ANALYSIS OF A DNA MIXTURE SAMPLE USING OBJECT-ORIENTED BAYESIAN NETWORKS

Nowadays, the use of DNA profiles in forensic identification problems is a very common procedure, in many and different situations. The observance of a mixture trace resulting of a crime that has been committed is a very interesting and challenging phenomenon for the laboratories and for the judicial systems. Using all adequate probabilistic concept to approach this king of problems we want to present and discuss the treatment and responses one can give in a case of a mixture with a single trace observed for several different markers. The construction and use of Bayesian networks to analyze complex problems of forensic identification inference had its beginning in 2002 and since then it has been seen as a very efficient tool to handle these problems. After presenting the case, we formulated the hypothesis of interest and have performed the analysis appealing to the use of software to construct the object-oriented Bayesian networks and analyze the data. In the case analyzed it was considered the possibility of unknown contributors in the mixture. We started by building a network for a single marker and extended it to a network being able to analyze the information available for all the markers considered.