Age estimation of unaccompanied migrant adolescents for judicial purposes: a bayesian approach

Judicial authorities are increasingly concerned by determining the age of unaccompanied adolescents or young adults under suspicion. When other approaches are considered insufficient, courts demand appropriate medical tests. Part of those tests aim at determining the civil age through skeletal age. However, this determination is subject to a large uncertainty and it is difficult to assess its amount of accuracy. To a large extent the problem is statistical and the previous statistical proposals suffer from serious drawbacks. For instance, they rest, at least partially, on assumptions or conventions rather than actual data or use the name of Bayes without implementing a genuine Bayesian procedure. Instead, we present an actual Bayesian procedure resting not only on reference data (as all previous proposals) but also on data that permit to estimate the age distribution of the subjects submitted to examination in a given context. The method provides estimates of the civil age with their precision under the form of a probability density. It is ethic since it estimates the global behaviour of the examined subjects without prejudging anything about it (e.g. with respect to the alleged age). The method is illustrated by a real-life case (data gathered in the Department of Forensic Medicine, Jean-Verdier Hospital, Bondy, France, from 2017 to 2020). We used data from hand/wrist radiographs but the presented approach could be applied to data from any other anatomical site.