Incorporating the Weathering Hypothesis and Virtual Imaging to Multivariable Macroscopic Adult Age Estimation for Forensic Anthropology

Adult age estimation in forensic anthropology is critical to the identification of unknown individuals across medico-legal, political violence, and humanitarian contexts. Despite decades of methodological development, current approaches remain accurate but imprecise—yielding wide age intervals that limit practical utility—particularly for middle-aged adults between 40 and 70 years. This limitation stems from four compounding problems: reliance on single-indicator methods; insufficient integration of multi-indicator approaches; an exclusive focus on skeletonized remains that excludes virtual imaging modalities; and a failure to account for the holistic life course and phenotypic heterogeneity of individuals. Notably, the age range most resistant to precise estimation overlaps almost exactly with the period Geronimus identifies as the window of greatest morbidity and mortality disparities between marginalized and privileged populations—a convergence that forms the theoretical core of this dissertation. Growing evidence suggests that stress-related variables, framed by the weathering hypothesis, contribute to accelerated skeletal aging among structurally vulnerable populations, yet these factors have not been systematically incorporated into age estimation models.This dissertation addresses these gaps through three interconnected specific aims, applied to a diverse international sample of 597 individuals (278 females, 319 males) from Portugal (CEI-XXI, n=92), Chile (COSS, n=195), and the United States (NMDID, n=297; Maxwell Museum, n=11). Data were collected on age indicators from two multivariable methods—Transition Analysis 3 (TA3) and Deep Random Neural Networks for adult age estimation (DRNNAGE)—alongside a battery of aging-influencing variables rooted in the weathering hypothesis, including sociodemographic, early- and later-life stress, and body-size indicators, evaluated across both dry bone and postmortem computed tomography (CT) modalities. Specific Aim 1 evaluated the transferability of dry bone age indicator scores to CT-derived virtual assessments in a cadaveric subsample scanned prior to skeletonization, demonstrating adequate agreement for most variables while identifying systematic problems in the assessment of texture-dependent and porosity-related features. Specific Aim 2 employed Spearman correlation, mutual information, elastic-net regression, and a DRNNAGE surrogate model to quantify variable contributions, confirming that traditional and non-traditional age indicators contribute meaningfully across the lifespan, and that key aging-influencing variables—including adult socioeconomic status, antemortem tooth loss, and caries counts—are informative in multivariable contexts. Specific Aim 3 constructed predictive deep random neural network models comparing different variable combinations; models complemented with stress-selection sets achieved improved precision while maintaining accuracy, with gains concentrated precisely in the most challenging middle adult age range. Living weigh—and its dry bone proxies such as femoral head and midshaft measurements, antemortem tooth loss, osteoarthritis, and sex emerged as the most effective moderators of age estimation bias. Collectively, these results support the hypothesis that incorporating weathering hypothesis–grounded and life-course indicative variables improves the precision of adult age estimates without sacrificing accuracy. The dissertation treats the skeleton as a true holistic phenotype and contributes a theoretically informed, empirically validated framework for more effective age estimation in the diverse forensic contexts where identification is needed most.

Level thesis:degree_level

Doctorate Degree

Year dc:date.issued

2026

Author dc:creator

• Galimany Skupham, Jacqueline Lorna

Advisor dc:contributor.advisor

• Stull, Kyra E

Committee members dc:contributor.committeemember

• Pilloud, Marin A
• Hand, Emily M
• Boehm, Debbie A
• Navega, David S

• Age Estimation
• Skeleton
• Weathering Hypothesis

Language dc:language.iso

en_US, English

Repository record dc:identifier.uri

https://scholarwolf.unr.edu/handle/11714/11826

OAI identifier oai:identifier

oai:scholarwolf.unr.edu:11714/11826