AI Model Improves 5-Year Breast Cancer Risk Prediction

By: Women Entrepreneurs Review Team | Thursday, 10 September 2026

Researchers at NYU Langone Health and Perlmutter Cancer Center have developed an artificial intelligence model that can more accurately predict a woman’s five-year risk of developing breast cancer by analyzing mammograms taken over several years. The deep-learning tool, known as NYU-DRP, uses longitudinal 3D mammograms, also called digital breast tomosynthesis, to assess changes over time.

A new study found that NYU-DRP performed better than models that rely on only the latest 3D mammogram or analyze 2D mammograms. The AI model correctly identified women at higher risk of developing breast cancer within five years 72 percent of the time, compared with 70 percent for single 3D mammograms and 68 percent for AI-assisted 2D mammograms.

Published online in the American Journal of Roentgenology on August 12, the findings suggest that analyzing a woman’s mammograms across multiple years could improve breast cancer risk assessment. Researchers say the approach could help identify women at greater risk and support more personalized screening strategies.

Key Highlights:

• NYU-DRP predicts five-year breast cancer risk using 3D mammograms
• The AI model achieved 72% accuracy in identifying higher-risk women
• The tool could support more personalized breast cancer screening

"Our study shows how AI models like NYU-DRP can be used to reliably determine a woman's future risk of breast cancer based on existing 3D mammograms, which hold information on how the breast tissue has changed across multiple screenings over time," said study lead investigator Yanqi Xu, PhD, a postdoctoral research fellow in the Department of Radiology at NYU Grossman School of Medicine and Perlmutter Cancer Center.

NYU-DRP was created from 313,531 yearly 3D mammograms from 161,165 women without breast cancer who had tests performed at NYU Langone hospitals between 2016 and 2020.

When researchers compared NYU-DRP against a common breast cancer risk assessment tool, called the Tyrer-Cuzick risk assessment, NYU-DRP proved to be even more effective, correctly predicting those who would be at greater risk after five years 67 percent of the time, compared to the Tyrer-Cuzick tool, which did so 56 percent of the time.

Tyrer-Cuzick does not make use of AI or mammograph scans – it uses personal and family medical data, including age, genetic mutations, breast density and more, supported by biopsies. The researchers compared the results of 432 women, half of whom were carefully matched to women with similar backgrounds and ages who did or did not develop breast cancer after 5 years to determine which method performed better, NYU-DRP or Tyrer-Cuzick.

The researchers noted that less than 3 per cent of women who were tested during the study, which ended in 2025, developed breast cancer.

One of the study's other conclusions was that the breast density alone was not linked to a woman's estimated risk. A dense breast is a risk factor for cancer. For women who had extremely dense breasts, the NYU-DRP model identified a 37.6 percent risk of average breast cancer risk, whereas real-world cases after five years had a rate of 0.7 percent. In contrast, the NYU-DRP model classified 15.5 percent of women with low breast density and high fat content as high risk, while the actual 2.5 percent were found after five years.

"Our findings demonstrate that repeated 3D mammograms contain information about a woman's future breast cancer risk that is not fully captured by either breast density or a single mammogram on its own," said study senior investigator Yiqiu "Artie" Shen, PhD, an assistant professor in the Department of Radiology at NYU Grossman School of Medicine and Perlmutter Cancer Center.

"If future experiments in other women with breast cancer prove successful, then AI-assisted 3D mammograms like NYU-DRP could help physicians better tailor screening to a woman's actual risk, by identifying those women who may benefit from additional screening while avoiding unnecessary supplemental tests for those at lower risk," said study co-investigator Laura Heacock, MD, an associate professor in the Department of Radiology at NYU Grossman School of Medicine and Perlmutter Cancer Center.

The team next wants to use the longitudinal DBT program to monitor women's health proactively, and see how the program affects their health, and who gets breast cancer, Dr. Shen said. They also will share and compare the tool with information from other academic health centers and data from various manufacturers of 3-D mammograms. Only breast imaging equipment by Hologic Inc. Marlborough, MA was used for all the testing conducted throughout the current study.

An estimated 1 of every 8 women in the United States will get breast cancer at some point in their lives, of which about 382,640 women will be diagnosed in 2026. When breast cancer is detected in its earliest stages, however, the relative survival rate for 5 years is more than 99 percent. With new advances in detection and treatment, many people are living with breast cancer today, with more than 4 million survivors living in the United States.