Advocacy groups and legal experts are demanding transparency around a new government tool being developed to score the needs of people with disability.

The tool procured to “accurately predict support needs” was recently unveiled, prompting calls over the following days from the Australian Federation of Disability Organisations (AFDO) and People With Disability Australia (PWDA) for more information about the inputs used in assessments.

Every Australian Counts independent chair Dr George Taleporos said that the material released did not assuage participants’ “fears about being assessed by strangersand whether the process could be used to limit supports.”

“The NDIS must never reduce us to data points in a secret algorithm – people with disability are not numbers”.

Predicting need

The Minister for the National Disability Insurance Scheme (NDIS) Jenny McAllister said that the University of Melbourne and the Centre for Disability (CDS) had won the eight-month tender process to develop a new assessment framework that will roll out from mid-2026.

The organisations will develop a needs-centred model to replace the current system that assesses participants’ functions based on reports from their own specialists and “take a long time and cost a lot of money for people to source”, McAllister said.

“This is a crucial step towards ensuring that we use the latest technology to make sure that we meet the needs of Australians with disability.”

CDS revealed that it will adapt a tool called I-CAN v.6 for use by certified government contractors in NDIS plan reviews and eligibility assessments that last up to three hours.

I-CAN rates the level and frequency of need against 12 domains, such as “communication” and “mobility”, but is “semi-structured” and “designed to” home in on individuals’ "most relevant support" and "relevant domains", CDS said.

Defining inputs

Laws passed last year stated NDIS plans "must have regard to any information and reports requested...using the assessment tool" and "may have regard to any other information", which 6,272 signatories of a petition said risks prohibiting using “professional reports that the participant submits” to challenge “substandard needs assessments”.

PWDA noted it remained unclear whether the I-CAN would be “part of a broader process…where participants retain the ability to provide reports from their own health and allied health specialists.”

AFDO president Grant Lindsay said I-CAN could "help structure planning" but "cannot replace the nuanced understanding that comes from lived experience and the clinical evidence gathered over a lifetime."

Although CDS’s example “I-CAN statement” shed light on information that would be captured and scored, it did not reveal “how those scores will actually determine participants’ budgets,” University of NSW senior lecturer Georgia Van Toorn told Information Age.

“As yet, there’s no clear process for the determination of budgets, no constraints on the use of algorithms for such decisions, no obligation to report their use, and no proactive testing for potential harms”.

Marie Johnson, the former head of technology authority for the NDIS raised a similar concern when opposing the bill that the government passed to create the needs-based assessment model.

“This bill puts an algorithm – the budget calculation instrument, which is yet to be defined – beyond the reach of administrative review,” she told parliament last year.

The National Disability Insurance Agency (NDIA) declined to comment.

70 per cent inaccurate

The Coalition trialled a model that, like the needs-based system, required government–contracted allied health professionals – unknown to participants – to conduct assessments that calculated budgets by assigning NDIS participants to one of 400 “personas”.

It is unclear how standardisation models and underlying assumptions in current systems calculate budgets and determine NDIS plans, but the rate at which the administrative appeal tribunal overturns NDIS decisions — 70 per cent during H3 2024 — suggests they are largely inaccurate.