Interval and doubly censored data and doubly censored current status data are fundamental incomplete data types in HIV/AIDS studies, where the exact time of the disease event is rarely known and the event is usually first detected during a clinical visit. For these types of data the nonparametric maximum likelihood estimator (NPMLE) is a more accurate estimator of the survival function and possesses other nice statistical properties. Recently the applicants have developed an efficient algorithm for computing the NPMLE in interval and double censored data. They propose to develop commercial software incorporating this algorithm and other recent advances in this area into INDUCE, a new product based upon S-PLUS data analysis and graphical environment. Using profile likelihood approach INDUCE will, for the first time, provide for routine joint estimation of the survival function and regression coefficients in a Cox proportional hazards model for interval and doubly censored data, and for doubly censored status data. The efficiency of the algorithm will also make it feasible to put bootstrap confidence bands around the NPMLE and perform significance tests.
