Creating a Research Repository
When Precision Driven Health first put out a call for research proposals in July 2016, we were a little staggered to receive twice the number of submissions we’d expected. In total, 70 proposals on new ways of using data to drive better outcomes in healthcare were put forward for consideration.
As we began to assess each proposal, our conversation always came back to data, data, data. More often than not, data access and checking is more challenging than the data science itself. How could data from one project be useful for another? How do we ensure that data is de-identified, and only accessed by qualified people? What happens if a health provider wants a researcher or clinician to search their records for patterns, without seeing personal details? We took up the challenge to make data management easier and more transparent – for consumers, clinicians, and researchers.
It was then that the idea of research repositories began to take shape. We are fortunate in that our commercial partner Orion Health has created an open data platform called Amadeus. It integrates, stores and analyses existing healthcare data and, having been built using modern technology such as Cassandra and Elastic Search, is being made capable of storing new data sources such as genomic, social and environmental information.
Having technical capability ensures we are well on our way toward safe data analysis, but healthcare data is extremely sensitive, so a critical part of establishing the research repository idea will be ensuring that the appropriate governance, consent and security mechanisms are in place to protect individuals’ data.
We are now in the process of carrying out a privacy impact assessment and creating governance and ethics policies that are guided by the New Zealand Data Futures partnership so that they meet international best practices.
While this work currently involves data in New Zealand – a country at the forefront of digitally enabled healthcare – we are confident that what’s created here will be applicable in other markets. Indeed, the future of health will include real-time research, where people explore data to answer new questions all the time. Our ultimate goal is to one day link to similar repositories around the world and therefore exponentially grow the amount of data available to researchers. Not only will this make health research more cost effective, it will also provide for significantly more accurate models and algorithms to optimise machine learning analysis. This, in turn, will result in more precise decision support tools for healthcare providers as medical practice evolves from population health management to personalised health care.