Proposal Description

In recent years, advancements in high-throughput sequencing technologies have democratized the acquisition and comprehensive characterization of microbiota data. Consequently, progress in clinical research has shed light on the significant role of the human microbiome in health and disease, effectively bridging the gap from the bench to the bedside. The composition and diversity of human microbiota have emerged as promising biomarkers for human health and predictors of disease. Indeed, utilizing microbiota as a biomarker offers non-invasive, cost-effective approaches for disease risk assessment, personalized treatment strategies, and monitoring therapeutic efficacy.

For example, research into the connections between the gut microbiome and the host immune response is paving the way for microbiota-based cancer prognosis. Additionally, investigations into the gut-brain microbiome axis are striving to identify new biomarkers and therapeutic targets for neurological and psychiatric conditions. Moving beyond the gut microbiome, promising examples include the vaginal microbiome-based prediction of preterm birth. Also, significant research initiatives have been undertaken to identify respiratory microbiota biomarkers associated with exacerbations of respiratory diseases such as asthma, as well as to evaluate their potential in predicting therapeutic outcomes of modulators, particularly in conditions like cystic fibrosis.

To achieve accurate predictions, researchers are increasingly turning to machine learning and deep learning techniques. More traditional approaches involve measuring a few biomarkers repeatedly over time to dynamically update estimated outcomes. Yet, the complexity of microbiome data presents significant challenges for analysis. Factors such as compositionality/hierarchical structure, zero inflation/high skewness, high-dimensionality, cross-kingdom interactions, and longitudinal nature all contribute to these challenges. Therefore, there is an urgent need for the development of rigorous statistical tools capable of addressing these complexities and effectively answering clinical questions.

The proposed session brings together researchers from diverse backgrounds and countries who are actively engaged in improving statistical microbiome data analysis techniques, collaborating closely with clinicians. The session will be facilitated by both statistical and clinical experts who are involved in projects where microbiome-based biomarkers have gained significant attention as potential prognostic tools.