Premium
Semiparametric dose finding methods: special cases
Author(s) -
Clertant M.,
O’Quigley J.
Publication year - 2019
Publication title -
journal of the royal statistical society: series c (applied statistics)
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.205
H-Index - 72
eISSN - 1467-9876
pISSN - 0035-9254
DOI - 10.1111/rssc.12308
Subject(s) - parameterized complexity , interval (graph theory) , semiparametric model , bayesian probability , computer science , confidence interval , nuisance parameter , function (biology) , econometrics , mathematics , statistics , nonparametric statistics , algorithm , artificial intelligence , combinatorics , estimator , evolutionary biology , biology
Summary A broad structure for the design and analysis of early phase clinical trials has recently been presented. The approach is described as being semiparametric in that the dose–toxicity function is modelled through a parameter of interest and a nuisance parameter. Although very general, the semiparametric method SPM allows for the possibility of specific calibration. In particular, it is shown that we can obtain identical operating characteristics of more richly parameterized designs such as the continual reassessment method. Here, we consider several other designs that have weaker parameterizations than the continual reassessment method, in particular the cumulative cohort distributions design, the modified toxicity probability interval design, the Bayesian optimal interval design and the keyboard design. We show that all of these designs are included, as special cases, in the semiparametric framework. It becomes immediately apparent how to structure any investigation into the operating characteristics of these designs as well as how to tune any design further with the purpose of improving on these characteristics. Simulations are provided to give added support to these findings.