Project Description

I've had the privilege of witnessing Lyn Brodie's work since she was CEO of the Lowitja Institute back around 2010. At a breakfast with her in Adelaide early 2019, she discussed some of her current projects as CEO of Optometry Australia (OA) and floated the idea for a strategic data project and asked if I'd take a look. She had the vision for the project but didn't have the technology expertise to know if it was feasible or affordable.

Sitting with her team a few months later, they began unpacking what they'd ideally like to achieve. The project was indeed visionary for a health association. They wanted to know if it was possible to take the patient records held by practices around the country and analyse these to generate insights useful to the sector. It was a great idea that could leverage decades worth of data already sitting there. But the project posed a number of challenges:

  1. The first was privacy. Any data analysed would first have to be de-identified, and we would need to draw up a number of legal documents to protect Optometry Australia from any accidental breaches.
  2. The second was the different IT systems each practice used. Although each Optometrist was entering data during each appointment, each practice used different software and hardware. This meant we had to create an extraction solution that took all of this into consideration.
  3. The third was data quality. Different optometrists log their notes in different ways. There is not a huge amount of standardisation between practitioners even within the same practice. Even common diseases like Macular Degeneration may be written in different forms of shorthand unique to each practitioner making it hard to write an AI script to reliably perform the analysis.

Believing these challenges were all surmountable, I suggested we undertake a feasibility project to provide more reliable insights into what would be possible. The analysis included:

  • Attending a range of clinics to interview their practitioners, review their systems, and perform a de-identified manual extraction of their data.
  • Creating initial AI algorithms to analyse half-a-million patient records we used as our initial sample.
  • Providing Optometry Australia with a report on data creation and management within the industry, as well as options and implications available to OA as next steps.

If successful, this project would enable Optometry Australia to better lobby government, provide stronger CPD training to members and help practitioners better manage their practices.

Initial Success

Although still ongoing and being scaled up, this project has had a number of significant wins.

  1. We have demonstrated the ability to create a piece of software that could sit in every Optometry clinic in Australia and confidentially and securely mine live health data and centralise this into a national data warehouse
  2. We have shown that our AI algorithm can accurately tell if a patient has or doesn't have a disease 95% of the time.
  3. We can show how disease prevalence has changed through time across different demographics.

We are currently reviewing a set of questions set by Optometry Australia to see just how far we can provide a range of public health and practice management insights.

I hope to take the learnings from this project to help other associations and in countries beyond Australia.

Dashboard concept developed as part of scoping project
No items found.