Big data breakthrough to boost child protection services
Mo Data stashed this in Big Data in Healthcare
What began as a simple administrative tool for Australian child care providers may now provide the missing link in protecting vulnerable children from abuse.
More than 1500 child care services nationwide have signed up to the HubWorks portal since the federal government mandated daily reporting of children's attendance in 2008, following legislative changes to the child care rebate framework. The providers use the portal to manage attendance records, enrolments and payments.
Now the woman behind HubWorks, Ruby O'Rourke, is leveraging world-first Australian technology to analyse data held on the portal and prevent children from slipping through the cracks.
It's a labour of love for the Melbourne mum, who suffered abuse herself as a child. With her own daughter now eight years old - the same age as O'Rourke when she became a ward of the state - the passionate businesswoman has come full circle.
Advertisement
HubCare, HubWorks' parent company, is using machine learning to analyse casual observations entered onto the HubWorks portal by the providers and produce a "heat map" of problem areas to identify children at risk. Machine learning explores the use of intelligent computer systems to learn and improve from patterns in data.
For instance, if a childcare worker notices a child has no appetite, they may wish to enter a note to parents. Parents have access to the portal, where they can view information that is relevant to them.
In isolation, such an event is unlikely to lead to intervention. But if other stakeholders in the child's life also enter observations – such as a swimming school instructor seeing bruises on the child's legs – a pattern may begin to emerge.
Ms O'Rourke likened the system to an online Neighbourhood Watch service, allowing citizen observations to be weighted from light to strong to "tell a story" about a child and pinpoint at-risk individuals before reaching the point where an adult formally reports abuse.
"Mandatory reporting is onerous and obliging, with high fragmentation of skill and capability of reporters," Ms O'Rourke said.
In contrast, the HubCare platform worked in a "very native way" similar to what people were used to on social media – with the key difference being that it contained critical data, she said.
Ms O'Rourke said a two-week pilot of the child protection initiative earlier this year identified 20 children at major risk of harm or abuse.
"This indicated our trial was successful in the early stages of assisting government identifying children who otherwise might slip through the cracks," she said.
Some 126 childcare providers participated in the nationwide trial. They recorded 4415 observations and added 1965 comments to children's records.
HubCare wants to formalise arrangements with government to be able to forward its findings to child protection authorities.
The technology has been developed in partnership with Australia's peak IT research body, NICTA, and the CSIRO's digital productivity and services flagship.
Director of Skills and Industry Transformation at NICTA, Simon Kaplan, believes the technology is world first.
"It's very much an emerging project but the hope is that if you can do this well you can save a lot of kids from a lot of trauma," Mr Kaplan said.
"It will never automatically report anyone but will simply act as a flag to humans to then make decisions - and that's extremely important because we don't want to be falsely accusing people because of a pattern in a computer."
CSIRO research leader Alan Dormer said the HubCare platform, with its existing network of community services and 1.4 million individual records, had the potential for broader application as a more efficient, citizen-driven model of government service delivery.
HubWorks customers access the platform daily to manage enrolments, payments and report children's attendance to the Department of Education, as mandated in 2008 following legislative changes to the childcare rebate framework.
Current members include childcare providers, local governments, educational institutions, sports clubs, Uniting Care, Save the Children, the YMCA, HubCare shareholder Community Child Care Co-Operative NSW and more.
Mr Dormer said the team had had some "very encouraging discussions" with both the Department of Communications and the Department of Finance, both of whom have carriage of the federal government's digital engagement strategy.
"It fits very well with the model of reducing costs, allowing the private sector to be more engaged and also providing competition and contestability in the marketplace," Mr Dormer said.
HubCare estimated that sharing critical data from children's records on its systems could cut government expenditure on child protection research and data collection in half.
That could prove crucial for the Department of Human Services, which recently came under fire from residential care groups for failing to protect at-risk children as a result of understaffing and underfunding.
CSIRO has been working with DHS as part of a $16 million, five-year strategic research alliance to improve the efficiency and quality of national service delivery using innovative technology.
"If the machine learning can do the analytics then that enables staff to sustain their energy in actually performing what they're meant to perform - which is intervention and assisting families," Ms O'Rourke said.
Sandie de Wolf, chief executive officer of children's welfare organisation Berry Street and former senior DHS staffer, said technology played "an important role" in supporting child protection system.
A central repository for critical citizen data would also facilitate better sharing of information between the different child protection authorities in each state, and reduce record duplicationThe HubCare platform was 100 per cent opt-in and compliant, with security measures above industry standard, Ms O'Rourke said.
NICTA's machine learning unit was recently recognised as among the top five in the world.
Stashed in: Big Data, Social Solutions
2:31 PM Sep 14 2014