The people are there to run the Integrated Assessment Tool algorithm, a Senate inquiry has heard.
If humans can’t override the Integrated Assessment Tool algorithm used to classify Support at Home clients’ needs, and their qualitative input isn’t taken into account, then what’s the point of the assessors?
That was independent Senator David Pocock’s question to representatives of the Department of Health, Disability and Ageing during last week’s inquiry into the assessment tool.
The hearing heard the decision to remove the ability of human assessors to override decisions of the IAT was made in the lead-up to the 2024-25 budget.
The committee heard that the ability of human assessors to override priority and classification came into effect on 1 November last year, as a feature of the Support at Home package under the new Aged Care Act.
This article originally ran on TMR’s sister site, Health Services Daily. TMR readers can sign up for a discounted subscription.
Legislation forbids human assessors from overriding Support at Home classifications generated by the IAT’s algorithm.
Julia Atkinson, assistant secretary of home support operations at DoHDA, said the assessment process could take up to three hours for a client with complex needs.
“Under the legislation, the assessment delegate is not permitted to override the classification generated by the algorithm,” she said.
“So what’s the point of the assessor delegate if they can’t override the tool?” Mr Pocock asked.
“They’re basically handed information, and they can’t change it, but they’re the decision maker. What are they doing?”
Ms Atkinson said the point of the assessor delegate was to provide “a point of check and balance within the assessment”.
“It’s assuring that the tool has been … populated in the way that assessors are trained to do,” she said.
“The outcome of the algorithm is not the decision taken by the assessor.
“The decision taken by the assessor is to run the algorithm and to say … this is an accurate representation of the individual that I have just assessed.”
DoHDA representatives were asked who made the decision to use an algorithm to assess individuals for Support at Home packages, and who made it.
“I believe the decision was made by government,” Ms Atkinson said.
“It was always a part of the framework of … what was being developed for the IAT to have a classification algorithm.”
When asked who made the decision and when, Mr Pugh said it was made “through normal budget deliberate processes” in the 2024-25 budget.
And when asked who made the decision to remove the human override function, Mr Pugh said it was “made by government”.
Mr Pugh could not answer the question of when the government was first briefed on the possibility of using an algorithm, saying: “I’ll take that one on notice”.
Mr Pugh also told the committee that around 24% of IAT assessments were conducted over the phone.
Opposition health spokesperson Senator Anne Ruston asked whether all individuals on Home Care packages had been reassessed when they were transferred to the new Support at Home packages.
Ms Atkinson said people were not reassessed when they transitioned to Support at Home. “That was not practical,” she said.
Mr Pugh was asked whether the department had been asked to provide more oversight and override determinations of the IAT, in light of feedback from individuals who report being worse off under the new system.
“We are very alive to what’s happening, and so we will provide advice to government on an ongoing basis,” Mr Pugh said.
DoDHA representatives could not answer the question of how many times assessor delegates had challenged the outcome of the algorithm.
The inquiry also heard that data entered into “free text fields” during initial assessments were not included in the final assessment.
Ms Atkinson said assessors were able to enter some data as free text fields, but those free text fields were not pulled into the algorithm.
“That’s not practical,” she said.
But Ms Atkinson said the free text fields were “not in any sense meaningless”.
“The free text fields assist the assessor in developing a support plan at the end of the assessment process.
“The assessor will finalise that conversation, enter the data, they’ll run the algorithm and understand what funding level the person is eligible to receive, understanding that they will then sit down and write a support plan and an assessment summary for the older person.”
Greg Pugh, first assistant secretary, Access and Home Support at DoHDA, said free text fields were “a very important component of being able to understand the individual’s overarching story and feed into what that support plan development looks like”.
But Mr Pocock asked, “How important can they be if they … have absolutely no bearing on the level of care or priority?”
Related
Ms Atkinson said the assessment was “more than just a level of care or priority”.
“It is more than that. It is a generation of a support plan with a qualified assessor who can help the older person understand how what supports they can access, how they can best plan those supports and to … recommend to them this is what the aged care system has to offer you, and this is what you’re able to … access through the aged care system.”
Mr Pugh said since 1 November 2025, DoHDA had registered around 800 requests for internal review received under the Aged Care Act 2024 in relation to decisions made by assessment organisations.
The committee heard that of those 800, 181 reviews had been finalised and 24 had received a different decision.
The committee heard that development of the IAT began in 2020 with initial prototyping, followed by a small “living lab” trial in 2022 overseen by Flinders university and involving assessors using the tool and actors playing older people being assessed.
The human override function was available during that trial period, Ms Atkinson said, and free text fields were part of that trial.
“The existence and use of the free text fields was always a part of the prototyping, because we knew that that was important,” Ms Atkinson said.
“But through the Living Lab trial, and later the … national live trial, one of the key pieces of feedback we got from assessors in the fields was that they valued … those free text fields.”
In an opening statement, Mr Pugh said “the old system wasn’t working”.
“The high degree of discretion often resulted in people with similar clinical needs receiving vastly different outcomes based on where they live and the human assessor who conducted the assessment.
“This inconsistency created equity across the system, and in some cases, led to people being allocated more funding than needed, and we see this in the $4 billion unspent funds which were accumulated in the old Home Care packages program.
“In 2024-25, approximately 20-25% of Home Care packages were approved at level four, and this was as high 60% in some jurisdictions.”
“Human assessors continue to play a critical role in achieving high quality assessment outcomes by using their clinical judgment, strong communication and interpersonal skills to complete the IAT during the assessment.
“The IAT is only as strong as the information entered into it.”
