Queensland Health Artificial Intelligence Framework S5: Stage 5, deployment and integration (Gate 4): confirm implementation, integration, rollout and operation plans are approved and current, communicate risk management plans with controls in place, complete user and system testing satisfactorily, prepare training resources and schedules, establish and communicate technical and other support, deliver project assurance and status reporting to governance, update the risk assessment; have system owners deploy and monitor the solution, give production data access to beta testers only, keep version control of source code and models for audit on demand and update the AI solution registry; implement change management with documented new or changed roles meeting HR and industrial relations requirements, train users before deployment, pilot with early adopters, report beta results for approval and roll out in stages assessing impact between them, using a silent trial for complex or clinical use cases, until all impacted staff have access

What does S5 require?

Stage 5 moves the approved solution into deployment and integration with existing workflows so that it reduces rather than adds complexity, and ends with the decision maker satisfied the solution is ready for business-as-usual. The plans for implementation, integration, rollout and operation developed earlier are approved and current; risk management plans are communicated to responsible areas with controls in place; user and system testing is completed with satisfactory results; user training resources and schedules are developed; technical and other support mechanisms are established and communicated to end users; project assurance is delivered to governance against endorsed scope, budget and timelines; status reporting to the steering committee or equivalent begins at agreed intervals; and the AI Risk Assessment Tool is updated for new risks. For technical integration the relevant system owners are asked to deploy and monitor the solution, access with production data is granted to beta testers only, version control is in place for source code and machine learning models to allow auditing on demand, and the AI solution registry is updated with the deployed solution. For the workforce, change management strategies prepare staff, any new or changed roles are documented with clear responsibilities and meet HR and industrial relations requirements, comprehensive training and support are provided before deployment where multiple users take on new or changed roles, pilot testing is conducted with early adopters, a report including beta testing results is submitted for approval, and a phased rollout to all staff follows with impact assessed between phases; for complex or clinical use cases a silent trial may precede the staged rollout. Progression requires the solution to be deployed and integrated, beta testing completed successfully with trained users, and access enabled for all other staff who interact with it, with data and application custodianship documented.