Survey questions accuracy of skin cancer apps

Survey questions accuracy of skin cancer apps

18 JAN 2013

A survey by the University of Pittsburgh Medical Center has expressed concern about a number of smartphone apps that give an assessment of whether a skin lesion is benign or malignant.

The survey is concerned because patients may substitute such analysis, whose accuracy it questions, for visiting a dermatologist. The study was published online by Jama Dermatology.

Three of the four apps assessed by the center had no physician involvement but even the best-performing of them managed to classify 18 of 60 known melanomas as benign.

Overall, one of the four apps (easily the best performing one) successfully identified melanomas 98 percent of the time. The worst of the apps only managed a success rate of 6.8 percent.

The two remaining apps scored around 70 percent which is inferior to the rate a trained dermatologist would expect to assess (about 90 percent) but compares favorably to the pick-up rate of a non-specialist doctor.

The identity of the four apps was not revealed in the survey.

The three less successful apps used algorithms to analyse a patient’s moles, with no input from dermatologists. They all cost $5 or less. Meanwhile the most successful app cost $5 per mole for analysis, so clearly a more expensive option although a more accurate one. This app was based on forwarding images for dermatologists to take a judgement.

The study was funded by the National Institutes of Health.


Richard Handford

Richard is the editor of Mobile World Live’s money channel and a contributor to the daily news service. He is an experienced technology and business journalist who previously worked as a freelancer for many publications over the last decade including...

Read more