Single-tooth resolved, whole-mouth prediction of early childhood caries via spatiotemporal variations of plaque microbiota.

Early childhood caries (ECC) exhibits tooth specificity, highlighting the need for single-tooth-level prevention. We profiled 2,504 dental plaque microbiota samples from 89 preschoolers across two cohorts, tracking compositional changes with imputed functional trends at a single-tooth resolution over 11 months. In healthy children, dental microbiota exhibited an anterior-to-posterior ecological gradient on maxillary teeth and strong bilateral symmetry. These patterns were disrupted in caries-affected children due to caries-driven microbial reorganization. Leveraging tooth-specific disease-associated taxa and spatially related clinical/microbial features, we developed spatial microbial indicators of caries (spatial-MiC or sMiC) using machine-learning techniques. sMiC achieves 98% accuracy in diagnosing ECC at a single-tooth resolution and 93% accuracy in predicting new caries 2 months in advance in perceived-healthy teeth. This high-resolution spatiotemporal microbial atlas of ECC development disentangles the microbial etiology at the single-tooth level, identifies a characteristic microbial signature for each tooth, and provides a foundation for tooth-specific ECC prevention strategies.