An AI scribe purpose-built for busy emergency departments is transcribing consultations at Adelaide’s Queen Elizabeth Hospital, in a trial run by the Central Adelaide Local Health Network, the network says.
The tool was developed with the Australian Institute of Machine Learning and the Commission on Excellence and Innovation in Health, and trained on South Australian health data rather than a generic overseas model. After positive feedback from clinicians, the trial is being expanded to more emergency staff and to the nearby Sefton Park Urgent Care Hub, with around 500 patients expected to take part over a six-week research phase.
The premise is simple: documentation eats enormous amounts of clinician time. Hand it to a scribe that listens and transcribes, and doctors get minutes back for every patient.
Why it matters
Emergency departments are among the most pressured corners of the health system. A tool that reliably shaves documentation time is not a gimmick — it is capacity, at a moment when the system has little to spare.
Training on local data is the smart part. A model built on South Australian records is better tuned to local terminology and workflows, and keeps sensitive health data closer to home.
The guardrails still have to hold. Clinical AI only earns trust if accuracy is checked, errors are caught and a human stays accountable for what goes in the record. A careful, measured trial like this is how that trust is built.
Sources: Central Adelaide Local Health Network; Premier of South Australia
