Analysis of complex family cases with Probabilistic Expert Systems

Forensic identification with genetic markers is currently experiencing various scenarios of inference, spanning from direct to indirect comparisons in matter of family relationship and criminal analysis. The discrimination power of current genetic polymorphisms is such that the inferential process underlying this field of human identification can efficiently be deployed even in cases where there is a lack of direct knowledge on the genetic asset on one of the terms in comparison. The most promising of these software facilities are the probabilistic expert systems (PES). To verify the power of these systems to confirm relationships among individuals in defective cases, we analyzed four different groups of complex families with known relationships. To test the goodness of the software employed (FINEX), we created different pedigrees by deleting one or more individuals from each family and calculating the different probability values obtained.