Аннотация:The estimation of the age of infant skeletons istypically based on dental development and postcranialmetrics. But numerous studies show thatboth methods display a high level of uncertaintydue to inter-individual variability. The assessmentof dental age is also largely dependent on theskills and experience of the observer.The rst ve years of life is the time of the mostrapid growth in all craniofacial dimensions. Thegrowth trajectories of the dimensions are notlinear and vary substantially. Thus, the combinationof the levels of maturation of differentdimensions can potentially provide quite a preciseestimation of the age-at-death.High-resolution clinical CT scans of 501 individualsof both sexes were studied, including 171infants of the 1st year of life, 261 children from 2ndto 6th years, and 69 adults (reference). The datasetwas divided into 4 age cohorts for the 1st year oflife, and 5 yearly cohorts for the older subadults. Aset of linear measurements describing the mainmorphological features of the facial skeletonwas calculated based on 3d landmark data. Anensemble of random forest and SVM supportvector machine algorithms of machine learningwas employed. The results were evaluated via10-fold cross-validation.In both sex-specic and combined samples, theproportion of fully correct estimations (i.e. exactlymatching the actual cohort) was 70-80%, partiallycorrect (to neighboring cohorts) - higher than 90%.Fairly high precision of the estimation could bereached even if use only a few of the dimensions.