A new study shows that artificial intelligence (AI) could make breast cancer screenings much cheaper and more efficient—but not by replacing human radiologists.
Instead, the research suggests that the best way to use AI in mammography is through a collaboration strategy called “delegation.”
In this model, AI helps identify low-risk mammograms, allowing human radiologists to focus on more complicated or high-risk cases. This partnership could cut screening costs by up to 30% without sacrificing patient safety.
The study, published in Nature Communications, was co-written by experts from the University of Illinois Urbana-Champaign, the University of Texas at Dallas, and the NYU Grossman School of Medicine.
According to Mehmet Eren Ahsen, a professor of business administration at Illinois, the findings highlight the real value of AI in health care—not as a replacement for humans, but as a tool to support them.
“We often ask if AI can replace certain professions,” Ahsen said. “In this case, the answer is no, but it can certainly help.”
The researchers compared three different ways of using AI in breast cancer screening. The first was the traditional method, where only expert radiologists read every mammogram.
The second was a fully automated approach, where AI did all the screening.
The third method, called delegation, allowed AI to handle the initial scan and flag any ambiguous or high-risk cases for further examination by radiologists.
Using real-world data from a global AI competition, the study found that the delegation model saved the most money—up to 30.1%—while still catching potential cancer cases effectively.
The reason this strategy works so well is that AI is particularly good at identifying low-risk mammograms that are easy to interpret.
This reduces the workload for radiologists, who can then focus their time and expertise on more complex cases.
According to Ahsen, nearly 40 million mammograms are performed each year in the United States alone. With a typical false positive rate of 10%, around four million women are asked to return for additional screenings, tests, and even biopsies that may not be necessary. This process is not only expensive but also stressful for patients, who often wait weeks for results.
The delegation model could change that. With AI quickly identifying suspicious cases, patients could be flagged for follow-up while still at the hospital, streamlining the entire process.
“It has the potential to be much more efficient,” Ahsen said, adding that it could save time and reduce anxiety for patients waiting for answers.
The study also touches on how this strategy could be useful in places with fewer radiologists, like developing countries, where access to health care is limited.
However, Ahsen noted that legal questions remain, especially around liability if an AI system misses a diagnosis. If AI is held to stricter standards than human doctors, some health organizations may hesitate to adopt the technology, even if it is cost-effective.
The researchers believe their findings could be applied to other medical fields, such as pathology and dermatology, where early detection and accurate diagnosis are crucial.
As AI continues to improve, its role in health care is likely to grow.
However, the study emphasizes that the best results come from humans and AI working together, not separately. Ahsen believes that this partnership could make health care not only cheaper but also better and more accessible for everyone.
If you care about breast cancer, please read studies about how eating patterns help ward off breast cancer, and soy and plant compounds may prevent breast cancer recurrence.
For more health information, please see recent studies about how your grocery list can help guard against caner, and a simple way to fight aging and cancer.
Source: UIUC.
