A large-scale Bayesian validation of FIDBV2 and FIDBV3 methods for adult age-at-death estimation using root dentin translucency in single-rooted teeth.

Statistical calibration of the results in adult age-at-death estimation is a fundamental yet challenging task in forensic anthropology and bioarchaeology. This study conducts a large-scale Bayesian validation of two probabilistic models for adult dental age-at-death estimation: the established FIDBv2, using root dentin translucency and periodontal retraction, and a novel simplified counterpart, FIDBv3, relying solely on translucency. Analysis of 8,595 single-rooted teeth from a global sample of 3,801 individuals demonstrates that FIDBv3 consistently outperforms its predecessor, reducing overall mean inaccuracy by 22.52% and increasing total sample accuracy by 8.11 percentage points. This superior performance across diverse populations indicates that periodontal retraction introduces more statistical noise than a reliable biological signal. The strategic exclusion of this variable, combined with the inclusion of a logarithmic transformation that better models the non-linear trajectory of translucency development, yields a more biologically realistic and universally applicable method. This simplification also streamlines analysis, while deployment as a free, open-access web platform democratizes global access to this robust, probability-based estimation. Despite persistent challenges at the extremes of the lifespan, FIDBv3 establishes a new standard for transparent, reliable and practical dental age-at-death estimation in forensic and archaeological contexts.