Estimate The Survival Function By Using The Genetic Algorithm

Abstract

  Survival analysis is the analysis of data that are in the form of times from the origin of time until the occurrence of the end event, and in medical research, the origin of time is the date of registration of the individual or the patient in a study such as clinical trials to compare two types of medicine or more if the endpoint It is the death of the patient or the disappearance of the individual. The data resulting from this process is called survival times. But if the end is not death, the resulting data is called time data until the event. That is, survival analysis is one of the statistical steps and procedures for analyzing data when the adopted variable is time to event and time. It could be days, weeks, months, or years from the start of the term registration until the event.

  This research is concerned with the question of estimating the survival function of observational data using one of the most important artificial intelligence algorithms which is the genetic algorithm and that In order to obtain optimum estimates for Weibull distribution parameters, this, in turn, is reflected in the estimation of survival function, whereby the genetic algorithm is employed in the maximum likelihood method, moment method, the least-squares method and the modified weighted least squares method. And for the capabilities of more efficient than traditional methods, and then will be a comparison between the roads depending on the experimental side is evaluated the best way depending on mean square error criterion of survival function, it will also be applied methods on the fact that data for patients with lung cancer and bronchitis.

The study found that the best way to estimate the Weibull distribution parameters and the survival function produced by the experimental side is the hybrid method of the least squares using the genetic algorithm.