Using a Penalized Likelihood to Detect Mortality Deceleration

TL;DR

This paper introduces a new penalized likelihood method for detecting mortality deceleration, which improves accuracy and reliability over traditional approaches by avoiding p-values and hypothesis testing.

Contribution

It proposes a novel penalized likelihood approach for the gamma-Gompertz frailty model, enhancing detection of mortality deceleration without relying on asymptotic distributions.

Findings

More accurate detection of mortality deceleration

Provides reliable estimates of model parameters

Outperforms traditional likelihood inference

Abstract

In this paper, we suggest a novel method for detecting mortality deceleration. We focus on the gamma-Gompertz frailty model and suggest the subtraction of a penalty in the log-likelihood function as an alternative to traditional likelihood inference and hypothesis testing. Over existing methods, our method offers advantages, such as avoiding the use of a p-value, hypothesis testing, and asymptotic distributions. We evaluate the performance of our approach by comparing it with traditional likelihood inference on both simulated and real mortality data. Results have shown that our approach is more accurate in detecting mortality deceleration and provides more reliable estimates of the underlying parameters. The proposed method is a significant contribution to the literature as it offers a powerful tool for analyzing mortality patterns.