A Newborn Bayesian

Project Report: A Newborn Bayesian

My name is Joseph Hanna and I’m currently working on a collaborative project involving members from the Kolling Institute of Medical Research (Kolling) and the Centre for Translational Data Science (CTDS). I’m a trainee biostatistician from the NSW Ministry of Health on a 6 month placement with the Kolling as part of the NSW Biostatistics Training Program. Members on the analysis team for this project include: Prof Sally Cripps, Dr Lamiae Azizi, Prof Jonathan Morris, Assoc Prof Jane Ford and Dr Deborah Randall. The project is looking at trying to find the most important factors leading to an increase over recent years in early planned births (37-38 weeks gestational age births by induction of labour or caesarean). This is important since there is increasing evidence to suggest that delaying planned birth, where possible, is associated with better infant health outcomes.

Previous research conducted by the Kolling has approached the problem using frequentist statistics. While some factors have been identified as contributors to the increase in early planned births, the team are keen to identify whether a Bayesian approach can identify additional factors. This project will adopt a Bayesian variable selection algorithm to gauge which covariates are the most important predictors of having an early planned birth. The Bayesian variable selection algorithm has been presented in a paper by members from the CTDS (David Kohn et al). The algorithm is a great way to assess each covariates importance for prediction since it computes the probability of a particular covariate being included in the final model. An overarching aim of the project is to compare the results of the analysis using the Bayesian methodology to the previous analysis.

Since I’m new to the Bayesian world of statistics and furthermore that this project is around early planned births, I have called myself a newborn Bayesian!

Thank you to Joe for the update on his project!