UNIVERSITY OF GLASGOW STATISTICS SEMINAR PROGRAMME Wednesday, 11th October, 3 pm Fractal and aggregate tessellations Sergei ZUYEV (University of Strathclyde) Wednesday, 25th October 2000, 3 pm (Glasgow-Strathclyde Partnership Seminar) Perfect simulation and stochastic geometry Jesper M{\O}LLER (Aalborg University, Denmark) Wednesday, 22nd November, 3 pm Cost-effectiveness of new drugs Tony O'HAGAN (University of Sheffield) Wednesday, 6th December, 3 pm Statistical aspects of BSE and nvCJD: implications for the public health Sheila BIRD (MRC Biostatistics Unit, Cambridge) Seminars take place in Room 1f(203), Mathematics Building, University of Glasgow For further information please contact the seminar organiser: Ilya Molchanov University of Glasgow : e-mail: [log in to unmask] Department of Statistics : Ph.: + 44 141 330 5141 Glasgow G12 8QW : Fax: + 44 141 330 4814 Scotland, U.K. : http://www.stats.gla.ac.uk/~ilya/ ABSTRACTS FRACTAL AND AGGREGATE TESSELLATIONS Voronoi tessellation is one of the simplest model for many systems employing subdivisions of space. Quite often, however, more realistic modelling calls for more complex tessellations that could take into consideration the observed irregularity of cells. Let T_0, T_1,... be a sequence of tessellations which cells are associated with nuclei. For each T_0-cell C(x) with nucleus x the aggregate cell C_0^1(x) of level 1 is the union of those cells of T_1 which nuclei lie in C(x). The second level aggregate cell C_0^2(x) is the union of those T_2-cells which nuclei lie in C_0^1(x), etc. Such aggregate tessellations appear naturally, for instance, in modelling of hierarchical telecommunications networks, where the the cells C_0^n represent the service zones of switches (n+1)-levels above in the hierarchical chain. Even if all T_i are Voronoi tessellations, the aggregate tessellations' cells may be empty, they are not, in general, convex nor connected nor contain their nuclei. We present results for aggregate tessellations that are based on stationary random tessellations, mainly Poisson-Voronoi ones. We find expressions for a typical aggregate cell's coverage probability, give bounds on variation of its boundary and study conditions assuring existence of the limit fractal tessellation as n grows to infinity. The talk will (hopefully) be assisted by a computer program PVAT downloadable from the author's web-page, enabling construction and visualisation of aggregate tessellations. PERFECT SIMULATION AND STOCHASTIC GEOMETRY Ordinary MCMC methods are only correct in the limit where an infinite number of steps in the simulations have been performed. A recent topic, which has drawn great attention after the seminal work of Propp and Wilson (1996), is perfect simulation where one is assured that equilibrium has been attained. Perfect simulation seems particular useful for models used in stochastic geometry. In the talk I'll review recent developments on perfect simulation and stochastic geometry based on my own and others research. COST-EFFECTIVENESS OF NEW DRUGS Whereas regulation of new drugs has concentrated in the past on their effectiveness, there is increasing interest in looking at cost-effectiveness. In the UK, for instance, the National Institute for Clinical Excellence has been created to advise the NHS on the best and most cost-effective treatments and procedures. This talk will outline some Bayesian methods for analysing both cost and efficacy data from a clinical trial. I will focus on one example where the importance of prior information is clear, and where I will argue that conventional methods cannot handle 'outliers' in the cost data properly. STATISTICAL ASPECTS OF BSE AND NVCJD: IMPLICATIONS FOR THE PUBLIC HEALTH After a brief historical account which includes statistical aspects of quality control at abattoirs, three current statistical issues of public health importance are addressed: 1. maternal transmission, such as from BSE-dam to calf; 2. whether age-related consumption of bovine meat products is sufficient to explain the younger age of nvCJD cases; 3. design considerations in estimating prevalence of pre-clinical nvCJD. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%