Breast cancer is one of the most common cancers among women worldwide. For decades, mammography screening has been one of the most important tools for finding breast cancer early, often before symptoms appear.
Early detection matters because cancers found at an early stage are usually easier to treat and less likely to spread. However, traditional screening is not perfect. Some cancers are missed during screening and are only discovered months or years later, sometimes when they have already become more aggressive.
A new large study from Sweden now suggests that artificial intelligence, or AI, could help improve breast cancer screening in a meaningful way.
The study, led by researchers from Lund University and other European institutions, found that AI-supported mammography not only finds more cancers during screening, but also reduces the number of cancers diagnosed between screening visits.
This is the first randomized controlled trial of its kind and involved more than 100,000 women. The results were published in the medical journal The Lancet.
To understand why this matters, it helps to know how breast screening usually works. In many European countries, each mammogram is read by two radiologists to reduce the chance of missing cancer. Even with this careful approach, some cancers are still overlooked.
These cancers may appear after a woman has received a normal screening result but before her next scheduled screening. Doctors call these “interval cancers.”
Studies suggest that about 20 to 30 percent of these cancers could have been seen on the earlier mammogram. Interval cancers are often more aggressive and harder to treat, which makes reducing them a key goal of screening programs.
In this Swedish trial, women attending regular breast screening were randomly assigned to one of two groups. One group received standard screening, where two radiologists reviewed each mammogram without AI. The other group received AI-supported screening.
In this group, a specially designed AI system first analyzed the mammogram images. Images that looked low risk were sent to one radiologist, while higher-risk images were reviewed by two radiologists. The AI system also highlighted areas that looked suspicious, helping radiologists focus their attention more effectively.
The AI used in the study was not experimental or untested. It had already been trained and checked using more than 200,000 mammograms from hospitals in over ten countries. This helped ensure that the system could recognize many different breast patterns and cancer types.
The results were encouraging. During the screening appointments themselves, the AI-supported group had more cancers detected than the standard group. Over time, this early detection led to fewer cancers being diagnosed between screenings.
During two years of follow-up, the AI group had 12 percent fewer interval cancers than the group that received standard screening. This means fewer women were diagnosed with cancer after being told their earlier scan was normal.
The AI group also had fewer serious cancers. There were fewer invasive cancers, fewer large tumors, and fewer aggressive cancer types compared to the standard screening group.
Importantly, the number of false alarms, where women are called back for further testing but do not have cancer, was almost the same in both groups. This suggests that AI helped find more real cancers without causing unnecessary worry for patients.
Another benefit was a reduction in workload for radiologists. Earlier results from the same trial showed that AI reduced the number of images radiologists had to read by about 44 percent. This is important because many countries are facing a shortage of trained radiologists, leading to longer waiting times and increased pressure on healthcare systems.
The researchers were careful to point out that AI is not meant to replace doctors. A human radiologist was still involved in every screening. Instead, AI acted as a support tool, helping doctors work more efficiently and focus their skills where they were most needed.
Like all studies, this one has limitations. It was carried out only in Sweden, used one type of mammography machine, and tested only one AI system.
The radiologists involved were experienced, so the results may not fully reflect what would happen in settings with less experienced staff. Information about race and ethnicity was also not collected, which means it is unclear how well the results apply to more diverse populations.
Overall, this study provides strong evidence that AI-supported mammography can improve breast cancer screening by detecting more cancers early and reducing the number of dangerous cancers found later. The findings suggest that AI could make screening programs more effective while also easing pressure on healthcare workers.
However, careful monitoring, further research, and cost studies are still needed before AI is widely adopted everywhere. If future studies continue to show similar benefits, AI could become an important part of routine breast cancer screening and help save more lives.
If you care about cancer, please read studies that a low-carb diet could increase overall cancer risk, and vitamin D supplements could strongly reduce cancer death.
For more information about health, please see recent studies about how drinking milk affects the risks of heart disease and cancer and results showing higher intake of dairy foods linked to higher prostate cancer risk.
Copyright © 2025 Knowridge Science Report. All rights reserved.
