Keyboard: A novel Bayesian toxicity probability interval design for phase I clinical trials

Fangrong Yan, Sumithra J. Mandrekar, Ying Yuan

Research output: Contribution to journalArticlepeer-review

43 Scopus citations


The primary objective of phase I oncology trials is to find the MTD. The 3 + 3 design is easy to implement but performs poorly in finding the MTD. A newer design, such as the modified toxicity probability interval (mTPI) design, provides better accuracy to identify the MTD but tends to overdose patients. We propose the keyboard design, an intuitive Bayesian design that conducts dose escalation and de-escalation based on whether the strongest key, defined as the dosing interval that most likely contains the current dose, is below or above the target dosing interval. The keyboard design can be implemented in a simple way, similar to the traditional 3 þ 3 design, but provides more flexibility for choosing the target toxicity rate and cohort size. Our simulation studies demonstrate that compared with the 3 + 3 design, the keyboard design has favorable operating characteristics in terms of identifying the MTD. Compared with the mTPI design, the keyboard design is safer, with a substantially lower risk of treating patients at overly toxic doses, and has the better precision to identify the MTD, thereby providing a useful upgrade to the mTPI design. Software freely available at http:// facilitates the application of the keyboard design.

Original languageEnglish (US)
Pages (from-to)3994-4003
Number of pages10
JournalClinical Cancer Research
Issue number15
StatePublished - Aug 1 2017

ASJC Scopus subject areas

  • Oncology
  • Cancer Research


Dive into the research topics of 'Keyboard: A novel Bayesian toxicity probability interval design for phase I clinical trials'. Together they form a unique fingerprint.

Cite this