PhD Final Seminar – Aleysha Thomas

Ensemble statistical modelling of risk factors in health

Student: Aleysha Thomas

When: Tuesday, 31st October 2017 10:00 AM-11:00 AM

Where: GP-E Block, Level 4, Room 410

Supervisors:

  • Distinguished Professor Kerrie Mengersen (Principal)
  • Dr Nicole White
  • Dr Leisa-Maree Toms
  • Professor George Mellick

Panel Members:

  • Distinguished Professor Kerrie Mengersen (Chair)
  • Dr Nicole White
  • Dr Leisa-Maree Toms
  • Professor George Mellick
  • Associate Professor James McGree

Abstract:

Parkinson’s Disease (PD) is a multi-factorial neurodegenerative disorder that affects 1-2% of the population over the age of 60 years. PD is well studied but much of its aetiology and pathogenesis is still largely unknown. The restrictive nature of the symptoms of the disease results in a poorer quality of life which further deteriorates during its progression. PD is expected to become a predominant social and economic burden on our societies as the population ages. The disease is difficult to diagnose and its onset is often is often mistaken for healthy ageing or as symptoms of other diseases.  There is currently limited literature on the age at PD onset. A clearer understanding of what influences age at PD onset could lead to better clinical assessments and timely disease management. Non-genetic risk factors have been shown to have a strong influence on the incidence of PD. The individual effects of some factors have also been studied in association to the age at PD onset, including organochlorine pesticide (OCP) exposure. This thesis aims to identify and characterise the combined effects of non-genetic risk factors, particularly quantitative measurements of OCP exposure, on the age at PD onset using an ensemble modelling approach. This utilises a combination of statistical models so that inferences from the first model inform the next model. The statistical models adopted for the ensemble approach are linear models, decision trees, hierarchical models, meta-analyses and Bayesian networks. The ideal practice to investigate the non-genetic PD risk factors is to study a population based cohort which are expensive to follow and examine. Only multiple disparate data sources on PD and OCP exposure were available for this thesis. Therefore each data source is individually analysed before its inferences are integrated using ensemble approaches. Results of the PD data sources demonstrates the complexities of predicting age at PD onset with non-genetic risk factor information as well as the stratification of relative influence of risk factors on age at onset by the different age ranges. Inferences from the serum OCP exposure data source fills a research gap on the effect of interactions between age group, gender and time on the overall decreasing trend OCP exposure in humans. These inferences on the association between age at PD onset and non-genetic risk factors as well as quantitative OCP concentrations are integrated with combined odds ratios on the association between age at PD onset and non-genetic risk factors from previous literature to ascertain the combined effects of risk factors on the age at PD onset. We infer that OCP exposure has a strong impact on an earlier age at PD onset along with a family history and prior head injury.  Although the results of the analysis are not conclusive due to the disparate nature of the data sources and the assumptions required to run the final ensemble model, these results justify further study on the combined effects of quantitative non-genetic risk factors, particularly OCP exposure, in a well-designed population study. Additionally, the analysis indicates the usefulness of a meta-analysis and BN to prospectively integrate disparate data to understand associations between variables when it is expensive and time-consuming to coordinate a large scale study.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s