Support Vector Machines for Current Status Data

In this paper we aim to develop a general, model free, method for analyzing current status data using machine learning techniques. In particular, we propose a support vector machine (SVM) learning method for estimation of the failure time expectation for current status data. SVM was originally introduced by Vapnik in the 1990's and is firmly related to statistical learning theory (Vapnik, 1999). The choice of SVMs for current status data is motivated by the fact that SVMs can be implemented easily, have fast training speed, produce decision functions that have a strong generalization ability and can guarantee convergence to the optimal solution, under some weak assumptions (Shivaswamy et al., 2007). Current status data is a data format where the failure timeT is restricted to knowledge of whether or notT exceeds a random monitoring timeC . This data format is quite common and includes examples from various fields. Jewell and van der Laan (2004) mention a few examples including: studying the distribution of the age of a child at weaning given observation points; when conducting a partner study of HIV infection over a number of clinic visits; and when a tumor under investigation is occult and an animal is sacrificed at a certain time point in order to determine presence or absence of the tumor.