Title:An Empirical approach to Survival Density Estimation for randomly-censored data using Wavelets

Abstract:Density estimation is a classical problem in statistics and has received considerable attention when both the data has been fully observed and in the case of partially observed (censored) samples. In particular, the non-parametric approach for density estimation has become a popular topic of research in statistics due to its robustness against assumptions about the underlying distribution. In survival analysis or clinical trials, a typical problem encountered in the data collection stage is that the samples may be censored from the right. The variable of interest may be prevented to be fully observed due to the presence of a set of events that occur at random and potentially censor the data. Consequently, developing a methodology that enables a robust estimation of the lifetimes in such setting is of high interest for researchers. In this paper, we propose a non-parametric linear density estimator using empirical wavelet coefficients that are fully data driven. Indeed, we derive an asymptotically unbiased estimator based on an inductive bias correction procedure for an estimator constructed from the complete sample. Also, we provide upper bounds for the bias and analyze the large sample behavior of the expected $\mathbb{L}_{2}$ estimation error based on the approach used by Stute (1995), showing that the estimates are asymptotically normal and possess global mean square consistency. In addition, we conduct a simulation study to evaluate proposed approach using different exemplary baseline distributions with different sample sizes. In our study, we choose a censoring scheme that yields an average censoring proportion of 40\%. Finally, we apply the proposed estimator to real data-sets previously used in other published results in the field of density estimation, showing that the wavelet estimator provides a robust and useful tool for such endeavor.