By Billy Candra 
A groundbreaking study spearheaded by researchers at Washington University School of Medicine in St. Louis has unveiled an innovative artificial intelligence (AI) driven methodology for analyzing mammograms, which dramatically enhances the precision of predicting an individual’s risk of developing breast cancer over the subsequent five years. This novel approach, which integrates up to three years of an individual’s prior mammogram images, has demonstrated its ability to identify high-risk individuals with 2.3 times greater accuracy compared to conventional risk assessment methods. The standard protocols currently in use predominantly rely on questionnaires that evaluate clinical risk factors such as age, race, and family history of breast cancer. This significant advancement, detailed in the December 5th publication of JCO Clinical Cancer Informatics, marks a pivotal moment in the quest for more effective early detection and personalized prevention strategies against breast cancer.
Pioneering AI Leverages Longitudinal Mammogram Data
The new method represents a paradigm shift from current practices, which often overlook the rich, longitudinal data embedded within sequential mammogram images. Instead of treating each mammogram as an isolated snapshot, the Washington University team’s AI algorithm is designed to discern subtle, cumulative changes over time, transforming the predictive power of routine screening. Dr. Graham A. Colditz, MD, DrPH, senior author of the study and associate director of Siteman Cancer Center, based at Barnes-Jewish Hospital and WashU Medicine, and the Niess-Gain Professor of Surgery, emphasized the overarching goal: "We are seeking ways to improve early detection, since that increases the chances of successful treatment. This improved prediction of risk also may help research surrounding prevention, so that we can find better ways for women who fall into the high-risk category to lower their five-year risk of developing breast cancer."
This research builds upon a foundation of previous work led by Dr. Colditz and lead author Dr. Shu (Joy) Jiang, PhD, a statistician, data scientist, and associate professor of surgery in the Division of Public Health Sciences at WashU Medicine. Their prior investigations revealed that historical mammograms contain a wealth of information regarding nascent signs of breast cancer development that are imperceptible to the human eye, even that of a highly trained radiologist. Key among these subtle indicators are changes over time in breast density – a crucial measure reflecting the relative proportions of fibrous versus fatty tissue within the breast.
Beyond the Human Eye: Unlocking Hidden Biomarkers
For this latest study, the research team engineered an advanced AI algorithm capable of identifying these minute, evolving differences across a series of mammograms. This machine-learning tool goes beyond merely assessing breast density, incorporating changes in other intricate patterns within the images, including texture, calcification, and asymmetries within the breast tissue. Dr. Jiang elaborated on the AI’s unique capability: "Our new method is able to detect subtle changes over time in repeated mammogram images that are not visible to the eye, yet these changes hold rich information that can help identify high-risk individuals."
Current risk-reduction options for breast cancer are relatively limited. They can include pharmaceutical interventions like tamoxifen, which, while effective in lowering risk, may carry undesirable side effects. More commonly, women identified at higher risk are advised to undergo more frequent screening examinations or to supplement mammography with additional imaging modalities, such as Magnetic Resonance Imaging (MRI), in an effort to detect cancer at its earliest possible stage. However, a significant gap has persisted in predicting who will develop breast cancer, rather than just when to screen more aggressively.
Dr. Debbie L. Bennett, MD, a co-author of the study, associate professor of radiology, and chief of breast imaging for the Mallinckrodt Institute of Radiology at WashU Medicine, underscored the transformative potential of this research: "Today, we don’t have a way to know who is likely to develop breast cancer in the future based on their mammogram images. What’s so exciting about this research is that it indicates that it is possible to glean this information from current and prior mammograms using this algorithm. The prediction is never going to be perfect, but this study suggests the new algorithm is much better than our current methods."
Rigorous Validation Across Diverse Populations
To develop and validate their sophisticated machine-learning algorithm, the researchers meticulously trained it using a comprehensive dataset of mammograms from over 10,000 women who received breast cancer screenings through Siteman Cancer Center between 2008 and 2012. These individuals were subsequently monitored through 2020, during which period 478 women received a breast cancer diagnosis.
Following the training phase, the team rigorously applied their predictive method to a separate, independent cohort of patients to validate its efficacy. This validation set comprised more than 18,000 women who underwent mammograms at Emory University in the Atlanta area from 2013 to 2020. Over the follow-up period, which also concluded in 2020, 332 women in this cohort were diagnosed with breast cancer.
