AI system audits ambulance service patient care records

NEWS - eHealthNews.nz editor Rebecca McBeth 

Hato Hone St John has developed an artificial intelligence (AI) system that can clinically audit and triage its patient care records.

The proof of concept, developed in partnership with Spectrum, IBM, and KPMG, finished in early 2025 and Hato Hone St John is now working to rebuild the model in-house. 

The organisations involved say the AI-powered clinical audit system is a world-first in pre-hospital care.

The ambulance service generates around 430,000 electronic patient care records annually from face-to-face patient contacts.

Kirsty Reekers, Hato Hone St John clinical review and assurance manager, says the organisation must audit 5-20 percent of these cases to meet regulatory compliance requirements.

Around 250 volunteers spend 7,500 hours annually completing these, plus one and a half full-time employees.

"It is a huge personnel burden and a huge staff burden at the moment just to keep up with our regulatory compliance requirements," she explains.

The AI system does an initial audit of all patient care records, then triages cases that need urgent assessment by clinicians.

The proof of concept focused on falls patients because they are one of the highest-risk groups, often involving older people and longer ambulance response times.

Reekers says data security was critical when developing the system because of the sensitive nature of patient information. It uses sovereign AI technology, operating within a completely controlled environment without exposing data externally.

The AI system was trained on St John's clinical practice guidelines and treatment protocols using IBM's InstructLab platform, helping it to learn whether care provided was safe and reasonable, rather than simply checking compliance.

"We needed a large language model that could process all of our electronic patient care records, like how clinicians would be able to review them," she says. 

Training involved creating de-identified, synthetic patient care records based on real cases, then staff giving peer review assessments to teach the AI how to make decisions. 

Reekers says an AI system could also potentially be used for speech-to-text integration to audit telephone consultations, real-time decision support prompts, and identifying trends in clinician practice.

"Audit and quality assurance is prevalent across every single healthcare provider worldwide,” she says. 

“This is hopefully going to be a bit of a game changer and pivotal to how we do our quality assurance activities in the future."

To comment on or discuss this news story, go to the eHealthNews category on the HiNZ eHealth Forum

You’ve read this article for free, but good journalism takes time and resource to produce. Please consider supporting eHealthNews by becoming a member of HiNZ, for just $17 a month

Read more AI & Analytics news

Return to eHealthNews.nz home page