Abstract: Screening important features based on ultrahigh-dimensional data has become an important task in statistical analysis. As such, several screening procedures have been roposed for various types of studies or data, including complete data and right-censored failure time data. In this study, we consider ultrahigh-dimensional interval-censored failure time data. Such data occur frequently in medical follow-up studies, among others, and include right-censored data as a special case, but for which few works exist. For the problem, a distance correlation-based sure independent screening procedure is proposed. The new approach is model-free and does not require estimating survival functions, unlike most existing nonparametric screening procedures for failure time data. We establish the sure screening property and the ranking consistency of the proposed method, and conduct an extensive simulation study, which suggests that the proposed procedure works well for practical situations. Finally, we apply the proposed method to a set of real data on Alzheimer's disease, which motivated this study.

Key words and phrases: Distance correlation, interval-censored data, model-free screening, sure screening property, ultrahigh-dimensional data.