Performl | AI analytics for preventive care

Outer problem

Preventive care addressing the social determinants of health is proven to reduce the cost of care systems, while improving people’s lives. Governments release open data about people’s needs so the care industry can efficiently target prevention. But this vision has gone unrealised due to information complexity.

For example, the Australian Government has released 250 million open data points to help target $49B in preventive disability care. But the high level of information complexity has stopped thousands of care providers and industry groups from effectively using the data.

Consequently, Australians with disabilities missed out on $11.4B in preventive NDIS care last year, causing $4.6B in lost savings from reduced load on other systems. Australia’s wider $350B care industry faces the same challenge, with spending on health and social services growing 15% last year, while GDP rose just 2.2%.

Inner problem

There is more open data than ever before about people’s health and social support needs to help target preventive care. This immense resource is growing at an accelerating rate. So too is its complexity.

Open data storage is not centralised. Formatting changes unpredictably. Files are incomplete and compromised with errors. Context and metadata is unstructured and unmanaged. Historical records vital for longitudinal analysis are arbitrarily removed. Reporting standards across time and geography change asynchronously, causing major continuity and concordance challenges. Ontology is ungoverned, meaning data is continually relabelled, often incorrectly, by different custodians at different times.

As a result, over time, the same information mutates into unrecognisable, undiscoverable, and unlinkable forms. The data varies from minimally useful to completely unusable. Basic data queries cannot be supported. More valuable queries, involving complex linkages across time, place and cohorts, are totally unachievable. Advanced data science applications, including ML and AI, face massive technical barriers before they are appropriate to attempt, much less leverage with confidence.

The Productivity Commission's Data Availability and Use inquiry in 2017 comprehensively examined how open data could improve competition, increase productivity, and support sustainability in the care industry. It found there could be $64 billion in value if open data was better used, and that open data in the care industry is “an underutilised resource that could be saving lives”, but instead “leads to data gaps and unnecessary expenditure”.

Our solution

Performl is leading a program of deep tech R&D to solve these systemic open data challenges and build the breakthrough solutions needed. Our research program over the last two years has developed a suite of ML and AI tools to help discover and integrate vast open data and power advanced analytics. In particular, we have developed novel data structures to integrate and use hyper-relational data about people's support needs. This graph-like system is influenced by Bronfenbrenner's ecological model of human development. We are developing our technology in partnership with research (UTS Startups), philanthropy (The Snow Foundation) and industry SMEs (e.g. Karitane and Marathon Health).

We have developed a technical thesis to progress our technology to a scalable AI analytics platform for the wider care industry. To achieve this, a complex set of interdependent policy, technical and ethical research problems must be solved.

Research is needed to establish new open data governance, discovery and integration methods, create robust data integration infrastructure, develop a novel query engine for complex data interactions, and innovate an AI query portal for intuitive language-based data inquiries.

Our AI analytics platform is set to change the way thousands of care industry stakeholders use open data to target preventive care. For instance, a care provider looking to target child and family services could simply ask our AI query portal, “How many new mothers in Australia have mental health support needs?”, instead of spending hours combining open datasets.

Going a step further, the platform will handle complex queries like, “What is the cost of this issue? Of the suburbs with a higher than average prevalence, where have rates decreased over time, indicating better outcomes? What evidence-based services are available in these places? What suburbs are likely to see the greatest increase in the number of new mothers with mental health support needs over the next decade?”.

Our platform will unlock sophisticated data science insights from diverse open data sets, providing care industry stakeholders with detailed analysis and actionable knowledge to target preventive care.

This platform will empower a broad group of care industry stakeholders with new knowledge, including government policy teams, service providers, academic researchers, philanthropic organisations, and population health professionals.