Research
Pharmaceutical research, Stochastic methods and software for the prioritisation of projects in pharmaceutical research, for compound selection and for determining the size of a clinical trial.
My PhD thesis was called Optimal Resource Allocation in Chemical Research, and this has remained my main interest. A large chunk of theoretical work is set out in my 1989 book on allocation (or Gittins) priority indices. These turn out to lead to computationally tractable solutions to a variety of problems, of interest to statisticians, operational researchers and economists.
In recent years my small research group has been developing stochastic models and using them to write software which may be used for planning purposes in pharmaceutical research. The currently available software is as follows.
OPRRA (Optimising Pharmaceutical Research Resource Allocation).
A preclinical research project is described as a sequence of stages, each of which must be completed successfully in order to produce one or more candidate drugs on which to carry out clinical trials. Each stage has a probability of successful completion, and its duration depends on the number of scientists allocated to it. OPRRA calculates the expected profitability of different allocation policies and the optimal policy, in terms of widely used economic indicators. It is described in the 2007 EJOR article.
BeBay (A Behavioural Bayes procedure).
This is a decision theoretic procedure for optimising the size of a Phase III clinical trial. It maximises the expected net benefit, either in monetary or social cost/benefit terms, taking account of the dependence of likely sales upon the strength of the evidence provided by the trial, and hence upon its size, and of the need to satisfy regulatory requirements. It is described, for example, in the 2000 Statistician article.
CPSDAI (Current Probability of Success – Dynamic Allocation Index).
In preclinical research many compounds a screened in the search for those which show sufficient interesting activity to proceed to more thorough investigation, and ultimately to clinical trials. CPSDAI is a statistical procedure which assesses the promise of any given family of compounds for producing a compound with some target level of interesting activity, using the indices CPS and DAI. The idea is to steer the selection of compounds towards those families which will make the number of compounds which need to be screened as low as possible. It is described in the 2003 paper. Dynamic allocation indices, in a more general setting, are the subject of the 1989 book.
More information is available via the linked publication list. Further enquiries are very welcome.
Printer Friendly Version 