General Notices

The Department of Mathematics and Statistics invites the campus community to attend a Ph.D. Seminar presented by Ms. Ebiakpo-Aboere Sonron of the Faculty of Science and Technology.

Title: A Bayesian gamma-Weibull Frailty Model with Application to Cataract Surgical Data                           

Seminar Details

·         Presenter: Ms. Ebiakpo-Aboere Sonron

·         Department/Faculty: Department of Mathematics & Statistics, Faculty of Science & Technology

·         Date: Monday, 2 February 2026

·         Time: 11:00 a.m. – 12:00 p.m.

·         Venue: Department of Mathematics and Statistics Seminar Room, 2nd Floor, Natural Sciences Building, FST.

Online (Zoom): 

https://sta-uwi-edu.zoom.us/j/99102961395?pwd=QGqaG1XNVPUeK7Wxjo20eYmXDFznEV.1  

Meeting ID:  991 0296 1395                                      

Passcode: 253912

Abstract

In survival analysis, frailty models are used to take into consideration the unobserved heterogeneity in each person’s risk of illness and death. In order to detect unobserved heterogeneity and investigate variables that affect the length of cataract surgery, this study presents a univariate multiplicative gamma-Weibull (G-W) frailty model. The Markov Chain Monte Carlo (MCMC) approach is used in the Bayesian estimation procedure to estimate the model's parameters. The best frailty model is chosen by comparing several loss functions and then used to analyse cataract surgery data from a public tertiary care facility in Trinidad. Further comparative analysis is conducted with various regression and Cox proportional hazard (PH) models. Akaike information criterion (AIC) and Bayesian information criterion (BIC) were used to select the best models. The results show that frailty is significant and the presence of unobserved heterogeneity in cataract surgery duration. The factors age, technique, and hypertension were significantly associated (p<0.05) with cataract surgery duration. The study highlights the advantage of using a frailty model compared to traditional regression models when unobserved heterogeneity is present, as it can account for latent risk factors not captured by observed variables.