Status of Mandibular Third Molar Development as Evidence in Legal Age Threshold Cases

Applying artificial intelligence to determination of legal age of majority from radiographic data

Forensic odontologists use biological patterns to estimate chronological age for the judicial system. The age of majority is a legally significant period with a limited set of reliable oral landmarks. Currently, experts rely on the questionable development of third molars to assess whether litigants can be prosecuted as legal adults. Identification of new and novel patterns may illuminate features more dependably indicative of chronological age, which have, until now, remained unseen. Unfortunately, biased perceptions and limited cognitive capacity compromise the ability of researchers to notice new patterns. The present study demonstrates how artificial intelligence can break through identification barriers and generate new estimation modalities. A convolutional neural network was trained with 4003 panoramic-radiographs to sort subjects into 'under-18' and 'over-18' age categories. The resultant architecture identified legal adults with a high predictive accuracy equally balanced between precision, specificity and recall. Moving forward, AI-based methods could improve courtroom efficiency, stand as automated assessment methods and contribute to our understanding of biological ageing.

Paleodemography: From archaeology and skeletal age estimation to life in the past

Much of paleodemography, an interdisciplinary field with strong ties to archaeology, among other disciplines, is oriented toward clarifying the life experiences of past people and why they changed over time. We focus on how human skeletons contribute to our understanding of preindustrial demographic regimes, including when changes took place that led to the world as we know it today. Problems with existing paleodemographic practices are highlighted, as are promising directions for future work. The latter requires both better age estimates and innovative methods to handle data appropriately. Age-at-death estimates for adult skeletons are a particular problem, especially for adults over 50 years that undoubtedly are mistakenly underrepresented in published studies of archaeological skeletons. Better age estimates for the entirety of the lifespan are essential to generate realistic distributions of age at death. There are currently encouraging signs that after about a half-century of intensive, and sometimes contentious, research, paleodemography is poised to contribute much to understandings of evolutionary processes, the structure of past populations, and human-disease interaction, among other topics.

Conditional independence assumption and appropriate number of stages in dental developmental age estimation

When estimating the age of an individual it is critical that 1) age ranges are as narrow as possible while still capturing the true age of the individual with an acceptable frequency, and 2) this frequency is known. When multiple traits are used to produce a single age estimate, the simplest practice is to assume that the traits are conditionally independent from one another given age. Unfortunately, if the traits are correlated once the effect of age is accounted for, the resulting age intervals will be too narrow. The frequency at which the age interval captures the true age of the individual will be decreased below the expected value to some unknown degree. It is therefore critical that age estimation methods that include multiple traits incorporate the possible correlations between them. Moorrees et al. (1963) [1] scores of the permanent mandibular dentition from 2607 individuals between 2 and 23 years were used to produce and cross-validate a cumulative probit model for age estimation with an optimal number of stages for each tooth. Two correction methods for covariance of development between teeth were tested: the variance-covariance matrix for a multivariate normal, and the Boldsen et al. (2002) [2] ad-hoc method. Both correction methods successfully decreased age interval error rates from 21% to 23% in the uncorrected model to the expected value of 5%. These results demonstrate both the efficacy of these correction methods and the need to move away from assuming conditional independence in multi-trait age estimation.

Performance of three mathematical models for estimating age-at-death from multiple indicators of the adult skeleton

The mathematical method which will achieve the most accurate and precise age-at-death estimate from the adult skeleton is often debated. Some research promotes Bayesian analysis, which is widely considered better suited to the data construct of adult age-at-death distributions. Other research indicates that methods with less mathematical complexity produce equally accurate and precise age-at-death estimates. One of the advantages of Bayesian analysis is the ability to systematically combine multiple indicators, which is reported to improve the age-at-death estimate. Few comparisons exist between Bayesian analysis and less complex mathematical models when considering multiple skeletal indicators. This study aims to evaluate the performance of a Bayesian approach compared to a phase-based averaging method and linear regression analysis using multiple skeletal indicators. The three combination methods were constructed from age-at-death data collected from 330 adult skeletons contained in the Raymond A Dart and Pretoria Bone Collections in South Africa. These methods were tested and compared using a hold-out sample of 30 skeletons. As is frequently reported in literature, a balance between accuracy and precision was difficult to obtain from the three selected methods. However, the averaging and regression analysis methods outperformed the Bayesian approach in both accuracy and precision. Nevertheless, each method may be suited to its own unique situation-averaging to inform first impressions, multiple linear regression to achieve statistically defensible accuracies and precisions and Bayesian analysis to allow for cases where category adjustments or missing indicators are necessary.

Minor or adult? Introducing decision analysis in forensic age estimation

Nowadays, forensic age estimation takes an important role in worldwide forensic and medico-legal institutes that are solicited by judicial or administrative authorities for providing an expert report on the age of individuals. The authorities' ultimate issue of interest is often the probability that the person is younger or older than a given age threshold, which is usually the age of majority. Such information is fundamental for deciding whether a person being judged falls under the legal category of an adult. This is a decision that may have important consequences for the individual, depending on the legal framework in which the decision is made. The aim of this paper is to introduce a normative approach for assisting the authority in the decision-making process given knowledge from available findings reported by means of probabilities. The normative approach proposed here has been acknowledged in the forensic framework, and represents a promising structure for reasoning that can support the decision-making process in forensic age estimation. The paper introduces the fundamental elements of decision theory applied to the specific case of age estimation, and provides some examples to illustrate its practical application.

Age estimation in the living: cervical ring apophysis development in a Turkish sample using CT

With increases in migration across borders, age estimation in living individuals of not (reliably) documented identity becomes all the more important. Unfortunately, there are not many age indicators that can be used for this purpose, and human variation requires specific methodical approaches. In this paper, a recently proposed age marker to assess the age around the critical age limit of 18 completed years is tested. The method uses apophyseal development of cervical vertebrae 2, 3 and 4. Here CT scans of a large sample of Turkish individuals (n = 1276) were assessed, and likelihoods of being 18 years at a given stage were calculated. The likelihood of being at least 18 years for stages 0, 1 and 2 were zero or close to zero in both males and females. By the time that stage 4 was reached, the likelihood to be 18 years were between 65 and 70% (depending on the vertebra) in females and 81 and 90% in males. In comparison to South Africans, the Turkish individuals developed earlier, but the likelihoods of being 18 years were lower at stage 4 as some individuals were still judged to be in stage 3 well into their twenties. Although fairly variable, this method is a valuable new addition to the modalities that can be used for age assessment in the living. CT scans seemed to provide good visualization of the structures in question, although in actual forensic cases the high radiation dose may be problematic.