For many objects in nature there are certain relations between their structure and various properties, e.g. physical, chemical, biological and many others. In most of the cases, revealing of such relations is quite complex, but new analytical methods, based on soft modeling approach, allow to make descriptions of dependencies between structure of an object and its properties. In the present work the analysis of skeleton bones structure has been used for estimation of age of victims' remains in forensic medicine.

The skeleton bones are stable to putrefying transformations and have different age-related information features, first of all connected with osteoporosis process. However the quantitative estimation of related structure properties is not trivial and required a lot of experiments. The other issue here is that the relations between structural characteristics and age of individuals are not linear, as it has been shown by experiments with light absorbance.

The preliminary investigations showed that the combination of different methods for structure features extraction (like Angle Measure Technique and statistics of wavelet coefficients) and Partial Least Squares for modeling the relations between these features and age gives quite promising result. In this work we would like to present the developing of the method. The new results obtained using a novel method for quantitative structure analysis, based on fractal dimension, and Support Vector Machines regression. Three sample sets were used to evaluate the model - calibration, validation and a control set for prediction. The comparison with previous results as well as with one of approaches used in forensic medicine will be shown.