NSW Health uses AI to help map staff access to digital patient record system

Marc Pelusi (Image credit: AWS)

NSW Health is using artificial intelligence to assist with determining role-based access for staff to the state’s new single digital patient record (SDPR) as they are progressively onboarded.

Associate director of portfolio and AI delivery at the NSW Health SDPR Implementation Authority, Marc Pelusi, told the recent AWS Public Sector Summit that the AI uses Amazon Bedrock in Health’s AWS tenancy, which sits in the cloud provider’s Sydney region.

The core technology for the SDPR is supplied by Epic Systems and hosted in AWS.

Staff from local health districts and specialty health networks are being progressively introduced to the SDPR over five tranches between now and the end of 2028.

Palusi said that “one of the hidden challenges we have is making sure every staff member has the right access to the right parts of SDPR on day one.”

He said that meant “analysing hundreds of thousands of workforce records with different job titles and structures.”

“This work was manual and intensive,” he said.

“Working with AWS, we used Amazon Bedrock inside our secure SDPR environment to help review our workforce records and recommend job categories from existing systems into SDPR. 

“This really helped to rapidly allocate roles and we had our implementation specialists review every recommendation given how critical it is at go-live to have the right role.”

In the tranche A go-live, 6000 roles had to be mapped for role-based access to the SDPR.

Pelusi said that without the assistance of AI, this activity “would have taken roughly 30 days of intensive effort”.

With AI assistance, it was “reduced to just days, roughly an 80 percent improvement, while increasing consistency.”

“If you expand that to [our] 178,000-person workforce, the impact is huge. and I think our implementation specialists are much happier for the help with the tool, too,” Pelusi said.

Pelusi said that his approach to using AI as part of the SDPR program was always based on “asking where [it] could help us deliver the program better].”

“One thing we decided early was we weren't going to wait [to make use of AI],” he said.

“The [role review] example really sums up our philosophy for AI at this time. 

“We use it to remove repetitive work, improve consistency, and really free up our experts to focus where their judgment has the most value. 

“So, the AI recommends, and people remain accountable.”