The rapid growth of artificial intelligence (AI) in healthcare is creating new opportunities to enhance clinical care, improve efficiency, and support decision-making. But realising these benefits safely requires clear and consistent approaches to governance that consider not just algorithms, but the broader systems in which AI operates.
Despite the proliferation of AI ethics principles globally, healthcare organisations still lack practical, standardised ways to translate those principles into day-to-day procurement, approval, and clinical governance decisions.
To support this need, the Digital Health Cooperative Research Centre (DHCRC), the University of Technology Sydney, and Ausintelli Technology have commenced the second phase of a national research effort to adapt and operationalise the OECD AI Classification Framework (AICF) for use in Australia’s healthcare sector.
This initiative builds on phase one of DHCRC’s research project, which localised the AICF for Australian SMEs and delivered a prototype web-based tool to support early-stage AI categorisation in healthcare.
This next phase will refine and enhance the tool’s functionality and develop a quantifiable, logic-based matrix capable of delivering more granular classification of AI systems used in medical and clinical contexts.
The project will update the existing framework to reflect current regulatory, operational, and practical needs across the sector. It will also incorporate measurable dimensions relating to risk, trustworthiness, uncertainty, and explainability. These refinements will then be operationalised into a functional decision model and validated through real-world case studies to ensure accuracy, usability, and relevance.
DHCRC CEO Annette Schmiede said the project represents an important step toward strengthening responsible AI adoption in the health system.
“Ensuring AI is safe, transparent, and trustworthy in healthcare requires more than technical excellence,” Ms Schmiede said. “It demands governance models that reflect clinical realities and can be used consistently across diverse settings. This project brings together the right expertise to build a practical, scalable approach to assessing AI risk and trustworthiness.”
“By refining and validating the OECD AI Classification Framework for Australian needs, we’re helping establish the foundations for safe, consistent, and future-ready AI adoption across the health sector.”
Distinguished Professor Fang Chen from the University of Technology Sydney said the project will help translate AI ethics and governance principles into tools that work in practice.
“Ensuring AI is safe, transparent, and fit-for-purpose in healthcare requires frameworks grounded in real clinical environments. This project allows us to turn complex AI ethics and governance requirements into concrete, operational decision tools that health organisations can actually apply in procurement, approval, and clinical deployment. In effect, we are building the bridge between ethical principles and real-world clinical use,” said Professor Chen.
“By strengthening the classification logic and validating it through case studies, we’re creating a clear, evidence-informed pathway for responsible AI adoption across Australia’s health system.”
Mr He Huang, CEO of Ausintelli Technology, said the collaboration will advance the sector’s ability to assess AI solutions with greater precision.
“For AI to be trusted in healthcare, organisations need consistent ways to understand risk and make informed decisions. By integrating structured data inputs, model characteristics, and decision logic, we’re helping build a more robust and reliable foundation for assessing AI systems in clinical environments,” said Mr Huang.
The project’s outcomes will inform the development of an enhanced prototype tool that supports consistent procurement, governance, and assessment of AI applications. This tool could also underpin future registries that track key attributes of AI systems used in clinical care, improving transparency and supporting safe deployment across the sector.
By combining conceptual development with hands-on validation, this next phase of research aims to deliver a more comprehensive and internationally aligned framework to guide responsible and evidence-based AI use in Australian healthcare.
Image by Professor Fang Chen: Responsible AI in healthcare cannot be built on principles alone. This image represents the “implementation gap” between high-level AI ethics and safe clinical deployment. The bridge illustrates how governance, transparency, and model auditing must be engineered into AI systems to carry them from abstract policy to trusted, real-world use.
Through this DHCRC-UTS-Ausintelli project, that bridge is being operationalised as a validated, logic-based classification and governance tool that enables health organisations to assess risk, trustworthiness, and regulatory readiness in a consistent and evidence-driven way.
