|
Activity Number:
|
219
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 10, 2004 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
WNAR
|
| Abstract - #301398 |
|
Title:
|
Statistical Properties of Modified Algorithm-based Designs for Phase I Cancer Clinical Trials
|
|
Author(s):
|
Yong Lin*+ and Weichung J. Shih
|
|
Companies:
|
University of Medicine and Dentistry of New Jersey and University of Medicine and Dentistry of New Jersey
|
|
Address:
|
195 Little Albany St., New Brunswick, NJ, 08901,
|
|
Keywords:
|
algorithm-based design ; phase I cancer clinical trials ; maximum tolerated dose
|
|
Abstract:
|
Algorithm-based designs are still widely used in practice for phase I cancer clinical trials because of its simplicity in logistics for clinical investigators to carry out, compared to various model-based continual reassessment methods (CRMs). Lin and Shih discussed statistical properties of the basic algorithm-based designs. We introduce modified traditional algorithm-based designs, which we encountered in many of the phase I cancer clinical trials. Key statistical properties as well as clinical issues related to these designs are investigated, including: (i) the probability of a dose being chosen as MTD (maximum tolerated dose); (ii) the expected number of patients at each dose level; (iii) target toxicity level (expected dose limiting toxicity at MTD); (iv) expected number of toxicity at each dose level; (v) expected overall toxicity in a trial. The exact formulae for the corresponding statistical quantities will be derived. These statistical properties are important for clinicians who use these algorithm-based designs for a phase I cancer clinical trial to gain insights of the design before starting the trial.
|
