Queensland Health Artificial Intelligence Framework S1: Stage 1, problem identification, risk considerations and concept development (Gate 0): define the problem and SMART benefits, check the AI project repository for existing solutions, confirm strategic alignment, run an environmental and horizon scan, complete the AI Risk Assessment Tool including alignment with the guiding principles, identify options (in-house, partnership, procurement and do-nothing, standalone or statewide) with their resources, users and human-in-the-loop model, define high-level infrastructure and data needs and technical feasibility, identify stakeholders, governance including clinical governance, and regulation such as TGA software as a medical device rules, secure endorser support, and obtain the decision maker's written decision using a transparent prioritisation matrix

What does S1 require?

Stage 1 gives decision makers enough to judge whether an AI solution is worth having. The initiative owner defines the problem, issue or opportunity and the benefits sought in SMART terms; researches whether a related solution already exists inside or outside Queensland Health through the AI project repository; identifies alignment with departmental or Hospital and Health Service strategies and government objectives; conducts an environmental or horizon scan for existing solutions, industrial relations, patient safety and regulatory matters; completes the risk assessment using the Queensland Health AI Risk Assessment Tool or an approved HHS tool, which includes alignment with the guiding principles and informs the decision maker's choice of pathway and effort; identifies potential solutions, whether built in-house, developed with universities, technology companies or consultancies, or procured from a vendor, compared against the do-nothing scenario, noting whether each is site-specific, enterprise or statewide or a new stage of an existing project, the costs, infrastructure, data, skills and workforce each needs, who the users are, what understanding they need and whether the solution is human-in-the-loop, human-on-the-loop or human-out-of-the-loop; defines high-level infrastructure and data requirements and assesses technical ability to implement within or by adapting existing resources; identifies the clinicians, front line staff, consumers and ICT staff affected and whether support exists; identifies governance requirements including clinical governance for clinical settings; identifies regulation to follow, including the latest TGA guidance on software as a medical device and clinical decision support; and discusses the initiative with the endorsers using the concept template. The initiative owner seeks the project sponsor's approval before the decision maker, chosen under local financial delegations, decides transparently and consistently with the principles, using the prioritisation matrix (strategic alignment, alignment to guiding principles, resourcing, effectiveness, timeframe) or an equivalent local criterion, and responds in writing within the governance timeframe stating reasons; outcomes are progression to the next applicable stage, approval pending funding, or no approval, with clinical applications also meeting local clinical governance steps.