B1185
Title: A network-constrain Weibull AFT model for biomarker discovery
Authors: Italia De Feis - National Council of Research (Italy) [presenting]
Abstract: A novel network-constraint survival methodology is proposed and explored, considering the Weibull accelerated failure time (AFT) model combined with a penalized likelihood approach for variable selection and estimation. The estimator explicitly incorporates the correlation patterns among predictors using a double penalty that promotes both sparsity and the grouping effect. In order to solve the structured sparse regression problems, an efficient iterative computational algorithm is presented based on the proximal gradient descent method. The theoretical consistency of the proposed estimator is established, and its performance, both on synthetic and real data examples, is evaluated.
