medwireNews: Smartphone applications that claim to be able to diagnose melanoma should be treated with extreme caution, say researchers.
Laura Ferris (University of Pittsburgh Medical Center, Pennsylvania, USA) and colleagues found that three out of four applications they assessed classified 30% or more of melanomas as being "unconcerning."
The team adds that these applications are not currently subject to official regulation and that they could easily be misused by the general public.
"These tools may help patients be more mindful about their health care and improve communication between themselves and their physicians, but it's important that users don't allow their 'apps' to take the place of medical advice and physician diagnosis," said Ferris in a press statement.
Ferris and co-authors of the research, published in JAMA Dermatology, evaluated the accuracy of four smartphone applications designed to help nonclinicians assess whether skin lesions are benign or malignant using verified medical images of 60 melanoma and 128 benign control lesions.
The accuracy of the four applications varied considerably, with sensitivities ranging from 6.8% to 98.1%, specificities from 30.4% to 93.7%, positive predictive values from 33.3% to 42.1%, and negative predictive values from 65.4% to 97.0%.
Ferris and colleagues explain that the primary endpoint was sensitivity to melanoma characterization because most of the lesions that they remove in their practice are removed because of concerns about malignancy, so therefore they expected the specificity to be low.
Based on sensitivity alone, the most accurate app had a sensitivity of 98.1% and the least 6.8%. The most accurate app differed from the others in that it could be run on a smartphone or a website and each image was sent to a board-certified dermatologist for assessment over 24 hours, whereas the other three were designed specifically for smartphones and used algorithms to assess images instantly rather than relying on expert analysis.
Discussing the results with the press, Ferris said that the lack of accuracy of some of the applications is a concern. "If [patients] see a concerning lesion but the smartphone app incorrectly judges it to be benign, they may not follow up with a physician."
She added: "Technologies that decrease the mortality rate by improving self- and early detection of melanomas would be a welcome addition to dermatology. But we have to make sure patients aren't being harmed by tools that deliver inaccurate results."