A New Accelerated Failure Time Model with Censored and Uncensored Real-life Applications: Validation and Different Estimation Methods

This study introduces a novel exponential accelerated failure time (AFT) model, detailing its fundamental properties and characterizations. To evaluate the performance of various estimation techniques, we conduct simulation studies that assess the finite-sample behavior of the estimators. Additionally, we propose a modified chi-square goodnessof-fit test tailored for the new model, applicable to both complete and right-censored datasets. The model’s validity is examined using the theoretical framework of the Nikulin-Rao-Robson (NRR) statistic, with maximum likelihood estimation employed for parameter estimation. Two separate simulation studies are carried out: one to evaluate the proposed AFT model and another to assess the efficacy of the NRR test statistic. Furthermore, the practical applicability of the test statistic is demonstrated through analyses of three real-life datasets.