Australia is the example, not the boundary
Australia provides a useful worked example because its healthcare environment combines national policy, public and private providers, long-lived clinical systems, modern interoperability standards and strict expectations around privacy, continuity and trust.
The architectural problem is global. Health information must move safely across organisations, technologies and trust boundaries without interrupting care or forcing every participant onto a single platform. The same economic question applies in the United States, across the NHS and in other national and regional health systems: can interoperability be introduced by changing the interaction between systems rather than replacing the systems themselves?
The stickiness of the $14 billion problem
The promise of health information exchange is undisputed: reduced medical errors, elimination of redundant testing and streamlined administrative workflows. The path to machine-interpretable semantic exchange — "Level 4" interoperability — has historically been blocked by a prohibitive cost barrier.
In 2007, Sprivulis and colleagues published a definitive cost-benefit analysis for Australia, projecting a national roll-out cost of $14.2 billion to achieve Level 4 interoperability — and $21.5 billion for the less capable Level 3. The figures were driven largely by the need to replace legacy electronic medical records and build thousands of custom interfaces.
Nearly 20 years later, the friction remains. In 2024, the RACGP warned that endlessly inputting the same clinical information into different systems remains a critical burden for general practice, with the Chair of its Practice and Technology Management Expert Committee describing it as a huge problem with poor interoperability at its core.
The baseline: the high cost of the old maths
The 2007 model assumed a provider-centric, peer-to-peer exchange environment. Its costs were driven by two factors: interface construction and system replacement.
The interface tax. In the traditional model, interoperability requires a unique interface between every distinct pair of systems. Hospital interfaces were estimated at $50,000 each and GP interfaces at $20,000 — and a single provider requires between eight and twenty discrete interfaces to connect with their local ecosystem.
The rip-and-replace burden. To reach Level 4, the 2007 study assumed legacy systems that could not speak modern standards would simply have to be replaced: $5.7 billion projected for clinician office systems nationally and $1.7 billion for public hospital systems. The authors noted general practitioners were unlikely to undertake that investment without significant government incentives.
The current reality: the mandate has arrived. The maths has not.
Policy has moved decisively. The Health Legislation Amendment (Modernising My Health Record — Sharing by Default) Act 2025, passed in February 2025, established the framework for key health information to be shared to My Health Record by default. On 1 July 2026 the first requirements came into force: pathology and diagnostic imaging reports must now be uploaded by default, with Medicare benefits tied to compliance. The government projects indicative savings of $146 million over two years from reduced duplicate testing alone, and weekly report uploads already exceed five million.
The Australian Digital Health Agency's progress reporting shows the majority of actions in the National Healthcare Interoperability Plan complete, the Sparked FHIR Accelerator advancing the Australian Core Data for Interoperability, and thousands of healthcare services registered with Provider Connect Australia.
Yet the delivery mechanism still relies on the old maths. Providers wait for legacy vendors to implement complex upgrades. For many GPs on the ground, the experience remains piecemeal — a mandate that has arrived faster than the infrastructure economics that must carry it.
The paradigm shift: Programmable Data Agents
Data Mediation rejects the premise that applications must be rewritten to become interoperable. The Programmable Data Agent (PDA) is an intelligent control plane that lives on the network between systems.
Protocol-level translation, not application integration. The PDA intercepts data from legacy systems — in proprietary formats such as HL7 v2, XML or even print streams — and transforms it into standardised formats such as AU Core FHIR R4 in real time, before it reaches the destination. A standardised FHIR gateway acts as a facade, upgrading a legacy system to meet current standards without touching its source code. A $20,000–$50,000 custom development cycle becomes a mapping configuration task at a fraction of the cost.
Decentralisation and sovereignty. Data remains in the source system. The PDA network presents unified access to the ecosystem, addressing the burden of patients mentally carrying their history between practices — without centralising the data itself.
Economic analysis: flipping the model
Comparing the 2007 Level 4 model against PDA deployment under current conditions: interface unit costs fall by an order of magnitude, from custom development to configuration. System replacement — $5.7 billion of projected EMR replacement — falls away entirely; the legacy system stays, fronted by a standards-compliant facade. Roll-out timelines compress from a decade to weeks per gateway. And where the 2007 study counted only avoided costs, governed data mobility creates new value.
The AI multiplier. PDAs can mask production data in flight, enabling zero-production-data-in-test environments. Health researchers can train models on real-world longitudinal data without privacy breaches — a research capability that was architecturally impossible in 2007.
Conclusion
The $14 billion price tag for healthcare interoperability is an artefact of obsolete architecture. Policy and standards are now in place — the mandate is live. The implementation burden, however, still falls on providers and legacy vendors operating under the old cost model. By decoupling data mobility from application logic, Programmable Data Agents deliver the semantic standardisation of Level 4 interoperability without the generational debt of system replacement.
Australia is the working example, not the limit of the approach. Any healthcare system with mixed legacy and modern estates can apply the same principle: mediate the interaction, preserve the systems that still serve a purpose and introduce standards, policy and evidence in the data path.
The economic imperative is clear: stop building bridges between silos, and start installing translators on the network.
References
Sprivulis, P., Walker, J., Johnston, D., Pan, E., Adler-Milstein, J., Middleton, B. and Bates, D.W. (2007). The economic benefits of health information exchange interoperability for Australia. Australian Health Review, 31(4), 531–539.
Woodley, M. (2024). Progress being made on 'huge problem' for GPs. RACGP newsGP, 27 February 2024.
Health Legislation Amendment (Modernising My Health Record — Sharing by Default) Act 2025 (Cth); Health Insurance (Share by Default) Rules 2025; My Health Record (Share by Default) Rules 2025.
Australian Digital Health Agency (2025). Annual Progress Report: July 2024 – June 2025.