Anglicare Victoria secures its AI adoption

Anglicare Victoria is embracing the adoption of AI to support its community and care services work, with a mix of guardrails, usage guidelines, education and tooling in place to ensure security and safety as it progresses.

Earlier this year, the community services organisation commenced a structured three-month pilot of Heidi AI with a small group of frontline staff.

Heidi, essentially an AI digital scribe, is becoming widespread in its adoption across general practice clinics and hospitals to capture clinical notes, allowing doctors to focus more on the patient presenting.

Anglicare Victoria’s Director of IT Services, Dean Mills, told the iTnews Podcast that he saw parallels between clinical and community services settings, and wanted to test whether Heidi AI could prove to be of real benefit to the latter.

“Going back close to 18 months ago, I observed what Heidi AI was doingin the health sector, and my view was whilst the platform isn't developed necessarily for community services organisations, the use case is similar but different,” he said.

“If you think of a GP with their patient, the last thing they want to do is to be trying to write down everything while they're with the patient, and yet [they need to] try and enrich that engagement with the patient to get as much information as possible.

“What tends to happen is they don’t capture everything during that consultation, they’ve probably got a consultation straight after, and so it's hard to find time to then go back and add to their notes, or fully recall everything discussed.

“More importantly is the administrative overhead and the cognitive load that is placed on the healthcare professionals.”

Mills said that in community services, a case worker “may spend 45 minutes with a client, or family”.

“We also don't write down things in front of the family,” he said.

“[So], you're trying to get that enriched engagement, trying to get as much information as possible, you can't capture everything because you cannot possibly remember everything, and then you have to go on to the next client or next family.

“That adds up, and at the end of the day when you do your case notes, how do you remember absolutely everything?”

Mills saw a number of potential benefits in using Heidi AI in a community services setting.

“I saw this as a tool by which we could enrich that client engagement; reduce the administrative and regulatory overhead, which is required - for good reason - in the vertical in which we work; get consistent content, comprehensive content, structure in case noting; and lastly, remove that cognitive load [on case workers],” he said.

Technology-enabled care

Heidi AI is, in many ways, just the latest manifestation of a broader transformation goal that Anglicare Victoria set itself, to harness technology to improve its operations and frontline service ability.

“There's been a real transformation in Anglicare Victoria over the past seven-to-eight years, and I would say probably in the last two years there’s been a real acceleration, where we’ve recognised the importance of technology as an enabler of not only just service delivery for frontline staff, but as an enabler for business across the organisation,” Mills said.

“When I came into the organisation, it was about making sure that that foundational layer was solid: how we integrate, how we scale, and importantly, how we secure.

“In the past two-to-three years, the shift is around how we automate and how we leverage technology to enable our frontline staff to really do their work with the least disruption or manual work possible, which we've been able to do through the use of technology.”

Technology in this sense has included more mobile devices at the frontline, along with supporting applications such as a case management system that provides a single, integrated view of services provided to each client.

Heidi AI builds on these foundations - but given the often sensitive nature of community support programs, adoption of the technology has been accompanied by a significant focus on doing so securely and safely.

“We've taken a very measured approach [to AI adoption]. We've been very strict, very guarded, and very measured around how we are using it,” Mills said.

“We have usage guidelines on the use of Heidi AI.

“More broadly, we have generative artificial intelligence acceptable use policy. There's been communications throughout the various divisions or regions, so people have an understanding of what it is, how it works, the differences and risks with an open AI platform and a secure AI platform.

“On top of that, we've got controls in place with what we call zero trust secure access with TrendAI.

“We have our files, our information, our data classified, and there's three levels of data classification. Protected is the most sensitive, so it might have sensitive PII in there, and we’ve blocked ingestion of any file into an open AI platform that has that protected marking.”

In securing the use of AI within the organisation, Mills said a key starting point was understanding the “attack risk surface” presented by the use of AI tooling, and to act based upon that.

Anglicare Victoria uses TrendAI’s Vision One cyber security platform for threat defence, risk management, and AI security, and to apply appropriate security and risk controls across its environment.

“What it does is it pulls in telemetry from [across our environment] into a dashboard in real time, assesses what the risks, if any, exist within the environment, and applies a workbench rating to it, depending on what the risk is, and that's driven by AI within the platform,” Mills said.

On top of TrendAI Vision One is a managed detection and response capability for actually responding to any risks or potential threats detected in real time.

“No environment is bulletproof, but having visibility is where it starts, and it's also around the mitigating controls that you've got in place,” Mills said.

Mills noted that there are also security requirements built into the company’s tendering processes and evaluation criteria for potential new partners, which again reinforced its position on adopting emerging technologies like Heidi AI in a safe and secure manner.

Heidi AI, he added, had been configured and set up to be “secure by design” in the way it operated, and in the way that information is captured is checked and recorded in the central case management system. Any sensitive PII is also de-identified by design in the platform.

"We are now scaling the implementation of Heidi AI for frontline staff in partnership with Heidi Health in a staged manner of which we continue to measure the benefits with this further investment," Mills said.