Part II: A new NHMRC Project for measuring the impact of social and news media on health behaviours
As promised following Part I – and now that I am back from the burnt orange colours of the United States to the purple jacarandas of Sydney – another update. But first, a quotation from one of the books in a series that inspired most of the work I do. It starts with the Emperor of the Galaxy talking to a research academic:
‘I am given to understand that you believe it is possible to predict the future.’
Seldon suddenly felt weary. There was going to be this misinterpretation constantly. Perhaps he should not have presented his paper. He said, “Not quite, actually. What I have done is much more limited than that…”
We were lucky enough to have been awarded a new 3-year NHMRC Project. With co-investigator Julie Leask and a team of excellent associate investigators from Sydney and Boston, I will be developing a pipeline of interconnected methods capable of linking and tracking exposure to news media at unprecedented scales; estimating differences in information diets by geography and demographics; correcting for biases in each of the data streams we will mine; and connecting all of that information to real health outcomes. In the first instance, those outcomes will be related to vaccines…but we will be rolling out these methods into a range of other health outcomes as soon as we can.
Our major aims for the project are the following:
- To determine whether measures of exposure to news and social media can be used to explain geographical differences in health behaviours (especially vaccines).
- To determine the proportion of our information diets made up of evidence-based information, and characterise the quality of evidence in the news and social media that makes up our information diets.
If you are looking for an opportunity as a postdoctoral research fellow, and this kind of research sounds like fun to you, please get in touch. You will need to be very comfortable with developing machine learning methods for big messy datasets (GPUs and neural networks are a bonus), and have published innovative methods and released your code open source. And if you are considering a PhD, want a generous scholarship, and you can design a project that might align with these ideas (no matter whether your disciplinary background is computer science, medicine, epidemiology, statistics, psychology, sociology, or just about anything else), send me an email with your bio and ideas.