Artificial intelligence (AI) has the potential to transform cancer screening as we know it, but how and at what cost?


Currently, mammograms can correctly identify about 87% of women with breast cancer, according to the nonprofit Susan G. Komen, which notes that mammography sensitivity is higher in women older than 50 and lower in women with dense breasts. Health care providers throughout the United States now offer patients the option of having their mammograms read by both a radiologist and an AI model to increase the chances of catching cancer early and minimizing false positives.


Breast cancer is the second most common cancer among women, after skin cancer. Nearly 300,000 women are diagnosed with breast cancer annually, according to the American Cancer Society. Regular screening can detect breast cancer early, when it is easier to treat, and reduce mortality.


While mammograms have long been standard practice in breast cancer screening, many factors may give rise to inaccurate results. AI models can sometimes “see what we cannot see,” Katerina Dodelzon, MD, a radiologist specializing in breast imaging at NewYork-Presbyterian/Weill Cornell Medical Center, told the Times.


Indeed, these models can review hundreds, if not thousands, of mammograms daily, highlighting suspicious areas that can then be analyzed by a radiologist.


In Sweden, a study of AI mammography found that the AI model improved breast cancer detection by 20%. Of the 80,000 women involved in the trial, the software detected six cases of cancer in every 1,000 women compared with five per 1,000 detected by radiologists.


What’s more, standard mammograms are more likely to miss cancer in younger women and those with dense breasts as well as lead to false positives, which can result in unnecessary treatment.


A 2022 Danish study used a combined AI model and saw reduced false positives as a result. The study’s authors said that using AI to identify breast cancer would not only lead to early cancer detection but reduce the strain on the health care system due to a worldwide shortage of specialist breast radiologists.


Another study emphasized that the many steps required in current state-of-the-art risk models—such as blood work, genetic testing and extensive questionnaires—increase the clinic workload for health care teams. The study showed that using AI in breast cancer screening is not only safe but also saves time. In fact, because cancer risk may be assessed within seconds of screening, AI can cut radiologists’ workload in half.


Although AI may improve screening accuracy, some experts question how well AI tools can work across a diverse patient population, The New York Times reports.


The cost of implementing AI is another area concern. Patients who opt for an AI analysis may spend between $40 and $100 out of pocket because the service has not yet been assigned an insurance billing code, according to the Times. What’s more, AI is not yet adept at distinguishing between surgical scars and tumors.


While some experts note the importance of combining AI models with standard cancer screening, others believe these tools need additional vetting.


To read more about AI, click #Artificial Intelligence. There, you’ll find headlines such as “Artificial Intelligence Analysis of DNA Fragmentomes and Protein Biomarkers Noninvasively Detects Ovarian Cancer,” “Study Provides First Look at Oncologists’ Views on Ethical Implications of AI in Cancer Care” and “How Artificial Intelligence Can Improve Colorectal Cancer Detection.”