Scholarlink

<p>The human gut microbiome plays a critical role in maintaining host health and homeostasis, and current literature suggests a bidirectional relationship between microbiome ecology and host well-being. DNA metabarcoding has emerged as a powerful tool for investigating microbiome imbalances (i.e., dysbiosis). While the prokaryotic microbiome has been extensively studied, the fungal counterpart – or mycobiome – remains largely unexplored, despite its recognized role from the perinatal stage onward. Here, we present a comprehensive survey based on DNA metabarcoding analysis of approximately 1,500 publicly available ITS1 samples. This survey integrates conventional statistical approaches with Machine Learning (ML) methods coupled with explainable Artificial Intelligence (XAI). ML models successfully predicted host health status with accuracies exceeding 80%, and fungal genera such as <i>Eurotium</i>, <i>Aureobasidium</i>, <i>Candida</i>, and <i>Cutaneotrichosporon</i> emerged as key classification features. This study introduces a cutting-edge multiview analytical framework applied to publicly available mycobiome data, highlighting the potential of fungal community profiling as a non-invasive tool to support health diagnostics.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-based assessment of the healthy human gut mycobiota landscape using ITS1 DNA metabarcoding data

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月2日

查看全文
  • Giuseppe Defazio,
  • Erika Lorusso,
  • Mariangela De Robertis,
  • Tommaso Mello,
  • Andrea Galli,
  • Graziano Pesole,
  • Bruno Fosso

摘要

The human gut microbiome plays a critical role in maintaining host health and homeostasis, and current literature suggests a bidirectional relationship between microbiome ecology and host well-being. DNA metabarcoding has emerged as a powerful tool for investigating microbiome imbalances (i.e., dysbiosis). While the prokaryotic microbiome has been extensively studied, the fungal counterpart – or mycobiome – remains largely unexplored, despite its recognized role from the perinatal stage onward. Here, we present a comprehensive survey based on DNA metabarcoding analysis of approximately 1,500 publicly available ITS1 samples. This survey integrates conventional statistical approaches with Machine Learning (ML) methods coupled with explainable Artificial Intelligence (XAI). ML models successfully predicted host health status with accuracies exceeding 80%, and fungal genera such as Eurotium, Aureobasidium, Candida, and Cutaneotrichosporon emerged as key classification features. This study introduces a cutting-edge multiview analytical framework applied to publicly available mycobiome data, highlighting the potential of fungal community profiling as a non-invasive tool to support health diagnostics.