People Research Data Commons

The data that health researchers need is distributed across government, research and health service sectors. Researchers and innovators face challenges knowing whether and where such data exists, and if so, how to access it.

For example, to dynamically update dietary guidelines on stroke management, we want to be sure we have the data we need, that it is governed and standardised for re-use, and that we can find it and request permission to use it.

Health information of patients and research subjects must be kept secure and not disclosed during the process of health research or health service improvement. Data custodians therefore require researchers to use secure environments for data analysis. Researchers don’t always have secure environments for the data they want to share or access, or when such secure environments exist, they can create data silos that hinder national-scale multidisciplinary collaborations.

For example, to transform our understanding of ageing by bringing together scores of longitudinal studies of brain ageing and dementia, we want to know how to give researchers secure access, and what systems and processes need to be in place to respect the privacy of the study participants.

To solve the grand challenges facing society, researchers seek insights by combining data from government, research, and health service providers.  Since these data are collected in a wide range of scientific, business or operational contexts, they use different standards and structures, which makes it difficult for researchers to bring these together.

For example, to improve outcomes for people who suffer traumatic injuries in car accidents, we want to integrate data from the trauma registry, car accident insurance claims, accident and emergency hospital records, ambulance records, research studies and the deaths registry. We need to consider any existing standardisation of data structures across these sources, whether common medical concept definitions are in place, and whether there are common identifiers.

Applying advanced analytics techniques like machine learning to national sensitive data collections across multiple secure data repositories raises data challenges that require innovative solutions.

For example, to support diagnostics and improve survival of cancer patients, we want to consider how nationally distributed machine learning over cancer data and imagery can be established. We want the images to remain securely within clinics and yet provide parameters to a model that is supported by data and coding standards and a well curated reference dataset.