Scientist, Cancer Care Ontario

• Spatial Statistics
• Spatio-temporal processes
• Longitudinal Data Analysis
• Computational statistics
• Multi-state models

• CHL 7001 H Models and Inference for Spatial and Longitudinal Data
• CHL 5222 H Longitudinal Data Analysis

Statistical research projects currently include:

• Spatio-temporal models:  With data collected over long time periods, changing boundaries mean models must be defined in continuous space.  The high dimension of the problem which results requires careful model specification and appropriate inference methods.  Current work aims to approximate continuous stationary spatio-temporal processes with discrete processes on a  lattice, and formulate the model in such a way that efficient Bayesian inference methods can be applied.  Related work involves developing smoothing methods for area censored data with changing boundaries.  This work is in collaboration with Jamie Stafford, Steve Fan and PhD student Lennon Li.
• Inhomogeneous clustered spatial point processes: In practise, ecological and epidemiological point process data is both inhomogeneous (with events occurring more frequently in some areas than others) and potentially clustered (with events tending to occur together, beyond what the intensity would predict).  Methods of inference for inhomogeneous shot-noise Cox process models are being developed.  A related problem is modelling and detecting clustering in residential histories, or longitudinal point processes.  This work is joint with Peter Henrys at Lancaster University.
• Multi-state models with time-varying rates, with Peter Neal (Manchester) and PhD student Gerald Lebovic.

Current projects relating the methods to applied problems include:

• Lupus in Toronto since 1965: A long followup period is needed to collect a substantial number of cases, but this introduces the problem of changing boundaries and potentially changing risk surfaces.  A spatio-temporal model is being developed to address these problems.  This is joint work with Paul Fortin, Lennon Li, and others.
• STI's in North Carolina: The goal of this project is to estimate the effect of risk factors, allowing for spatial dependence, and produce maps of risk surfaces over time.  This is joint work with Dionne Gesink-Law.
• Small-area variation in smoking prevalence using the CCHS, with Lutong Zhou.
• Individual-level models for cancer screening and cancer incidence, with Hedy Jiang.

• Henrys, P. A.; Brown, P. E. (2009) "Inference for Clustered Inhomogeneous Spatial Point Processes" Biometrics,
• French, N, Barrigas, M Brown, P E, Ribeiro, P J, Williams, N, Leatherbarrow, H, Birtles, R, Bolton, E, Fearnhead, P and Fox, A (2005) "Spatial epidemiology and natural population structure of Campylobacter jejuni colonising a farmland ecosystem." Environmental Microbiology, Volume 7, Issue 8, Page 1116-1126.
• Brown, PE, Christensen, OF, Clough, HE, Diggle, PJ, Hart, CA, Hazel, S, Kemp, R, Leatherbarrow, AJH, Moore, A, Sutherst, J, Turner, J, Williams, NJ, Wright, EJ, French, NP (2004) "The frequency and spatial distribution of environmental Campylobacter spp. isolated from systematic sampling of a large area of recreation and dairy farmland." Applied and Environmenta
• Brown, Diggle and Henderson (2003) "A Non-Gaussian Spatial Stochastic Processes Model for Opacity of Flocculated Paper". Scandinavian Journal of Statistics, 30, pp. 355 - 368.
• Brown, Diggle, Lord and Young (2001) Space-time calibration of radar-rainfall data, Applied Statistics. 50, pp221-242
• Brown, Kaaresen, Roberts and Tonellato (2000) Blur-generated non-separable space-time models, JRSSB. 62 pp.847--860