DEPARTMENT OF MEDICAL STATISTICS DE MONTFORT UNIVERSITY, LEICESTER Seminar Announcement Time: Wednesday, 16th February 2000, 1 p.m. Place: James Went Building, DMU Leicester Room JW 6-18 Speaker: Paul Nelson Glaxo Smithkline, Harlow Title: Applications of multivariate data analysis to drug discovery and development Abstract Ever increasing demands for information about drug products, processes performance and the utility or potential of techniques in drug discovery and development has led to the generation of a great many measurements and the identification of a great many variables that can simultaneously influence these measurements. The problem that soon arises is one of locating the relevant information among so much data and interpreting it reliably. The information of interest does not often lie in single variables, but in their combinations. The solution is multivariate data analysis by projection methods such as principal components analysis (PCA) and partial least squares (PLS). By analysing all measurements simultaneously, multivariate projection methods utilise correlations among variables to reduce the high-dimensional coordinate system to one comprising only a few relevant axes. The information can then be displayed in simple graphs. Applications of multivariate projection methods to pharmaceutical problems will be given to illustrate how they can be used to uncover structure-activity relationships, patterns, outliers, influential variables, groupings and separations that would not have been discerned had the measurements been analysed one at a time. See also http://www.cse.dmu.ac.uk/~jdenne/seminar.html Info: Nick Longford, [log in to unmask] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%