Australian-built AI tool slated to lead revolutionary reduction in breast cancer deaths

A landmark artificial intelligence tool developed to detect and predict the risk of developing breast cancer in Australian women could lead to the most significant reduction in deaths from the disease in more than 30 years.

The Australian-built predictive BRAIx tool has been trained to detect breast cancer in Australian women, as well as predict the likelihood of developing cancer in the next four years, by evaluating mammogram images.

Currently, mammogram images are reviewed by two radiologists, and sent to a third if their findings differ.

READ MORE: Son of ousted king positions himself to take over in Iran

The AI tool evaluates the same images, using an algorithm to detect signs or risks of cancer.

Researchers say it doesn't replace the role of a radiologist but assists them to improve the accuracy and efficiency of breast screening.

St Vincent's Breast Screen clinical director Dr Helen Frazer said the algorithm has been trained using real mammogram images from Australia's population screening program to detect signs, or patterns of risk of early cancer "that human eyes can't see".

"We've been able to train the algorithm on millions of mammogram images, many more than a radiologist would ever be able to see in their lifetime," Frazer said.

"It's using simple image classification algorithms to detect signs of cancer or the risk of cancer on a mammogram.

"We can create really accurate tools that have been trained on local Australian populations to improve the population screening program for Australian women."

READ MORE: Melbourne woman recalls 'appalling' sexual assault as police continue search for offender

The BRAIx detection tool is currently being used in a randomised controlled trial at Breast Screen Victoria and is soon to be rolled out at Breast Screen SA.

The technology has expanded to generate risk scores, predicting a patient's likelihood of developing breast cancer within the next four years.

The risk-prediction tool is in research and development, Frazer said.

"This is an Australian first for an AI, high-consequence decisions in health care," she said.

"This will provide us with the necessary evidence for trusts in AI tools for high-consequence healthcare decisions."

Currently, doctors analyse age, family history and breast density to determine the risk of developing breast cancer.

But research shows the BRAIx tool can predict cancer "with much greater accuracy" than ever before.

"One in 10 women that had a BRAIx risk score in the top two per cent of the population cohort that we tested went on to develop breast cancer in the next four years," Frazer said.

"And they had been given an all-clear at [mammogram] screening."

READ MORE: Apple unveils $899 MacBook Neo in surprise launch

"That really is a breakthrough finding... the future for AI-leveraged risk detection is incredibly exciting."

The tool could mean the most significant "quantum change" in breast cancer deaths in Australia since the national breast cancer screening program was introduced for women over 50 more than 30 years ago.

Despite the introduction of the screening program, 90,000 Australian women are expected to die from breast cancer in the next 25 years.

Frazer hopes the Australian-built tool can put a stop to the sobering statistic.

"What we've learned from the screening program is that if we detect cancer early, whilst it's still in the breast and hasn't spread to lymph nodes or distant organs, we have an almost perfect survival [rate] with modern-day treatments," she said.

"If we can make it even more accurate, detect cancer even earlier, and actively invite women at age 40, and maybe in the future at an even younger age, we have the potential to detect every woman's cancer early and save every woman's life."

NEVER MISS A STORY: Get your breaking news and exclusive stories first by following us across all platforms.

• Download the 9NEWS App here via Apple and Google Play
• Make 9News your preferred source on Google by ticking this box here
• Sign up to our breaking newsletter here