2001-2002 Session
October 17th 2001 at MANDEC (Manchester Dental Education Centre), Higher Cambridge Street at 2pm Joint meeting Manchester University's Biostats Group on "Drug trials and development" Sheila Bird (MRC Biostatistics Unit, Institute of Public Health, Cambridge) Trials and Tribulations: judicial, ethical and NICE This talk will range over issues such as (consumer principle of) randomization, confidentiality versus criminality, informed consent, informative drop-out, database linkage, non-randomized natural history comparison of effectiveness, epidemiology and design of effectiveness trials (pharmaceutical or judicial), costs, and public health risk assessment. Stephen Evans (UK Medicines Control Agency) Statistical contributions to assessing safety of medicines Statistical effort in the development of medicines has tended to be concentrated on assessing efficacy. When a new medicine is licensed its safety is always provisional. The major statistical issues relate to: combining evidence from randomised trials;These topics will be illustrated by reference to hormones and venous thromboembolism and cancer. Stephen Senn (University College, London) The identifiability problem: gene by treatment interaction, pharmacogenomics and cross-over trials It has been claimed that the human genome project will make clinical trials "cleaner" by permitting identification and selection or elimination (as the case may be) of responders and non-responders for treatment. This would then lead to so-called "theranostic" treatment strategies in which genetic diagnosis and gene-tailored treatment are equal partners. I give some grounds for suggesting that this may be more difficult than has been supposed. I also suggest that the generally despised (by biostatisticians) cross-over trial may have a role to play in some would-be theranostic strategies. Stephen
Evans' Presentation (this is a personal view) Dec 5th 2001 at MMU at 2.30pm Joint meeting with the RSS Quality Forum held in Manchester. Max Porter, GlaxoSmithKline, Ulverston Are Six-sigma, Statistics and statisticians fit for purpose? Brief outline of what we are doing within Actives Manufacturing Statistics to support Six-sigma and manufacturing compliance. Issues with limits setting and process capability as they are presented in Six-sigma training and in most statistics text books. Some of our solutions. Can black belts design experiments? Dealing with real experiments. Lessons for Statistics and statisticians. Richard Greatbanks, UMIST Six-Sigma applications in Manufacturing: Barriers and benefits. Six-Sigma is becoming a popular approach to process improvement. The implementation, training and application for several examples are discussed, examining both the barriers and benefits, which can be tangible or intangible. The paper comments on the contribution to improvement in process, product and people, and considers the future of the Six-Sigma technique. Tony Greenfield, Greenfield Consultants DoE by computer for manufacturers Those engineers and scientists in manufacturing industries I know are clear about what they want with design and analysis of experiments. I shall talk about the history of my program, WinDEX: why I developed automatic procedure and the choice of approach and design, show real industrial examples in the DOS version and then show how this has changed in the Windows version with a few suggestions about further developments. John Logsdon, Quantex Research So where do we go from here? Statistics in manufacturing and engineering has been the topic of endless discussion yet little progress seems to have been made. We examine the reasons why such a numerate and potentially useful discipline as statistics is largely ignored by manufacturers and look for solutions that will enable statisticians to become fully involved as well as improving the competitiveness of UK industry. Powerpoint and Publisher presentations available: Feb 5th 2002 at Xaverian College, Rusholme at 2.30pm - note Tuesday! Mathematical Models in Sport John Haigh, Sussex University The ways in which random chance plays a role have been incorporated into mathematical models of sports such as tennis, soccer, baseball, golf. Good models help players and spectators understand the game better, and can give pointers towards good tactics. Examples from several sports will be described. Directions to
Xaverian College March 6th 2002, MMU E32 4.30 for 5.00 Estimating cancer survival in populations - and public health applications Michel Coleman, LSHTM Randomised controlled trials measure the highest achievable survival, while population studies estimate the average survival actually achieved. This simple dichotomy has major implications for both the data and the methodology required to estimate cancer survival. It also affects the interpretation of survival estimates for public health and policy-making, their applicability to recent or current clinical practice, and the feasibility of examining long-term survival. Population-based survival estimates for cancer comprise one of the essential tools for monitoring the efficacy and equity of a national cancer treatment programme. The use of crude, net and relative survival estimates will be discussed in the context of their use as health service performance indicators, as well as for the study of cancer survival trends, socio-economic and geographic inequalities in survival, the proportion of patients who can be said to have been cured, and the number of deaths that would be avoidable if socio-economic inequalities in survival were eliminated. The public health and policy applications of cancer survival data will be illustrated with results from cancer survival studies in England and Wales, Scotland, Europe and the USA. April 16th 2002 (Tuesday) at CIS Limited, Miller St, Manchester M60 0AL ( directions) Joint meeting with Manchester Statistical Society at 5.25pm for 5.45pm (The MSS flyer for this meeting) The Future of Forecasting Robert Fildes, Lancaster University Forecasts are a mainstay of our lives. We live by them and sometimes die by them. Organisations pay large sums of money buying forecasts. Yet for many users they have the scientific status of astrology (and may be given even less credence). This presentation will discuss the key issues in scientific forecasting, using evidence from companies, macroeconomists, statisticians and social and technology forecasters. The story is a complex one - in some areas we have a good record. In others, our pretensions far exceed our abilities. What lessons can we learn - from our successes and our failures? Unfortunately researchers in forecasting have all too often taken the easiest course and neglected to learn from the empirical evidence, preferring the elegant to the effective. What then is the role of forecasting research in this picture? NB We need to let CIS know who will
be coming so please email Liz
Mitchelmore or telephone 0161 837 4011 in good time otherwise you
will be refused admission! Robert Fildes' Powerpoint presentation June 5th 2002 at MMU at 4.30pm for 5.00pm This meeting will include a very short AGM and report A graphic account of structure in models and data Peter Green, Bristol University There has been exciting progress in recent years in methods for structured statistical modelling, but often a perception that these developments are only of theoretical interest, removed from practical realities. I will try to idspel this myth, using ideas from graphical modelling to expose structure in data, and to assist in building models that offer flexibility, computational tractability, and the opportunity to express background knowledge about the phenomenon under study. |
