Stochastic Models in the Identification Process

0368507 - ÚI 2012 RIV CZ eng J - Journal Article
Slovák, Dalibor - Zvárová, Jana
Stochastic Models in the Identification Process.
European Journal for Biomedical Informatics. Roč. 7, č. 1 (2011), s. 44-50. ISSN 1801-5603
R&D Projects: GA MŠMT(CZ) 1M06014
Institutional research plan: CEZ:AV0Z10300504
Keywords : identification process * weight-of evidence formula * coancestry coefficient * beta-binomial sampling formula * DNA mixtures
Subject RIV: IN - Informatics, Computer Science

Objectives: The DNA analysis is now accepted by the broad public as a completely standard and faultless procedure but in some circumstances its reliability can decrease substantially. This paper deals with the process of identifying and determining the weight of evidence against the suspect. Main stochastic approaches to identification are shown. Methods: The weight-of-evidence formula was derived from Bayes theorem and its application in the model of the island problem was demonstrated. The beta-binomial formula derived from Dirichlet distribution was used for calculation of more complex situations. Results: From many various complications in the model of the island problem there was shown how to work with uncertainty in a population size. The beta-binomial formula was used to include a subpopulation structure and in issues of DNA mixtures. Conclusions: In particular, the influence of a population structure is now explored insufficiently. Using the results of H. Kubátová in this area, a new formula was derived.
Permanent Link: http://hdl.handle.net/11104/0202834