{"title":"通过机器学习整合呼吸道微生物组和宿主免疫反应,用于呼吸道感染诊断","authors":"Hongbin Chen, Tianqi Qi, Siyu Guo, Xiaoyang Zhang, Minghua Zhan, Si Liu, Yuyao Yin, Yifan Guo, Yawei Zhang, Chunjiang Zhao, Xiaojuan Wang, Hui Wang","doi":"10.1038/s41522-024-00548-y","DOIUrl":null,"url":null,"abstract":"<p>At present, the diagnosis of lower respiratory tract infections (LRTIs) is difficult, and there is an urgent need for better diagnostic methods. This study enrolled 136 patients from 2020 to 2021 and collected bronchoalveolar lavage fluid (BALF) specimens. We used metatranscriptome to analyze the lower respiratory tract microbiome (LRTM) and host immune response. The diversity of the LRTM in LRTIs significantly decreased, manifested by a decrease in the abundance of normal microbiota and an increase in the abundance of opportunistic pathogens. The upregulated differentially expressed genes (DEGs) in the LRTIs group were mainly enriched in infection immune response-related pathways. Klebsiella pneumoniae had the most significant increase in abundance in LRTIs, which was strongly correlated with host infection or inflammation genes TNFRSF1B, CSF3R, and IL6R. We combined LRTM and host transcriptome data to construct a machine-learning model with 12 screened features to discriminate LRTIs and non-LRTIs. The results showed that the model trained by Random Forest in the validate set had the best performance (ROC AUC: 0.937, 95% CI: 0.832–1). The independent external dataset showed an accuracy of 76.5% for this model. This study suggests that the model integrating LRTM and host transcriptome data can be an effective tool for LRTIs diagnosis.</p>","PeriodicalId":19370,"journal":{"name":"npj Biofilms and Microbiomes","volume":"28 1","pages":""},"PeriodicalIF":7.8000,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Integrating respiratory microbiome and host immune response through machine learning for respiratory tract infection diagnosis\",\"authors\":\"Hongbin Chen, Tianqi Qi, Siyu Guo, Xiaoyang Zhang, Minghua Zhan, Si Liu, Yuyao Yin, Yifan Guo, Yawei Zhang, Chunjiang Zhao, Xiaojuan Wang, Hui Wang\",\"doi\":\"10.1038/s41522-024-00548-y\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>At present, the diagnosis of lower respiratory tract infections (LRTIs) is difficult, and there is an urgent need for better diagnostic methods. This study enrolled 136 patients from 2020 to 2021 and collected bronchoalveolar lavage fluid (BALF) specimens. We used metatranscriptome to analyze the lower respiratory tract microbiome (LRTM) and host immune response. The diversity of the LRTM in LRTIs significantly decreased, manifested by a decrease in the abundance of normal microbiota and an increase in the abundance of opportunistic pathogens. The upregulated differentially expressed genes (DEGs) in the LRTIs group were mainly enriched in infection immune response-related pathways. Klebsiella pneumoniae had the most significant increase in abundance in LRTIs, which was strongly correlated with host infection or inflammation genes TNFRSF1B, CSF3R, and IL6R. We combined LRTM and host transcriptome data to construct a machine-learning model with 12 screened features to discriminate LRTIs and non-LRTIs. The results showed that the model trained by Random Forest in the validate set had the best performance (ROC AUC: 0.937, 95% CI: 0.832–1). The independent external dataset showed an accuracy of 76.5% for this model. This study suggests that the model integrating LRTM and host transcriptome data can be an effective tool for LRTIs diagnosis.</p>\",\"PeriodicalId\":19370,\"journal\":{\"name\":\"npj Biofilms and Microbiomes\",\"volume\":\"28 1\",\"pages\":\"\"},\"PeriodicalIF\":7.8000,\"publicationDate\":\"2024-09-12\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"npj Biofilms and Microbiomes\",\"FirstCategoryId\":\"99\",\"ListUrlMain\":\"https://doi.org/10.1038/s41522-024-00548-y\",\"RegionNum\":1,\"RegionCategory\":\"生物学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"BIOTECHNOLOGY & APPLIED MICROBIOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"npj Biofilms and Microbiomes","FirstCategoryId":"99","ListUrlMain":"https://doi.org/10.1038/s41522-024-00548-y","RegionNum":1,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"BIOTECHNOLOGY & APPLIED MICROBIOLOGY","Score":null,"Total":0}
Integrating respiratory microbiome and host immune response through machine learning for respiratory tract infection diagnosis
At present, the diagnosis of lower respiratory tract infections (LRTIs) is difficult, and there is an urgent need for better diagnostic methods. This study enrolled 136 patients from 2020 to 2021 and collected bronchoalveolar lavage fluid (BALF) specimens. We used metatranscriptome to analyze the lower respiratory tract microbiome (LRTM) and host immune response. The diversity of the LRTM in LRTIs significantly decreased, manifested by a decrease in the abundance of normal microbiota and an increase in the abundance of opportunistic pathogens. The upregulated differentially expressed genes (DEGs) in the LRTIs group were mainly enriched in infection immune response-related pathways. Klebsiella pneumoniae had the most significant increase in abundance in LRTIs, which was strongly correlated with host infection or inflammation genes TNFRSF1B, CSF3R, and IL6R. We combined LRTM and host transcriptome data to construct a machine-learning model with 12 screened features to discriminate LRTIs and non-LRTIs. The results showed that the model trained by Random Forest in the validate set had the best performance (ROC AUC: 0.937, 95% CI: 0.832–1). The independent external dataset showed an accuracy of 76.5% for this model. This study suggests that the model integrating LRTM and host transcriptome data can be an effective tool for LRTIs diagnosis.
期刊介绍:
npj Biofilms and Microbiomes is a comprehensive platform that promotes research on biofilms and microbiomes across various scientific disciplines. The journal facilitates cross-disciplinary discussions to enhance our understanding of the biology, ecology, and communal functions of biofilms, populations, and communities. It also focuses on applications in the medical, environmental, and engineering domains. The scope of the journal encompasses all aspects of the field, ranging from cell-cell communication and single cell interactions to the microbiomes of humans, animals, plants, and natural and built environments. The journal also welcomes research on the virome, phageome, mycome, and fungome. It publishes both applied science and theoretical work. As an open access and interdisciplinary journal, its primary goal is to publish significant scientific advancements in microbial biofilms and microbiomes. The journal enables discussions that span multiple disciplines and contributes to our understanding of the social behavior of microbial biofilm populations and communities, and their impact on life, human health, and the environment.