Early-life microbiome trajectories as biomarkers to predict health outcomes.

The early-life gut microbiome is tightly linked to different aspects of infant development. Microbial colonisation patterns have been repeatedly shown to play a role in a variety of paediatric outcomes, ranging from metabolism and immune function to neurodevelopment. Concomitantly, the identification of early-life biomarkers is crucial, especially considering that for various conditions, reliable diagnostic tools only emerge in early childhood. As such, microbiome data collected in the first two years of life may offer valuable prospects for early detection, prevention, quantification or even correction of adverse health trajectories. With the increasing availability of high-resolution microbiome data, researchers are leveraging both traditional statistical approaches and machine learning (ML) methods to analyse the evolution of these complex microbial communities. While statistical models are well-suited for identifying associations between microbiome features and health states, ML methods allow for predicting health outcomes from those features. This review explores the role of the early-life gut microbiome in infant health and development, with a focus on how data acquisition and analytical methods can shape current knowledge. We contrast statistical approaches with ML methods, summarising key findings on microbial succession and factors influencing it. By addressing current challenges and identifying areas for methodological refinement, we aim to discuss the potential of the microbiome in the assessment of current and future health states of an individual and aid in the development of more robust, clinically-relevant models for paediatric care.