In order to estimate the maximum tolerated dose (MTD), O'Qu-igley Pepe and Fisher (1990) proposed the continual reassessment method (CRM) based on the traditional up-and-down schemes in Phase I clinical trials.Recently, a Product-of-beta CRM for the problem is proposed by Gasparini and Eisele (2000,Biometrics 56,609-615). Based on the Product-of-beta CRM, Tao (2001) proposed the Step CRM method. Both the Product-of-beta and the Step CRM discard the specific dose-response curve used in the CRM and are all called the curve-free CRM.Both the CRM and the Product-of-beta CRM (PBP CRM) select a level among the six prespecified doses as the MTD,no consideration of how much drug for each of the dose. This will produce some questions.This paper summarize the background and the main subjects of the CRM firstly. Secondly ,the curve-free CRM are analysised. Lastly, a method for how much drug should be specified is proposed. |