The University of Pittsburgh studied four smartphone apps that evaluate melanoma by analyzing images submitted by the users.
One app works by having a board-certified dermatologist look at the photos, while the other three apps analyze the images by computer algorithm.
The best-performing of the computer-driven apps missed 30 percent of the melanoma cases, while the worst-performing missed 93 percent. The app that used actual physicians to diagnose the melanomas worked well, correctly identifying more than 98 percent of the submitted images.
The study's lead researcher, Laura Ferris, worries that misdiagnosis by smartphone could harm patients in the long run.
"If they see a concerning lesion but the smartphone app incorrectly judges it to be benign, they might not follow up with a physician," she said. "A three-month delay could be the difference between high survival and low survival."
The study did not identify the apps by name.
"There are more apps out there and more coming out all the time," said Ferris, assistant professor of dermatology at the University of Pittsburgh's School of Medicine. "It's better to make a statement on the fact that they're out there than to say, 'This one is good and this one is bad.' "
The findings were published online Jan. 16 in JAMA Dermatology.
The idea for the study came about when a patient asked Ferris what she thought about these types of smartphone apps. She and other dermatologists started playing around with the apps by submitting test images and eventually decided to conduct a formal study.
They used actual images of both melanomas and benign skin lesions taken from patients being treated at the university's medical school and evaluated only the apps that would allow submission of previously existing photographs.
Some of the apps were free, while others charged a small amount, such as $4.99 for unlimited images. The app that uses physicians to evaluate the images charged $5 per image, she said.
Ferris said she understands the temptation to find a low-cost, quick way to see a dermatologist. "I know I have patients who are pretty sensitive to cost, and it's very appealing to get this official-looking answer back without worrying about having to pay a co-pay," she said.
Dermatologists have some of the longest wait times of any medical specialists -- one study in the Journal of the American Academy of Dermatology reported an average wait time of 38 days -- while the three computer-based apps returned their answers in minutes. The app that used board-certified dermatologists returned answers in about 24 hours.
But time and cost aren't worth the possibility of postponing diagnosis on a fast-moving disease, Ferris said. "With melanoma, it's such a time-sensitive thing," she said. "It grows and gets deeper, but it's curable if you can get it earlier."
Medical apps are largely unregulated by the Food and Drug Administration, Ferris said, with the exception of apps that interact directly with devices that are already regulated. The FDA is currently discussing plans to regulate other devices, a change that Ferris would support.
"The people who made these apps weren't conscientious or thoughtful, with no validation," she said.
Copyright 2013 Scripps Media, Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.
Latest News Stories
The Martin County School Board considers cutting public access to a swimming pool to meet budget requirements.