AI-based tool estimates breast cancer risk within next four years

An international team, including Associate Professor Davis McCarthy and colleagues, has developed a breast screening artificial intelligence (AI) tool that estimates a woman’s risk of developing breast cancer more accurately than current methods.

Published today in The Lancet Digital Health, the AI-based tool – called the BRAIx risk score – identified women at high risk of developing breast cancer with remarkable precision. Nearly one-in-ten of those ranked in the top 2% by the tool were diagnosed with breast cancer within four years, despite previously being given the all-clear.

While traditional breast cancer screening has been very successful, it still largely takes a one-size-fits-all approach, with most women screened in the same way regardless of their personal risk. The study found that the BRAIx risk score outperforms factors that doctors typically rely on, such as age, breast density, and family history.

The breakthrough comes from the BRAIx program – a partnership led by Associate Professor Helen Frazer at St Vincent’s BreastScreen Melbourne, which brings together leading experts like Davis, along with St Vincent’s, The University of Melbourne, The University of Adelaide and BreastScreen Victoria, and funded by the Medical Research Future Fund.

Developed using mammograms from nearly 400,000 women screened at BreastScreen Victoria and then tested on data from almost 96,000 Australian women, the results were then confirmed in an independent Swedish population of over 4500 women.

The Lancet study was co-led by Davis, who, with the help of his Bioinformatics & Cellular Genomics team, drove the AI model development and statistical analyses underpinning the findings.

Davis says, “Our ability to develop this state-of-the-art AI technology here in Australia demonstrates both the depth of our local expertise and the broader value of investing in national capability in AI applications in healthcare.”

The risk score offers a promising way to customise screening by using information already contained in breast scan images to better identify risk level. This could help inform screening protocol, with women at high risk requiring closer monitoring, and those at very low risk requiring less frequent screening.

Helen says that she ultimately envisions an AI-supported breast screening program that sets a baseline screen for all Australian women at age 40 or even younger, and is personalised to their future risk, not their age.

“Our AI risk work offers the potential for the most significant step change in breast cancer mortality since screening was introduced in Australia in the early 90s,” she said.

“This study is a testament to what’s possible when clinicians, researchers and data scientists work together towards a common goal. It is built on the legacy of the late Professor John Hopper, whose many years of research in breast cancer aetiology and risk were integral to creating the tool,” said Davis.

While more studies are needed before it is considered for use in routine care, the BRAIx team is on track to achieve its goal of reducing the number of deaths from breast cancer to zero.

“Given the potential we’re seeing with AI in the BRAIx program – its accuracy, and the information it provides clinicians – zero deaths from breast cancer is a real and genuine long-term goal,” said Helen.