Wired |Business|01.27.2021 02:00 PM “These Doctors Are Using AI to Screen for Breast Cancer”. “During the pandemic, thousands of women have skipped scans and check-ups. So physicians tapped an algorithm to predict those at the highest risk.”
image from HopkinsMedicine.org
Read the article for more detail and see the academic paper in Science Translational Medicine 27JAN2021 A. Yala et al.
According to Constance Lehman (Massachusetts General Hospital) with the onset of the pandemic “around 20,000 women have skipped routine [mammogram] screening” and typically “five of every 1,000 women screened show signs of cancer” or “’…100 [breast] cancers that we haven’t diagnosed.” Further she comments, on work done with MIT over that last few years, “the AI approach has helped identify a number of women who, when persuaded to come in for routine screening, turned out to have early signs of cancer.” Those flagged by AI “were three times as likely to develop cancer.” Besides analyzing mammogram images over-time to incease accuracy, the AI approach, dubbed “Mirai” incorporates relevant information regarding cancer risk collected from patients. Of course, it’s not that simple even using different radiology devices can impart data that are not useful or even misleading, so complex behind-the-scenes work is involved in improving each version of the algorithm to ensure earlier detection while limiting false positives. Mirai was better than existing methods, identifying “42 percent of people who went on to develop cancer in five years compared with 23 percent for the best existing model.” Testing any AI model on multiple and diverse data sets is a standard part of validating and Mirai was successful with “patient data from Taiwan and Sweden.” Charles Kahn (Radiology at University of Pennsylvania) believes that “individual patients [will] ideally [be] receiving a clearer picture of their risk as well as a custom screening plan” but worries that it could “lead to biased care.” Lehman “hopes that AI methods she’s testing can benefit people who typically receive less medical attention.”
Figure 1 from ibid from Science Translational Medicine
Comments