Breast Cancer Risk Management – Past, Present, and Future
Description
Breast cancer prevention strategies rely upon effective tools to identify "at risk" women who are more likely to develop breast cancer. Risk-stratified screening has been proposed to optimize screening for patients and better use resources, which is equally important to the entire healthcare system and individual patient’s benefit. Improved breast cancer risk models enable targeted screening strategies that achieve earlier detection and less screening harm than existing guidelines. There are several available models that can help estimate a woman's risk of developing breast cancer, including Breast Cancer Risk Assessment Tool (BCRAT)/Gail model, Tyrer-Cuzick model, Claus model, and the Breast Cancer Surveillance Consortium (BCSC) model. These models incorporate only a small fraction of patient data available gleaned from questionnaires, including family history of breast cancer, hormonal and reproductive history. However, despite decades of effort, the discriminatory ability of breast cancer risk models used in clinical practice remains modest. With the development of artificial intelligence (AI) and improved accessibility of large-scaled screening data, deep-learning-based risk models have been developed and validated around the globe. With a free open-source model, Mirai, available for public use, one company (Clairity) has gained FDA approval and one company (Prognosia) has gained FDA Breakthrough Device designation. This session will provide a comprehensive review of traditional risk models, and the new development of AI based risk models. With invited physician speaker(s), the development, validation, implementation, and clinical impact of AI based risk models will be introduced in depth.