High accuracy seen using ordinary online photographs
By Lori Solomon HealthDay Reporter
WEDNESDAY, Aug. 21, 2024 (HealthDay News) — Artificial intelligence (AI) is able to distinguish Marfan from non-Marfan facial images using ordinary online photographs with an extremely high degree of accuracy, according to a study published in the July 15 issue of Heliyon.
Danny Saksenberg, from the Yale University School of Medicine in New Haven, Connecticut, and colleagues explored the potential of AI in diagnosing Marfan syndrome from ordinary online facial images. The model was trained on 80 percent of 672 facial images (182 Marfan and 490 control), while the other 20 percent of images were used as the test set.
The researchers found that overall accuracy was 98.5 percent (0 percent false positive; 2 percent false negative). For Marfan faces, the F1 score was 97 percent and the F1 score was 99 percent for non-Marfan faces. The area under the receiver operating curve was 100 percent.
“Clinical usefulness of this program is anticipated. However, due to the limited and preliminary nature of this work, this should be viewed as only a pilot study,” the authors write. “As AI continues to revolutionize health care, integrating such AI-based models into the diagnostic workflow could substantially enhance early detection and management of conditions like Marfan syndrome. It may be possible to deploy such models to allow for screening in nonspecialist settings, such as with general practitioners, nurses, schools and workplaces.”
Copyright © 2024 HealthDay. All rights reserved.
