Lehrstuhl Stochastik » Forschungsschwerpunkten » Drittmittelprojekte

NIHGMS-Grant: Cost efficient designs for practitioners

Optimal Experimental Designs

This research will develop efficient and versatile experimental designs for a broad class of models useful in many disciplines. Our focus is in biological and toxicological studies. Design strategies will be implemented on a website and practitioners can use resources on the site to do research on optimal designs and generate tailor-made designs for their problems. Specifically, the main aims are:

to develop innovative methods of constructing optimal designs through a theoretical framework that gives new insights to multiple-objective designs, optimal designs capable of performing a lack of fit test, maximin designs, optimal designs on a restricted interval and the number of support points required in optimal designs for a class of nonlinear models,

to work with environmental health scientists and develop optimal experimental designs for useful biological models, including compartmental models and Michaelis-Menten like models. A specific application is how to design a study for estimating the threshold level for an agent obtained from two compounds using limited data from the dose-response curves from the individual compounds and the joint compound itself,

to construct a user-friendly interactive website with lots of resources on optimal designs and their applications. A major component is to apply results from (I) to develop efficient computer algorithms for generating efficient designs tailored to a practitioner's problem. The website will contain existing and new algorithms for generating designs, links to specific design-related software available on other websites and will also evaluate efficiencies of user-supplied designs under a set of user-selected criteria.