One of the biggest challenges to researchers looking into health questions on a broader scale is getting access to individual health data from a lot of people.
But what if you could start tackling some of these health questions without having actually to access all those individual records?
Researchers from the ARC Centre of Excellence for Mathematical and Statistical Frontiers (ACEMS) at QUT developed new statistical methods for analysing summary data that can downloaded on many online health atlases.
Farzana Jahan is a PhD candidate with ACEMS at QUT and the lead author of the research that was just published in the Royal Society Open Science.
“There are many disease maps and atlases that provide summary measures of disease incidence or survival. Our research provides a method for further analysing that information in such a way that questions beyond the scope of those atlases can be explored,” says Ms Jahan.
Because of privacy and confidentiality constraints, the individual-level data obtained by groups that publish health and disease atlases are modelled and shown at a population level for certain geographic areas. The individual-level health data are not available for anyone to see, including researchers who would like to explore additional questions.
“If researchers could model the available online health data, there would be more opportunities to gain new insights that might benefit society," says Ms Jahan.
“For example, are there socio-economic or environmental factors that may be influencing disease incidence or mortality?”
The ACEMS researchers used Bayesian methods to model the estimates of disease without having access to the individual-level data.
farzana-atlas-image.jpg
Australian Cancer Atlas showing melanoma incidence and deaths
“The Australian Cancer Atlas team compared our results with those obtained from the individual-level data. Our proposed method was able to reveal differences in the distribution of geographical estimates by key variables such as Remoteness and Area disadvantage without needing access to those individual-level data,” says Ms Jahan.
The authors believe their method of using published aggregated health data could prove useful for other researchers looking to investigate health-related questions but aren't able to because of the difficulty or inability to obtain individual-level health data.
“Not only that, when researchers get access to individual-level health data, often the analysis is based on a particular project, and they may not have the additional time or the budget to conduct further research on that data," says Ms Jahan.
The other ACEMS/QUT researchers involved with the project are Dr Earl Duncan, Dr Susanna Cramb, and QUT Distinguished Professor Kerrie Mengersen, who is an ACEMS Deputy Director. Professor Peter Baade from Cancer Council Queensland is also on the team.
ACEMS Media Contact: Tim Macuga, 07 3138 6741, timothy.macuga@qut.edu.au