学术报告（梁宝生 8.16）

Abstract: In this article, we proposed a semiparametic gamma frailty intensity model to analyze a new type of repeated current status data. In such scenario, the exact onset time and the count of the event of interest were unobservable. The only available information was whether the recurrent event of interest happened or not during some specific observation intervals. The proposed model did not impose strong Poisson process assumption for the counting process of recurrent events. Theoretical properties of the proposed model were established and an efficient EM algorithm along with B-spline functions over sieve space was derived to obtain the maximum likelihood estimates. Simulation studies showed that the proposed method performs well with moderate samples. The approach was illustrated through an application to a STAR*D data.