{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

[生物书合集].Advances.in.Clinical.TBiostatisti

[生物书合集].Advances.in.Clinical.TBiostatisti

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
intercept parameter. The results suggested that estimation of the MTD was not significantly affected by the choice of prior distribution and that no one prior distribution performed consistently better than the others under a broad range of circumstances. An alternative formulation of the prior distribution, suggested by Tsutakawa (1975) and discussed by Patterson et al. (1999) and Whitehead (1997), is based on a prior assessment of the probability of DLT at selected dose levels. As a simple example, consider two prespecified dose levels z 1 and z 2 . These dose levels need not be available for use in the phase I trial, but often represent doses used in previous clinical inves- tigations. For i = 1, 2, positive constants t ( i ) and n ( i ) are chosen so that t ( i )/ n ( i ) corresponds to a prior estimate of the probability of DLT at the dose z i . The prior for N is then specified as h ð N Þ ¼ n P 2 i ¼ 1 p ð z i j N Þ t ð i Þ ½ 1 ± p ð z i j N Þ n ð i Þ± t ð i Þ where n is the standardizing constant rendering h
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}