Development of Interactive Software for Bayesian Optimal Phase 1 Clinical Trial Design
- PDF / 11,431,048 Bytes
- 10 Pages / 648 x 864 pts Page_size
- 76 Downloads / 149 Views
Williom F. Rosenberger, PCD Professor, Department of Mathematics and Statistics, University of Maryland, Baltimore County. Baltimore, Maryland, and Adjunct Professor, Department of Epidemiology and Preventive Medicine, University of Maryland School of Medicine, Baltimore, Maryland Gerold C. Confield, PhD Associate Professor, Department oflnfonnation Systems, University of Maryland, Baltimore County, Baltimore, Maryland
Development of Interactive Software for Bayesian Optimal Phase 1Clinical Trial Design*
89
We describe an interactive Webbased software, developed by an interdisciplinary team, to sepentially assign dose lev& to patients in a Phase I clinical tial. The methddogy is based on state-of-the-art Bayesian optimal design theq, and has been described in the biostutistics litemture. We desclibe the methoddogy and the development of the software, and give examples using the Web-based p r o g m m . We describe subtleties, such as the elicitation of prim, numerical integmtion, and werdosing constmints.
lnno Perevozskoyo, PhD Biometrician, Merck and Co., Rahway, New Jersey Lindo M. Hoines, PhD Professor, Head of Discipline and Programme Director: Statisticsand Biometry Program. Schod of Mathematics, Statistics. and Infonnation Techndoa, University of Kwazulu-Natal, Pietermaritzburg, South Ahca Petr Hausner, MD Assistant Professor, Department of Medicine, Greenebaum Cancer Center, University of Maryland Medical School, Baltimore. Maryland, and Chief of Hematology-Oncology, VA Medical Center, Baltimore, Maryland
Key Words Bayesian optimal design; Maximum tolerated dose; Prior elicitation; Quantile estimation; User interface Correrpondente Address Professor W E Rosenberger, Department ofMathematics and Statistics, University of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore. MD 21250 (e-mail: [email protected]). 'Grant support: Rosenberger, Canfield, and Hausner were supported by grant ROl -CA87746 from the National Cancer Institute. Haines was supported by the University ofNatal and the National Research Foundation of South Africa.
IN T R O D U C T l O N The focus of this paper is to describe an interdisciplinary team effort to create a user interface software for the sequential design of phase 1clinical trials. The team involved three statisticians, one with extensive computer programming skills, an information systems expert, and an oncologist. The methodology implemented was initially proposed and the preliminary statistical theory developed as part of the doctoral dissertation of Perevozskaya (1).The principal manuscript describing the methodology and developing the necessary statistical theory has been published (2). This work is of principal interest to statisticians, and the MATLAB programs that generate the designs are "userfriendly" only to the original programmer. Hence, we were faced with the task of creating a user-friendly interface for use by clinicians, possibly designing trials without statistical support. We found a number of subtleties in creating a Web-based interface which we
Data Loading...
