Artificial intelligence (AI) is helping to identify fast-growing, hard-to-treat breast cancers, according to results from a large Swedish trial involving almost 100,000 women. Detecting these aggressive cancers earlier could save lives and ease pressure on overworked radiologists.

The study, published in the journal The Lancet, analyzed women aged 40 to 80 who participated in the Swedish national breast screening program between April 2021 and December 2022. Mammography is considered the most effective method to detect breast cancer early and reduce loss of life, with European guidelines recommending that scans be reviewed by two radiologists.

However, researchers point out that 20–30% of breast cancers are diagnosed between routine visits, so-called "interval cancers," which typically grow faster, are more advanced at the time of diagnosis, and more difficult to treat.

How was AI used?

In the test, women were randomly divided into two groups:

Standard scan with two radiologists.
AI-supported scanning, where AI marked suspicious areas and the radiologist made the final decision.

The results were impressive. The AI-supported group recorded 11 fewer interval cancers over two years, a 12% reduction compared to standard screening. More cancers were detected during routine scans.

Researcher Jessie Gommers from Radboud University Medical Centre explained that AI functioned as a support tool rather than a replacement for radiologists, highlighting suspicious areas and aiding interpretation.

Workload reduction

Because only one radiologist reviewed the AI-supported scans, workload was reduced by 44%, a major benefit for understaffed institutions. Gommers emphasized the importance of reducing interval cancers, as these types are more aggressive and associated with worse outcomes for patients.

One in 20 women will be diagnosed with breast cancer during her lifetime.
In 2022, an estimated 2.3 million women were diagnosed with breast cancer globally.
About 670,000 women died from the disease that year.
In Europe alone, the WHO recorded almost 558,000 new cases in 2022.

Will AI replace doctors?

Although AI-supported mammography is already being used in parts of Sweden and Denmark, researchers emphasize that AI is not ready to replace human surveillance and more evidence is needed.

Dr Olga Oikonomidou, an expert at the University of Edinburgh Cancer Centre, said AI is intended to help radiologists work faster and more efficiently, not replace them. She also warned that AI systems vary widely and need to be tested with the same rigour as drugs, and not be seen as a low-cost alternative.