface samples to be tested.

[0057] PCR amplification reagents include 25 μL of PCR premix (containing Taq enzyme, dNTPs, and buffer), specific primer pairs for each target bacterial species (corresponding to each single bacterial species in the axillary bromhidrosis skin microbial marker combination in Example 1, with a final concentration of 10 μmol / L for each primer pair, 2 μL each), and 18 μL of sterile deionized water, used for specific amplification of characteristic gene fragments of the target bacterial species.

[0058] Sequencing aids include 10 μL of sequencing adapters (compatible with the Illumina sequencing platform) and one set of DNA quality testing reagents (nucleic acid dyes, standards), used to ligate amplified DNA fragments and verify their quality, in preparation for subsequent sequencing.

[0059] Example 5: Device for diagnosing or assisting in the diagnosis of axillary bromhidrosis Embodiment 5 of the present invention is an embodiment of the device for diagnosing or assisting in the diagnosis of axillary bromhidrosis provided by the present invention. The embodiment of the device includes: a detection unit and an evaluation unit.

[0060] The detection unit is used to perform the following: detecting the relative abundance value of each single bacterial species in a sample of the axillary skin surface to be tested, wherein the single bacterial species includes the combination of axillary bromhidrosis skin microbial markers obtained in Example 1; An evaluation unit is used to perform the following: inputting the abundance detected by the detection unit into a model constructed in Example 2 for diagnosing or assisting in the diagnosis of axillary bromhidrosis, and outputting the probability of having bromhidrosis.

[0061] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of this invention.

Claims

1. A combination of skin microbial markers for axillary bromhidrosis, characterized in that, Including Prevotella ( Faecalibacterium prausnitzii Clostridium bacteria ( Clostridiales bacterium ), *Cryptospira marines* ( Phocaeicola dorei ), Parabacterium dilatatum ( Parabacteroides distasonis ), human fecal parasartella ( Parasutterella excrementihominis Corynebacterium BCW_4722 ( Corynebacterium sp. BCW_4722), Propionibacterium acnes ( Cutibacterium acnes Staphylococcus epidermidis ( Staphylococcus epidermidis ) and Sphingosomalidone spp. 3F27F9 ( Sphingomonas sp. 3F27F9).

2. The use of the combination of skin microbial markers for axillary bromhidrosis as described in claim 1 in the preparation of products or devices for diagnosing or assisting in the diagnosis of axillary bromhidrosis.

3. The application according to claim 2, characterized in that, Prevotella ( Faecalibacterium prausnitzii Clostridium bacteria ( Clostridiales bacterium ), *Cryptospira marines* ( Phocaeicola dorei ), Parabacterium dilatatum ( Parabacteroides distasonis ) and human fecal parasartella ( Parasutterella excrementihominis The relative abundance of Corynebacterium BCW_4722 was significantly reduced in patients with bromhidrosis. Corynebacterium sp. BCW_4722), Propionibacterium acnes ( Cutibacterium acnes Staphylococcus epidermidis ( Staphylococcus epidermidis ) and Sphingosomalidone spp. 3F27F9 ( Sphingomonas sp. The relative abundance of 3F27F9 was significantly increased in patients with bromhidrosis; the statistical criteria for significance were a false discovery rate of <0.05 after Benjamini-Hochberg correction and a linear discriminant analysis score of >4.0 in LEfSe analysis.

4. A reagent kit for diagnosing or assisting in the diagnosis of axillary bromhidrosis, characterized in that, The kit contains reagents for detecting the combination of axillary bromhidrosis skin microbial markers as described in claim 1.

5. A product for diagnosing or assisting in the diagnosis of axillary bromhidrosis, characterized in that, The product includes primers, probes, antibodies, aptamers, or chips that are specific to the combination of axillary bromhidrosis skin microbial markers as described in claim 1.

6. The method for screening a combination of skin microbial markers for axillary bromhidrosis as described in claim 1, characterized in that, Includes the following steps: 1) Collect axillary skin surface samples from patients with bromhidrosis and healthy subjects, extract total microbial DNA, and perform shotgun metagenomic sequencing; 2) Perform quality control and species analysis on metagenomic sequencing data to obtain a table of relative abundance of microorganisms; 3) Based on the obtained microbial relative abundance table, the microbial species that showed significant differences between the two groups of patients with bromhidrosis and healthy subjects were selected as the combination of microbial markers for axillary bromhidrosis skin.

7. A method for constructing a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis, characterized in that, Using the relative abundance of the combination of skin microbial markers for axillary bromhidrosis as described in claim 1 as a feature, a machine learning model is constructed, trained, and evaluated to obtain a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis.

8. A method for constructing a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis according to claim 7, characterized in that, The machine learning model is a random forest classification model, a support vector machine model, a gradient boosting decision tree model, a logistic regression model, or a neural network model.

9. A method for constructing a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis according to claim 7, characterized in that, The machine learning model is a random forest classification model. The relative abundance of the combination of skin microbial markers for axillary bromhidrosis as described in claim 1 is used as a feature. The random forest classification model is trained and its parameters are optimized using 10-fold cross-validation. The area under the receiver operating characteristic curve is calculated and its performance is evaluated to obtain a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis.

10. A device for diagnosing or assisting in the diagnosis of axillary bromhidrosis, characterized in that, include: The detection unit is configured to perform the following: detecting the relative abundance value of each single bacterial species in a sample of the axillary skin surface to be tested, wherein the single bacterial species includes the combination of axillary bromhidrosis skin microbial markers as described in claim 1; An evaluation unit is used to perform the following: inputting the abundance detected by the detection unit into a model obtained by the method for constructing a model for diagnosing or assisting in the diagnosis of axillary bromhidrosis as described in any one of claims 7 to 9, and outputting the probability of having bromhidrosis.

Images