Measures of dependence for censored survival data

作者: JOHN T. KENT , JOHN O'QUIGLEY

DOI: 10.1093/BIOMET/75.3.525

关键词:

摘要: In the linear regression model with normal errors squared product-moment correlation provides standard measure of dependence between explanatory variable and response variable. Using concept information gain, a can also be defined for more general models used in survival analysis, such as Weibull or Cox's proportional hazards model. Further, this conveniently estimated even when is subject to censoring, simple approximation available.

参考文章(8)
S. Kullback, R. A. Leibler, On Information and Sufficiency Annals of Mathematical Statistics. ,vol. 22, pp. 79- 86 ,(1951) , 10.1214/AOMS/1177729694
P J Holdsworth, J Thorogood, E A Benson, A D Clayden, Blood group as a prognostic indicator in breast cancer. BMJ. ,vol. 290, pp. 671- 673 ,(1985) , 10.1136/BMJ.290.6469.671
S A Rashid, J O'Quigley, A T Axon, E H Cooper, Plasma protein profiles and prognosis in gastric cancer. British Journal of Cancer. ,vol. 45, pp. 390- 394 ,(1982) , 10.1038/BJC.1982.66
Daryl Pregibon, Data analytic methods for matched case-control studies. Biometrics. ,vol. 40, pp. 639- 651 ,(1984) , 10.2307/2530907
JOHN T. KENT, Information gain and a general measure of correlation Biometrika. ,vol. 70, pp. 163- 173 ,(1983) , 10.1093/BIOMET/70.1.163
JOHN T. KENT, The underlying structure of nonnested hypothesis tests Biometrika. ,vol. 73, pp. 333- 343 ,(1986) , 10.1093/BIOMET/73.2.333
Frank E. Harrell, Evaluating the Yield of Medical Tests JAMA: The Journal of the American Medical Association. ,vol. 247, pp. 2543- 2546 ,(1982) , 10.1001/JAMA.1982.03320430047030