Refinement of the Bayesian probabilistic framework for forensic adult age-at-death estimation using dentin translucency in intact-single rooted teeth

This study refines a Bayesian probabilistic framework for adult age-at-death estimation based on root dentin translucency (RDT). Model comparison using the Bayesian Information Criterion demonstrated that a univariate model relying solely on RDT was the most parsimonious and efficient, achieving an R² of 63.60%. The framework was validated through entirely external samples: a Peruvian case study, and samples from Brazil (N = 247), the UK (N = 113), and Guatemala (N = 302). Bayesian updating using data from the Forensic International Dental Database (FIDB; N = 4522) substantially increased the strength of the evidence, with Likelihood Ratios rising from a mean of 2160 to nearly 10,000, elevating the inference to a level of very strong support and extremely strong support. Results confirmed the inter-population stability of the model and its capacity to mitigate the characteristic trajectory effect of overestimation in young adults and underestimation in older adults. This approach provides a transparent, robust, and globally applicable tool for forensic anthropology, successfully balancing statistical rigor with practical utility. Furthermore, it establishes a foundational platform, named BayAD, that enables the future integration of additional skeletal traits within an age-at-death estimation framework.