Health NZ to invest $4.4M in AI solutions for breast screening

NEWS  - eHealthNews editor Rebecca McBeth

Health New Zealand | Te Whatu Ora has released a tender worth $4.4 million over three years covering AI reading for screening mammograms and breast density reporting.

The organisation expects AI reading to improve cancer detection rates whilst reducing recall rates and interval cancer rates.

Health Minister Simeon Brown says Health NZ is taking a careful, evidence-based approach to introducing AI into screening and the AI will support, not replace, skilled clinicians.

"Every diagnosis and follow-up decision will continue to include qualified health professionals," he says.

The Request for Proposal is for AI mammography solutions for BreastScreen Aotearoa, and solutions must integrate into the national breast screening programme’s existing blind double-read screening workflow. 

BreastScreen Aotearoa screens around 270,000 women annually, with numbers expected to increase as age extension rolls out to include women up to 74 years old.

Brown says the proposed AI technology would undertake one of the two independent reads currently required in the mammogram assessment process.

He says that international studies show that AI-supported mammogram reading can improve cancer detection rates, including identifying cancers that might otherwise be missed or detected between routine screening rounds.

“As demand for breast screening continues to grow, we need to ensure we are making full use of safe, proven technology to support timely, accurate results for women," says Brown.

The roll-out will be via three-phases, beginning with validation in July 2026. Phase one includes retrospective trials comparing AI reading performance against historical ground truth, including datasets using Māori, Pacific and Asian ethnicity mammograms, followed by prospective trials comparing AI performance against radiologist reading.

Phase two is implementation, including integrating AI solutions into existing breast screening workflows and technical platforms before deploying into production. The final phase covers post-implementation delivery across eight lead providers. 

Solutions must work with all BreastScreen Aotearoa mammogram machine types and integrate with both Sectra PACS and Intelerad systems used across the programme.

Data sovereignty requirements mean that patient health information must remain within New Zealand or Australia jurisdictions and all AI models must be locked, so they cannot learn from production data, with any updates requiring customer approval before release.

The deadline for proposals is 22 June 2026.

This RFP follows the release of an RFI in February this year.

You’ve read this article for free, but good journalism takes time and resource to produce. Please consider supporting eHealthNews by becoming a member of HiNZ, for just $17 a month.

Read more AI & Analytics news

Return to eHealthNews.nz home page