Comparison of parametric models using right censored data for breast cancer patients
In medical research, time-to-event data commonly happen as it reflects the time until an individual has an event of interest. The event of interest can be the occurence of disease, death or the side effect of the treatment given. However, right censoring is often arising when studying the time-to-ev...
Saved in:
Main Author: | |
---|---|
Format: | Thesis |
Language: | English English English |
Published: |
2018
|
Subjects: | |
Online Access: | http://eprints.uthm.edu.my/329/1/24p%20SYAHILA%20ENERA%20AMRAN.pdf http://eprints.uthm.edu.my/329/2/SYAHILA%20ENERA%20AMRAN%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/329/3/SYAHILA%20ENERA%20AMRAN%20WATERMARK.pdf |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | In medical research, time-to-event data commonly happen as it reflects the time until an individual has an event of interest. The event of interest can be the occurence of disease, death or the side effect of the treatment given. However, right censoring is often arising when studying the time-to-event data. The data are said to be censored when some individuals are still alive at the end of the study or lost to follow up at a certain time. One of the methods to handle the censored observation is the survival analysis. Hence, this study was carried out to analyze the right censoring survival data by using three different parametric models; exponential model, Weibull model, and log-logistic model. Data of breast cancer patients from general hospital in Johor Bahru were used to illustrate the right censoring data. When analyzing the breast cancer data, all three distributions were shown the consistency of data with the line graph of cumulative hazard function resembles a straight line going through the origin. Its show that the parametric models used in this study were appropriate to analyze the survival data. In order to determine the best parametric model in analyzing the survival of breast cancer patients, the performance of each model was compared based on the value obtained from corrected Akaike Information Criterion (AICc), Bayesian Information Criterion (BIC) and mean square error (MSE). Based on the model selections, the log-logistic model found to be the best model with smallest value in AICc, BIC, and MSE. Besides that, a simulation study was also carried out to see the performance of parametric models with a different number of sample sizes. The coverage probability was carried out to determine the accuracy of the simulations study. As the result, the log-logistic model was the best fitted parametric model for the survival analysis of breast cancer compared with the exponential and Weibull model. |
---|