Fisher Information in a Competing Risks Model Under Inhomogeneous Random Interval Censoring

Abstract The Fisher information is a fundamental concept in the theory of statistical inference. It is crucial to estimate statistical model parameters. The Fisher information gives metrics to the sufficiency of the observed data to estimate the unknown model parameters. If the censoring is random, some observations can be incomplete or censored so calculating Fisher information can be onerous. This calculation stays in the focus of numerous researches. We approached this issue through computing and analyzing the Fisher information applying a competing risks model when inhomogeneous censoring is made using a random interval. The results obtained suggest that under certain conditions, the Fisher information remains useful even for incomplete observations. This is crucial for improving the accuracy of parameter estimation in statistical models.

Hali tarjima qilinmagan