ENHANCING AND EVALUATING PREDICTION MACHINES
Mercoledì 21 Giugno 2017 alle ore 17.30 in Aula Magna al Collegio Morgagni si terrà il secondo incontro con il Prof. Daniel Gianola all’interno delle attività didattiche Scuola Galileiana di studi Superiori. Il titolo dell’incontro è: “ENHANCING AND EVALUATING PREDICTION MACHINES USING RESAMPLING: APPLICATIONS TO ANIMAL AND PLANT BREEDING” in cui dalla genetica applicata alle piante e agli animali si arriverà a parlare di medicina personalizzata e di nuove metodologie di applocazione e intervento.
L’incontro è aperto al pubblico fino ad esaurimento posti a sedere.
The process of inferring the genetic worth of animals and plants in the context of artificial selection (genetic improvement) programs can be casted as one of prediction of yet-to-be observed outcomes. For example, a battery of molecular markers together with phenotypes of interest (i.e., measurements on some target trait) is used to build a prediction equation, and the latter is employed to predict the future performance of a set of candidates for selection (e.g., artificial insemination bulls or crossbred performance in maize), conditionally on the realized values of the markers. Typically all one has at hand is a single data set; however, some measure of uncertainty is needed to explore the amount of variability to be expected if the process were to be repeated over and over. The presentation will illustrate how resampling procedures can be used to estimate the distribution of prediction errors, to compare alternative prediction models ( prediction machines), to improve the accuracy of predictions and to remove their bias, if any. The procedures discussed can also be applied to personalized human medicine, where diagnosis of diseases and on their progression and outcomes of interventions can be now predicted using a potentially enormous numbers of biomarkers.