Smartphone Apps Are Not Doctors

Melanoma was not accurately diagnosed by smartphone application

(RxWiki News) There’s an app for just about everything these days. But when it comes to diagnosing any type of cancer, nothing beats a real doctor. Seriously, go to a real doctor.

A recent study tested four smartphone applications designed to diagnose skin cancer with a photograph. The results of the study showed that smartphone applications would allow nearly one-third of dangerous skin cancers to go untreated.

"See a dermatologist for any skin lesions."

Laura K. Ferris, MD, PhD, professor in the Department of Dermatology at the University of Pittsburg Medical Center in Pennsylvania, worked with a team of colleagues and students to investigate the use of smartphone applications in diagnosing skin cancer.

There are applications on the smartphone market that advertise diagnostic capabilities for skin cancer. The applications allow users to photograph skin lesions, and then provide information on the likelihood that the skin lesion is cancerous.

This study focused on the applications' ability to diagnose malignant melanoma, the most dangerous form of skin cancer. Melanoma begins as a skin lesion on top of the skin that can be easily removed. But, if the melanoma is not removed before it spreads beneath the skin’s surface, tumors can spread at a deadly rate.

For the study, 60 melanoma lesions and 128 non-cancerous lesions were photographed for diagnosis by four different smartphone applications before being removed by a dermatologist.

Researchers looked at the diagnostic accuracy of the applications in four areas: sensitivity, specificity, positive predictive value and negative predictive value.

  • Sensitivity ranged between 7-98 percent
  • Specificity ranged between 30-94 percent
  • Positive predictive value (correct diagnosis of the presence of melanoma) ranged between 33-42 percent
  • Negative predictive value (correct diagnosis of the absence of melanoma) ranged between 65-97 percent

One of the applications sent the photographs of lesions to a board-certified dermatologist and this application had the highest rate of sensitivity.

The application with the lowest rate of sensitivity used an automated mathematical algorithm to analyze the images of the lesions.

The authors concluded, “The performance of smartphone applications in assessing melanoma risk is highly variable, and three of four smartphone applications incorrectly classifies 30 percent or more of melanomas as unconcerning. Reliance on these applications, which are not subject to regulatory oversight, in lieu of medical consultation can delay the diagnosis of melanoma and harm users.”

The study showed that smartphone applications may not be qualified to diagnose medical conditions.

The study was published in January in JAMA Dermatology. Funding came from The National Institutes of Health. Dr. Ferris reported working as an investigator and consultant for MELA Sciences, Inc., but no conflicts of interest were reported.

Review Date: 
January 15, 2013