Related videos:
An innovative artificial intelligence (AI) system developed in Australia promises to revolutionize breast cancer detection by identifying women at high risk of developing the disease, even after receiving a mammogram that is considered normal.
According to a study published in the journal The Lancet Digital Health, the tool, called BRAIx, analyzes mammographic images using deep learning algorithms and calculates an individual risk score for the years following the examination.
The research, conducted by the University of Melbourne, was based on data from more than 96,000 women who participated in the public BreastScreen program in Victoria.
Scientists calculated the BRAIx risk index based on mammograms, estimating the likelihood of each patient developing breast cancer at the time of screening or in the four subsequent years.
The results were revealing.
"In particular, the 2% of women with the highest risk index had nearly a 9.7% probability of receiving a breast cancer diagnosis in the following four years, a figure comparable to or even higher than the risk for some carriers of genetic mutations associated with the disease," the researchers point out.
This precision suggests that AI could help transition from a uniform screening model to a more personalized one, tailored to the risk of each patient.
One of the notable advantages of the BRAIx approach is its ability to extract risk signals directly from the mammographic image, surpassing traditional factors such as breast density, age, or family history.
Thus, women at higher risk could undergo additional tests such as MRI or contrast mammograms, while those at lower risk could space out their check-ups, reducing unnecessary procedures.
The study was also tested on a Swedish population of over 4,500 women, confirming the algorithm's ability to identify patients at high risk even after a negative mammogram result.
"Nearly one in ten of those who scored in the top 2% according to the algorithm were diagnosed within four years, despite having been discharged from medical care," reports Science Media Centre.
The authors of the study warn that further prospective trials and clinical evaluations are still needed to determine how to integrate this technology into existing screening programs and assess its real impact on early diagnoses and survival.
Breast cancer is the most common type of cancer among women, with approximately 2.3 million cases diagnosed annually worldwide, according to the World Health Organization.
In Australia, more than 20,000 women receive this diagnosis each year, and about 3,300 die from the disease, highlighting the importance of improving detection and prevention tools.
Filed under: