Pushpa Publishing House

Journal Menu

Content

Volume 25 (2025)

Volume 24 (2024)

Volume 23 (2023)

Volume 18 (2021)

Volume 17 (2020)

Volume 16 (2019)

Volume 15 (2018)

Volume 14 (2017)

Volume 13 (2016)

Volume 12 (2015)

Volume 11 (2014)

Volume 10 (2013)

Volume 9 (2013)

Volume 8 (2012)

Volume 7 (2012)

Volume 6 (2011)

Volume 5 (2011)

Volume 4 (2010)

Volume 3 (2009)

	Volume 3, Issue 3 Pg 171 - 256 (October 2009)
	Volume 3, Issue 2 Pg 77 - 169 (June 2009)
	Volume 3, Issue 1 Pg 1 - 75 (February 2009)

Volume 2 (2008)

Volume 1 (2007)

JP Journal of Biostatistics

JP Journal of Biostatistics
Volume 3, Issue 1, Pages 17 - 39 (February 2009)

SURVIVAL PREDICTION WITH GENE EXPRESSION PROFILES
Wenqing He (Canada) and Grace Y. Yi (Canada)

Abstract:
There is extensive research on prediction of various clinical phenotypes using gene expression profiles. Success has been demonstrated in molecular classification of different cancer types. However, relatively less attention has been paid to study the connection of gene expressions to time to event of patients such as time to tumour metastasis, an important problem in cancer research. One reason is that traditional survival analysis techniques may not be directly applicable in dealing with gene expression data, as typically the number of genes is much larger than the number of subjects. A primary objective of microarray studies is to identify informative or differentially expressed genes, and based upon them to make predictions on outcomes such as tumor type in cancer research. We develop methods for selecting survival relevant genes which may explain the time to event, and build prediction models, based on those genes, for the survival probability. Specifically, dimension reduction methods are invoked to pick out informative gene profiles that carry survival information. Cox proportional hazards models are utilized to conduct prediction, and the prediction accuracy is assessed by means of the Receive Operating Curve (ROC) method. Extensions to other survival models, such as accelerated failure time models, can be done along the same line. Simulation studies are conducted to evaluate the performance of the proposed methods under various conditions. A real microarray data set is analyzed using the proposed methods.

Keywords and phrases: Cox PH models, gene expression, microarray data, principal component analysis, Receive Operating Curve.

Number of Downloads: 282 | Number of Views: 570