The results from both the training and validation cohorts were compelling. According to the new prediction model, women categorized into the high-risk group were an astonishing 21 times more likely to receive a breast cancer diagnosis over the subsequent five years compared to those classified in the lowest-risk group. Specifically, within the high-risk category, 53 out of every 1,000 screened women developed breast cancer within five years. In stark contrast, only 2.6 women per 1,000 screened in the low-risk group developed the disease during the same timeframe. This represents a significant improvement over traditional questionnaire-based methods, which, in a comparative analysis, correctly classified only 23 women per 1,000 screened into the high-risk group. This striking discrepancy indicates that the new AI method identified an additional 30 breast cancer cases that the conventional approach would have missed.
Crucially, the mammograms utilized in the study were collected from both academic medical centers and community clinics, underscoring the method’s robust accuracy across varied healthcare settings. Furthermore, a concerted effort was made to build the algorithm with a strong representation of Black women, a demographic group historically underrepresented in the development of breast cancer risk models. The algorithm’s predictive accuracy was consistently maintained across all racial groups. Of the women screened through Siteman, a significant 27% were Black, while at Emory, Black women constituted an even larger proportion, at 42%. This inclusive approach is vital for ensuring health equity and delivering accurate risk assessments to all women.
Transforming Clinical Practice and Patient Care
The implications of this innovative AI-driven risk assessment tool for clinical practice are profound. By providing a far more accurate prediction of an individual’s five-year breast cancer risk, clinicians will be better equipped to personalize screening protocols. Women identified as high-risk could be offered more intensive surveillance, such as earlier or more frequent mammograms, or supplementary imaging like MRI, potentially leading to earlier detection and more favorable treatment outcomes. Conversely, women with a demonstrably low risk might be able to space out their screenings, reducing anxiety and exposure to radiation, while still maintaining effective surveillance.
This personalized approach aligns with the burgeoning field of precision medicine, where medical decisions are tailored to the individual patient based on predictive biomarkers. For patients, this could translate into less uncertainty, more informed decision-making regarding their health, and potentially, access to preventative measures or lifestyle interventions at an earlier stage. The ability to distinguish between high- and low-risk individuals with such clarity offers unprecedented opportunities for targeted prevention strategies.
Addressing Health Disparities Through Inclusive AI
One of the most commendable aspects of this research is its explicit focus on health equity. The deliberate inclusion of robust data from Black women in the algorithm’s development directly addresses a critical shortcoming in many existing medical AI models: the potential for biased outcomes when certain demographic groups are underrepresented in training data. Breast cancer outcomes disproportionately affect Black women, who have a higher mortality rate despite similar incidence rates to white women. An AI tool that performs equally well across racial groups has the potential to help close this gap, ensuring that all women receive equally accurate and effective risk assessments, thereby fostering more equitable healthcare.
The researchers are continuing their work, actively testing the algorithm in women of other diverse racial and ethnic backgrounds, including those of Asian, Southeast Asian, and Native American descent. This ongoing effort is crucial to guarantee that the method maintains its high level of accuracy and applicability for every woman, regardless of her background.
The Road Ahead: Commercialization and Future Research
The potential impact of this technology extends beyond immediate clinical application. It opens new avenues for breast cancer prevention research. By accurately identifying women at high risk, researchers can more effectively enroll participants in clinical trials for new preventative drugs or lifestyle interventions, accelerating the discovery of better ways to mitigate risk. Dr. Colditz’s statement about helping "research surrounding prevention" highlights this long-term vision.
Looking forward, the Washington University researchers are actively collaborating with WashU’s Office of Technology Management. Their goal is to secure patents and licensing agreements for this innovative method, with the ultimate aim of making it widely accessible wherever screening mammograms are performed. Furthermore, Dr. Jiang and Dr. Colditz are exploring the establishment of a start-up company dedicated to bringing this transformative technology to the broader healthcare market. This commercialization effort underscores the researchers’ commitment to translating their scientific breakthrough into tangible benefits for patients worldwide.
The work, titled "Development and validation of a dynamic 5-year breast cancer risk model using repeated mammograms," was published by Jiang S, Bennett DL, Rosner BA, Tamimi RM, and Colditz GA in JCO Clinical Cancer Informatics on December 5, 2024. This significant undertaking was supported by Washington University School of Medicine in St. Louis. It is important to note that both Dr. Jiang and Dr. Colditz currently have patents pending related to this work, specifically concerning the prediction of disease risk using radiomic images, signaling the proprietary nature and novel intellectual property embedded within their methodology. This marks not just a scientific achievement, but a step towards a future where AI plays a more integrated and impactful role in personalized preventative medicine.
