AI can identify women at a high risk of developing breast cancer more accurately than existing prediction models, according to a new study published in Radiology.
The study’s authors emphasized the importance of “reliable risk stratification” when it comes to breast cancer screening, noting that AI could potentially extract key information from mammographic images and help healthcare providers make key care-related decisions. If it is determined after a mammogram that a particular patient is at an increased risk, it may be time to screen that patient with an MRI scan.
The team developed a deep neural network, using it to explore data from more than 2,200 women between the ages of 40 and 74. While 278 of the women included in the study were later diagnosed with breast cancer, the remaining subjects were healthy controls.
Overall, the authors found that the deep learning risk scores provided by their deep neural network achieved a higher area under the ROC curve (AUC) and lower false-negative rate than density-based risk models.
“The deep neural network overall was better than density-based models,” lead author Karin Dembrower, MD, Karolinska Institute in Stockholm, Sweden, said in a prepared statement. “And it did not have the same bias as the density-based model. Its predictive accuracy was not negatively affected by more aggressive cancer subtypes.”
These findings suggest that AI could play a pivotal role in breast cancer screening and risk assessment, helping determine which patients may need to undergo MRI scans immediately for further evaluation.
Dembrower added in the same statement that AI-based risk models can be updated at any time with stronger data, meaning there is always potential for them to become even more useful. In fact, researchers are working to improve the model now—and there are plans “to test the model clinically next year by offering MRI to the women who stand to benefit the most.”
Michael has more than 16 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.