Biostatistics Research Group


Biostatistics deals with issues and data relating to biological sciences and medicine.

The Biostatistics Research Group focusses on areas of biostatistics including power analysis and sample size determination, multivariate data analysis methods such as structural equations modelling, as well as longitudinal data analysis and spatial statistics with applications to spatial epidemiology.

The group delivers on four core areas of education, methodological research, collaborative research and service.

  • Education: to educate people about biostatistics through mentoring and teaching graduates as well as offering multidisciplinary courses, continuing education courses and seminars
  • Methodological research: to enhance existing bodies of knowledge in theoretical and applied biostatistics
  • Collaborative research: to provide biostatistical expertise in research studies with scientists in health sciences from assistance in the planning and conduct of the study to analysis of data and interpretation of results
  • Service: to provide statistical support for research projects and participate on local and national committees

Key research areas

Spatial statistics and Geographic Information Systems (GIS) environment

The integration of GIS and spatial statistics in biomedical research started with a very advanced research program in population health involving innovative spatial analysis and knowledge translation using a geomatic approach. Applications were mostly chronic disease and the utilisation of health care services such as cardiovascular disease and risk factors. Research has focussed on:

  • Hierarchical Bayesian models to deal with longitudinal and spatially correlated count data (Binomial and Poisson data)
  • effects of folic acid food fortification on the prevalence of neural tube defects in Canada
  • infant mortality rates within Miami (Florida): racial, ethnic and geographic variations, and
  • childhood lead poisoning in the tri-county area of South Florida.

Multivariate statistics

The work in multivariate statistics focusses on structural equations models for dyadic data and other conceptual frameworks that involves mediation pathway models. Applications range from substance abuse, bio-psychosocial determinants of pregnancy length and fetal growth, students’ medical education as well as to racial disparities in HIV survival. Research has focussed on:

  • pregnancy planning (maternal outcomes) and bio-psychosocial determinants of pregnancy length and fetal growth (infant outcomes) exploring factors other than biomedical such as stress, self-esteem, and their associations with infant outcomes
  • data analysis of the 2003–2005 Youth Risk Behaviour Survey: HIV/AIDS and substance use risk behaviours among tri-ethnic adolescents of Florida
  • retrospective cohort study to estimate the contribution of socio-economic status, segregation, and rural residence to racial disparities in HIV survival, Florida, and
  • measuring socioeconomic inequality in the incidence of AIDS (rural-urban considerations).

Hierarchical and regression trees models

Work in this area focusses on deriving predictive models using recursive partition trees. Research has focussed on:

  • cardiovascular risk factors, in particular, tobacco use, and
  • family history of cardiovascular disease (CVD) as a risk factor of CVD.

Current research study

The current study, “Retrospective cohort study to estimate the contribution of socio-economic status, segregation, and rural”, aims to estimate the role of the contextual factors, namely, community deprivation, segregation and rural versus urban residence in explaining the survival disadvantage among African Americans. It aims to characterise the extent to which these factors change between the time of diagnosis with AIDS diagnosis and death.

The study utilises the HIV/AIDS dataset and some of the US Census indicators explored the social determinants of HIV/AIDS using multilevel modelling.

Areas of study and research

+ Click to minimise