Application of survival models for population studies

The paper deals with the analysis of the two survival models of the accelerated failure-time using two-parametrical log-logistic and Weibull distributions, and survival models using conditional generalized Weibull, log-logistic, and Smith and Bain distributions. The observed survival (number of deaths during the 30-year follow-up period among the study cohort) and the survival predicted by regression models (predicted number of deaths for the same period of time) were compared. Data on deaths occurring in random sample of men were obtained from the death register of the city of Kaunas. The best agreement between the predicted and observed survival was obtained with one of the modified Smith and Bain models.