AI Breast Cancer Risk Models Guide Triage and Nurse-Led Patient Counseling

Artificial intelligence (AI) applications in breast imaging are evolving beyond computer-aided detection of active lesions toward predicting long-term, individualized cancer risk. Speaking in an interview for Oncology Nursing News, Constance “Connie” Lehman, MD, PhD, Professor of Radiology at Harvard Medical School and Founder and CEO of Clairity, outlined the critical distinction between diagnostic AI tools and longitudinal risk projection models.

Diagnostic AI assists radiologists in interpreting screening mammograms to categorize findings as positive or negative for suspicious lesions. Conversely, risk-assessment AI evaluates mammographic tissue patterns to generate a percentile score along a continuous risk spectrum. This score places patients into risk categories—average, intermediate, elevated, or high—aligned with established national screening guidelines.

Oncology nurses and nurse practitioners serve as key educators in translating these algorithmic outputs into understandable, actionable care plans for patients.

Key Takeaways for Oncology Nurses

• Differentiate AI Modalities: Educate patients that risk-evaluation AI does not diagnose current breast cancer; it calculates future risk based on mammographic features.
• Frame Risk Scores Accurately: Clarify that a low or average risk score (such as 1.5%) does not indicate zero risk. Patients in this category must maintain standard screening mammography every 1 to 2 years.
• Provide Reassurance for High Scores: Explain that an elevated 5-year risk score (such as 5%) is not a definitive cancer diagnosis. Approximately 95% of women demonstrating high-risk tissue patterns do not develop breast cancer in the subsequent 5-year window.
• Perform Holistic Evaluation: Combine AI risk outputs with patient family history, breast density assessment, and lifestyle factor analysis to identify risk-reduction strategies.
• Implement Dynamic Re-Testing: Understand that unlike static germline genetic testing (e.g., BRCA1/2 mutation testing), which remains constant over a patient’s lifetime, AI breast risk assessment is dynamic.

Mammographic risk models evaluate changing tissue characteristics over time and should be repeated across routine screening cycles in accordance with National Comprehensive Cancer Network (NCCN) recommendations.

Communication clarity remains essential when presenting continuous risk models. Moving from binary outcomes to risk percentiles empowers clinicians to triage patients toward appropriate supplemental screening or chemoprevention counseling without inducing unnecessary patient anxiety.