AI-Assisted Mammography Identifies Cardiovascular Risk: Key Findings
AI-Assisted Mammography Identifies Cardiovascular Risk: Key Findings
Dr. Dapamede will present these findings at ACC.25 on Monday, March 31, at 9 a.m
Key Points:
• AI in Mammography for Heart Health: New research presented at ACC.25 explores how artificial intelligence (AI)-assisted mammography can predict a woman’s cardiovascular disease (CVD) risk.
• Detecting Breast Artery Calcifications (BAC): While BAC is visible on mammograms, it is often overlooked in cardiovascular assessments. Researchers trained an AI model to analyze BAC and generate a cardiovascular risk score.
• Data & Model Training: The AI system was trained using mammography images and electronic health records from over 56,000 patients treated at Emory Healthcare (2013–2020), with at least five years of follow-up data.
• Performance & Impact: The AI model successfully classified patients’ cardiovascular risk as low, moderate, or severe based on routine mammograms. It proved particularly beneficial for younger women (<60 years old), allowing for earlier intervention and cardiology referrals.
• Major Finding: Patients with severe BAC faced a 2.8 times higher five-year mortality risk compared to those without concerning BAC signs.
• Future Implications: Lead author Dr. Theo Dapamede, MD, PhD, emphasized that deep learning and AI enable opportunistic screening, offering women both cancer and cardiovascular risk assessments from a single mammogram.
