Muchatibaya, M C2022-12-212022-12-212022https://hdl.handle.net/10539/33920A research report submitted to the Faculty of Science, University of the Witwatersrand, in partial fulfilment of the requirements for the degree of Master of Science, 2022This research report involves the study of a dataset compiled at a single cancer centre of patients with the chronic disease known as Acute Myeloid Leukemia (AML). A semi-parametric model (i.e., the Cox Proportional Hazard (PH)) and four parametric models, namely: exponential, Weibull, lognormal, and the log-logistic were fitted to the data. In fitting the survival models, variables such as patient age at diagnosis, sex, hemoglobin levels, cytogenic categories, and infection status, as well as whether or not the patient had chemotherapy before treatment, were found to be significant in the models. Based on information criteria and forecast error metrics, the Cox PH model, the semi-parametric model performed best in comparison to the parametric models. The Cox PH model had the smallest Akaike’s information criterion (AIC) and Bayesian information criterion (BIC) values and Integrated Brier Score (IBS). TheCox PH model gave the best predictions.enThesis