Pessoal,
a conferencia do Peter Diggle no Forum Mineiro de PRobabilidade e Estatistica
http://www.est.ufmg.br/~msantos/forum/
sera' sobre assunto quentissimo em inferencia: como monitorar on-line
processos espaco-temporais procurando detctar anomalias o mais rapidamente
possivel. Seus exemplos cobrem a epidemia de febre aftosa na Inglaterra com
tecnicas de visualizacao maravilhosas.
Aqui vai o resumo:
Statistical Modelling for Real-time Epidemiology
Large volumes of data on a range of health outcomes are now collected
routinely by many
health care organisations but, at least in the UK, are often not
analysed other than for
retrospective audit purposes. Each data-record will typically be
referenced both in time and
in space; for excample, in the UK the temporal reference will be a
date, and in some cases a
time of day, whilst the spatial reference will usually be the
indiviudal’s post-code which, in
urban settings, corresponds to a spatial resolution of the order of 100 metres.
By real-time epidemiology, I mean the analysis of data-sets of this
kind as they accrue, to
inform clinical or public health decison-making. Such analyses would
be triggered and the
results posted automatically, for example on a web-site, by the
arrival of new data.
In this talk I will review work in spatial, temporal and
spatio-temporal modelling that seems
especially relevant to this general task, and will describe a number
of applications, including
some or all of:
• real-time syndromic surveillance (Diggle, Rowlingson and Su, 2005);
• tropical disease prevalence mapping (Crainiceanu, Diggle and
Rowlingson, 2008);
• early warning of incipient renal failure in primary care patients
(Diggle and Sousa, 2009).
CRAINICEANU, C.,DIGGLE, P.J. and ROWLINGSON, B.S. (2008) Bivariate modelling
and prediction of spatial variation in Loa loa prevalence in tropical
Africa (with Discussion).
Journal of the American Statistical Association, 103, 21–43.
DIGGLE, P.J., ROWLINGSON, B. and SU, T-L. (2005). Point process methodology for
on-line spatio-temporal disease surveillance. Environmetrics, 16, 423–34.
DIGGLE, P.J. and SOUSA, I. (2009). Real-time detection of incipient
renal failure in primary
care patients using a dynamic time series model. (in preparation)