IBM, Kaiser Permanente find combined AI, human expertise boosts mammogram readings

When used in tandem with radiologists' assessments, machine learning algorithms can improve the overall accuracy of breast cancer screenings, finds a study published March 2 in JAMA Network Open.

The study summarized the findings of a competition launched in 2016 by IBM Research, Kaiser Permanente Washington Health Research Institute, University of Washington School of Medicine and Sage Bionetworks. Participants in the Digital Mammography DREAM Challenge were tasked with designing artificial intelligence algorithms to "meet or beat" radiologists' accuracy in reading mammograms.

Researchers from the lead organizations tested all of the submitted algorithms, which were designed and tested using datasets from Kaiser Permanente Washington and the Karolinska Institute in Sweden.

The organizations found that none of the proposed algorithms were able to outperform human radiologists on their own. When combined, however, radiologists achieved higher levels of accuracy in reading mammograms than when unassisted by AI.

"Based on our findings, adding AI to radiologists' interpretation could potentially prevent hundreds of thousands of unnecessary diagnostic workups each year in the United States," Christoph Lee, MD, a professor of radiology at the University of Washington School of Medicine, lead radiologist for the challenge and co-first author of the study, said in a news release, adding, "Robust clinical validation is necessary, however, before any AI algorithm can be adopted broadly."

More articles on AI:
UCSF developing AI to speed prostate cancer diagnosis
There's not enough data to train AI symptom checkers for coronavirus: WSJ report
AMA developing EHR-connected AI tools to reduce physician burnout

© Copyright ASC COMMUNICATIONS 2020. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Featured Webinars

Featured Whitepapers