With hospital emergency departments facing chronic overcrowding, researchers are working with doctors on ways to reduce the number of people needing emergency care. One South Australian project is creating a way for doctors and patients to ‘get ahead’ of their symptoms and reduce the chances of ending up in hospital.
Actionable Intime Insights, or AI2, is an artificial intelligence program developed by Flinders University’s Digital Health research team, headed by Associate Professor Niranjan Bidargaddi who is supported by the Barossa Fleurieu Local Health Network. Created in consultation with doctors, clinicians and patients, AI2 aims to reduce the impact of mental illness on patients by allowing them to prevent relapses before they even occur.
Up to 80% of mental health patients will relapse within five years of their initial treatment, so reducing these setbacks will make a significant difference for patients and the health system itself.
AI2 compiles data from several digital sources, including Medicare, electronic health records, and My Health Record. It then analyses that data to recognise patterns that may indicate an impending relapse of symptoms before they become noticeable to doctors or the patient themselves.
As healthcare is increasingly digitised, the information collected and stored about patients is increasing substantially. However, as that information accumulates, often across separate platforms, it can conversely make it more difficult to recognise early warning signs and trends.
“The challenge is how we turn all that information into care that is more patient-centric and responsive to early warning signs, rather than reacting to a crisis that has happened,” says Bidargaddi. “AI2 is a data curation technology that transforms those accumulating health reports through algorithms and turns them into intervention responses for health professionals.”
“AI2 is an attempt to provide the right information at the right time.”
With early-warning signs detected in patient records, AI2 updates healthcare providers about a patient’s status and prompts them to follow up with the patient about their care. However, this risks overburdening an already stretched health system by flooding it with notifications, admits Bidargaddi. Instead, his team are now extending AI2 to first provide ‘nudge’ suggestions to patients that help them take steps to self-manage their condition. If the algorithm detects no improvement for the patient, it only then escalates to a healthcare provider.
That app is being created with patient groups and will effectively work as an everyday navigator, leading the patient along a strategy depending on their circumstances. “That’s the missing link now. Patients currently go to a GP or a hospital. But in between, there is nothing.”
The Digital Health team partnered with SA Health to trial the program, and AI2 is now used by mental health clinics in four regions around the state. The algorithms and how they’re used are being further developed in partnership with Digital Health SA and the national Digital Health Cooperative Research Centre.
Another partnership vital for the research, says Bidargaddi, was with Health Translation SA. HTSA invested over $1 million in funding for the project from three Australian Government Medical Research Future Fund rounds. And this, he says, was invaluable in helping the project progress.
The three funding rounds helped the Digital Health team develop and build a prototype, turn that prototype into a pilot study in one clinic, and then expand the trials across multiple clinics. Without the funding, the research would not have been able to develop to the point where it can help patients.
However, HTSA provided more than just funding, say the researchers. In particular, HTSA helped the team shape the project and build relationships with consumers to provide input based on their lived experiences. While the researchers came up with ideas and research directions, this consumer input helped refine those ideas and boost the effectiveness for those who use AI2.
The project has been so positively received that the health systems using it have adapted their services to incorporate AI2’s input more effectively. This includes offering more proactive support, rather than reacting to patient crises.
One clinician who has used AI2 is Jörg Strobel, a Senior Consultant Psychiatrist with the Barossa Fleurieu Local Health Network. “It (AI2) gives you a pattern of behaviour, and deviations from what you expect… is a starting point for a conversation: ‘Tell me more, help me understand’.”
“If you want a quality-driven health system, you need to gather data, and you need to look at that risk data longitudinally,” says Strobel.
With the successful rollout of AI2 helping mental health care, the researchers are now turning their attention to other chronic health conditions. Using the same approach, Bidargaddi’s team believe they can have a similar impact on diabetes, cardiovascular disease and other chronic conditions, which often occur together in patients.
“We want to extend this platform to people with multiple chronic conditions… to set this up to help GPs to optimally support groups that have those multiple conditions,” says Bidargaddi. “It’s very complex for GPs to work out how to prioritise (disease management and care) for someone who has 4 or 5, which is the bulk of people with chronic conditions.”
“It comes down to making sure you have the right medical appointments at the right time, the right tests at the right time, and the right medications.”
“The system we have built is designed to identify when there are gaps occurring and turn that into an intervention response.”
And that, he says, will mean less stress on the health system and better care and outcomes for patients.