Evaluation of Kinship Identification Systems Based on STR DNA Profiles

In this paper we show how to evaluate a kinship identification system, a probabilistic tool devoted to obtain the likelihood ratio (LR) for the hypothesis that an individual is a specific member of a family, conditional on the available familial DNA evidence. The paper considers the LR as a random variable and focuses on the expected possibilities of identifying a candidate. In a pre-experimental perspective, we evaluate whether a system fulfils the requirements of the parties involved. The proposed system evaluation is specific for each case, does not require any additional laboratory costs and is carried out before the identification trial is performed. Special attention is devoted to the computational aspects of the problem: since high dimensional spaces are involved, matters concerning approximations are discussed. A number of real cases taking advantage of the proposed methodology are presented and discussed in the paper.