Does anyone have experience or thoughts on the suitability of any of these general ML packages for implementing methods for risk prediction with a censored time to event response? I would like a framework for high-dimensional model tuning, nested cross-validation, prediction on new data, and model evaluation using for example C-statistic, for arbitrary algorithms in the survival context. -- Levi Waldron Post-doctoral fellow Department of Biostatistics Harvard School of Public Health Building 1, room 412C 655 Huntington Avenue Boston, Massachusetts 02115 phone: (617) 432-4926 fax: (617) 432-5619