2003-2004 Session
October 15th 2003, 2pm to 5pm at MANDEC (Manchester Dental Education Centre), Higher Cambridge Street Joint meeting with Manchester
University's Biostats Group Theme: "Spatial modelling in epidemiology" Modelling of under-detection of cases in disease surveillance TREVOR BAILEY, School of Mathematical Sciences, University of Exeter This talk describes modifications to Bayesian small area disease mapping modelswhichincorporate censoring of case detection in suspect districts, thus enabling estimationof the under-reporting of cases in these areas. It is applied to leprosy incidence in Northern Brazil, producing useful targetting information for improvements to the surveillance programme. Spatially varying coefficient models in ecological studies of environment and health NICKY BEST, Department of Epidemiology & Public Health, Imperial College School of Medicine, University of London Random coefficient regression models may be used to capture geographical heterogeneity in the effects of risk factors on disease outcomes when effect modification occurs at an area level. This talk discusses Bayesian models and explores their sensitivity for detecting different patterns of ecological association. Composite Likelihood Cluster Modelling of Small Area Health Data ANDREW B LAWSON, Dept of Epidemiology and Biostatistics, Arnold School of Public Health, University of South Carolina, USA This talk covers newer developments in non-focussed cluster analysis in relation to environmental risk assessment, and examines the use of very flexible Bayesian pseudo-likelihood models for clustering without prior knowledge of sources of risk, closely linked to nonparametric approaches to smoothing risks. Simulation studies and data will be presented. December 3rd 2003 at MMU, 4.30pm for 5.00pm, Room E30 for tea, E34 for the talk Statistical resources for Business and Industry John Logsdon, Quantex Research Ltd, Manchester Business and industrial applications
have
long been statistical Cinderellas. Fragmented industry, sparse
perceived
demand, disinterested supply and a tendency to use some packaged
approach
has meant that advanced statistical techniques are late at the Ball. HTML
Flyer
for the meeting February
4th 2004, 2pm for 2.30pm at Xaverian College, Rusholme At Headlingley in the Ashes test in
2001 Australia declared their second innings and set England a final
innings target of 315. England won. This was small consolation however,
as they had already lost the series. This is an example of a
decision made in
sport at the highest level that could have benefited from a 'bit
of
Statistics'. Declarations in test cricket are
considered
in this talk and we use a 'bit of Statistics' in the shape of
regression
modelling to develop an aid for captains. In fact, decision problems in
sport abound - when to commit a professional foul in football is a
classic
- early on in a game model leaving your side with ten players would be
foolish;
however late on, when the score is close, it may be sensible to do so. Sports men and women have to make
decisions
during competition, and the extent to which they do so varies from
sport
to sport. In a more obscure sport - mountain marathons - competitors
are
continually making decisions: 'where am I', 'which way?' and 'how
fast?'.
Mountain marathons are mountain running events with navigation from
point to point. Route choices have to be made on the run. Again a bit
of that
old favourite, regression, can help and is used to model running times.
Sailing and Formula One motor racing
are
other good examples where modelling can be beneficial. Sport is
big
business now, so we are not just doing the statistics here 'for sport',
but
we are doing serious applied statistics in sport! March 17th 2004 at MMU, 4.30 for 5.00pm in Room E32 Establishing the probability of Rapid Climate Change Peter Challenor, Southampton Oceanography Centre, Southampton University/NERC North West Europe has a relatively mild climate because the ocean circulation pulls heat North through the Atlantic. If as a consequence of global warming this circulation were to stop our climate would become much colder. In this talk I present some the challenges in estimating the probability of this happening using dynamical models. Peter's slides (4.4Mb)April 14th 2004 at MMU, 4.30 for 5.00pm in Room E32 Statistics, Warrington and their Place in the Evolution of Medicine as an Evidence-Based Science Andy Grieve, Pfizer The famous controlled trial by James Lind in 1747 in which he investigated the effect of oranges and lemons on scurvy is well-known and is often seen as standing alone. In fact relatively recent research has shown that far from being an isolated event, Lind's trial was but one example of many attempts to introduce quantitative measurement into medicine in the 18th century with Warrington providing a particular hot-spot. In this talk I will look at other examples from the 18th century and look at the parallels that there are with modern evidence-based medicine. 26th May 2004 at MMU, 4.30 for 5.00pm in Room E32 (preceeded by a short AGM) Is there Global warming? Time Series Analysis of Global Temperature anomolies. Tata Subba Rao, UMIST One of the hottest(!) topics of the present day (besides terrorism) is Global warming. A number of scientific articles have been written and every day we come across an article by science corespondents with scientific evidence showing that there is significant global warming. We refer to the recent article by Mark Henderson (The Times,Friday March 5,2004) entitled: Past ten summers were the hottest in 500 years. In our present paper we consider time series analysis of the monthly Global Temperature anomalies (observed over the past 150 years) and show through nonstationary models that there is significant increase. Using CUSUM charts and nonstationary spectral methods, we try to detect change points. We believe the change points correspond to significant historical changes in the climate. Our analysis is only a starting point in this new direction. |
