Globally, more than 600,000 women do not survive breast cancer each year.
Health organizations worldwide support screening mammography to achieve early cancer detection, and it has proven its value, reducing mortality by 20-40%.
Mammography is the best tool we have for early detection, but challenges of false positives, false negatives, human variation in image interpretation and lack of access to specialized radiologists show opportunities to improve.
At Clairity, we believe that opportunity lies in the application of precision medicine to our current screening paradigm.
While “big data” has revolutionized healthcare in many ways, the mammography screening paradigm has not changed since the 1960s. Our current one-size-fits-all approach is generic, based primarily on gender (female) and age.
Many intelligent and well-respected industry thought leaders and national organizations have sought opportunities for improvement but, to date, there is little agreement on the best approach. As a result, conflicting recommendations about when to begin screening, how often to be screened, eligibility, and risk confuse patients. This negatively impacts compliance and ultimately, patient outcomes.
In the 1980s, risk models combining a number of factors loosely correlated with the development of breast cancer were developed to identify women who might need supplemental screening. Although these models are only moderately accurate, they are incorporated into NCCN guidelines, used by clinicians as part of shared decision making, and serve as thresholds for eligibility.
Unfortunately, recent research comparing the output of these traditional risk scores shows wide variability across models and, more importantly, significant racial and ethnic bias. In addition, patients with a prior history of breast cancer, unknown family history, and those who have had a single mastectomy are not eligible for risk assessment.
The introduction of artificial intelligence (AI) and deep learning neural networks offer significant promise in further advancing the mammography screening process. The ability to process large volumes of data and analyze visual patterns at the pixel level provides radiologists with new and more precise insights into which individuals could be at risk of developing breast cancer.
Many of the significant advancements made in the domain of personalized oncology are linked to advancements in the domains of genomics and proteomics – using specific genetic or protein data to identify and predict which cancer patients have tumors that will respond to treatment with a specific drug. The domain of radiomics is similar, but in this case a high number of quantitative features are extracted from medical images. Artificial intelligence and deep learning models are particularly effective at evaluating these massive amounts of data, learning from patterns that emerge and making predictions.
Built on decades of research in the domain of radiomics and applied AI, Clairity’s rigorous scientific approach to breast health will improve the accuracy of the risk assessment process, allowing clinicians to fulfill the promise of precision medicine and tailor a healthcare experience that is unique to each patient.
Deep learning breast cancer risk models demonstrate improved accuracy compared to traditional risk models. This research highlights the promise deep learning models to support equitable, risk-based mammography screening.
Accurate density assessments could help health care systems more appropriately use limited supplemental screening resources and help better inform traditional clinical risk models.
The COVID-19 pandemic took a toll on mammography screening volumes among women of color, slashing exam uptake by more than 50% between 2019 and 2020.
