Int J Paleopathol. 2025 Dec 1;52:32-43. doi: 10.1016/j.ijpp.2025.11.004. Online ahead of print.
ABSTRACT
OBJECTIVE: This paper demonstrates computational modeling’s value as a tool for mapping the impact of hidden variables and evaluating the accuracy of statistical methods in bioarchaeology.
MATERIALS: As a working example, this paper presents an agent-based model of a 1,000-person cohort of individuals who can form an unspecified skeletal lesion at any age between birth and ten years and enter a simulated cemetery at the end of their lives. Skeletal lesions either have no effect on mortality risk (scenario 1) or are associated with doubled mortality risk (scenario 2).
METHODS: The agent-based model simulates data on individual age at death and lesion status. Kaplan-Meier survival analysis is run on each simulated dataset, comparing survival estimates for individuals with and without lesions.
RESULTS: Survival analyses underestimate the true value of lesion-associated mortality risk in early life in scenario 2 and produce a false lesion-associated survival advantage under the null conditions of scenario 1.
CONCLUSIONS: Researchers should account for the ages of a skeletal lesion’s developmental window, where known, when assessing lesion-associated mortality. Survival analyses return accurate results when they exclude individuals in the ages of active lesion formation.
SIGNIFICANCE: Modeling experiments can identify which archaeologically unmeasurable variables have the greatest impact on estimates of population health and outline the ways in which they bias estimates of past health from the skeletal record.
LIMITATIONS: The only limits on modeling are limits of imagination and common sense.
SUGGESTIONS FOR FUTURE RESEARCH: Many other archaeologically hidden variables remain to be explored with this approach.
PMID:41330016 | DOI:10.1016/j.ijpp.2025.11.004
