Applications of machine learning in microbial source tracking

Microbial source tracking (MST) has become important in environmental ecology, food safety, and forensic investigation. With advances in microbiome technologies, MST has shifted from single-indicator methods to community-level inference, creating demand for stronger analytical frameworks. Machine learning (ML) now plays a central role in handling large-scale microbiome data and capturing complex relationships between microbial communities and their sources. This review summarizes major ML methods used in MST, representative tools, and applications in pollution tracing, geospatial attribution, food safety, and forensics. Current studies show that ML substantially improves the accuracy, resolution, and scalability of MST. We also discuss key challenges, including limited interpretability, ecological dynamics, and a lack of benchmark datasets with explicit ground truth and evaluation criteria, and highlight how next-generation AI, especially deep learning, may further advance robust and intelligent MST.