Known unknowns and the osteological paradox: Why bioarchaeology needs agent-based models

Area of Science:

• Bioarchaeology
• Computational Modeling
• Paleopathology

Background:

• Assessing past population health relies on skeletal records.
• Statistical methods are commonly used but can be influenced by unmeasured variables.
• Skeletal lesions can provide insights into disease and mortality but require careful interpretation.

Purpose of the Study:

• To demonstrate computational modeling as a tool for bioarchaeology.
• To evaluate the accuracy of statistical methods in the presence of hidden variables.
• To map the impact of unmeasurable variables on health estimates from skeletal remains.

Main Methods:

• An agent-based model simulated a 1,000-person cohort with skeletal lesions.
• Two scenarios were modeled: no mortality risk and doubled mortality risk associated with lesions.
• Kaplan-Meier survival analysis was applied to simulated datasets.

Main Results:

• Survival analyses underestimated mortality risk when lesions doubled risk (Scenario 2).
• A false survival advantage was observed under null conditions (Scenario 1).
• The age of lesion development significantly biased survival estimates.

Conclusions:

• Researchers must consider the developmental window of skeletal lesions when assessing mortality.
• Excluding individuals in ages of active lesion formation improves survival analysis accuracy.
• Computational modeling can identify and quantify biases in skeletal health assessments.