In the Journals

App-based teledermatology may improve early diagnosis of melanoma

Machine-learning systems could improve early melanoma diagnosis based on the implementation of a teledermatology program in Bergamo, Italy, according to a research letter in Journal of the American Academy of Dermatology.

The free app was advertised in Bergamo from March to July 2017. Participants aged at least 18 years (n = 232) used the app to send pictures of their suspicious lesions for a dermatologic assessment and were invited to undergo a free full-body examination by a different dermatologist.

Online assessment classified 56 lesions as suspicious. After direct clinical examination, 14 (25%) were confirmed as suspicious. One lesion (0.5%) classified as nonsuspicious online was considered suspicious on direct clinical examination, according to the study.

Lesions that were classified as suspicious on clinical examination were assessed histologically. Six lesions (2.6%) confirmed as melanomas were also classified as suspicious online.

Researchers determined the diagnostic accuracy of the suspicious lesions via online teledermatology assessment compared with direct clinical examination was 81% (95% CI, 75.4-85.9), sensitivity was 92.9% (95% CI, 66.1-99.8), specificity was 80.3% (95% CI, 74.4-85.3), positive predictive value was 23.2% (95% CI, 13-36.4) and negative predictive value was 99.4% (95% CI, 96.9-100).

The researchers added that teledermatology may help promote patient empowerment. Those who used the app reported that their main motivators were a suspicious lesion (67.2%) and the ease of access to the system (50.5%). About 70% of participants would not have seen a dermatologist without the program, according to the study.

The sample was biased toward those who were more concerned about their health and also familiar with technology.

“Our system, which involved the use of remote clinicians to judge lesions, was expensive and time-consuming,” Anna Di Landro, MD, from the Study Center of the Italian Group for Epidemiologic Research in Dermatology, and colleagues wrote. “In the near future, machine-learning systems connected with an app could revolutionize melanoma early diagnosis.” – by Abigail Sutton

 

Disclosures: The authors report no relevant financial disclosures.

 

Machine-learning systems could improve early melanoma diagnosis based on the implementation of a teledermatology program in Bergamo, Italy, according to a research letter in Journal of the American Academy of Dermatology.

The free app was advertised in Bergamo from March to July 2017. Participants aged at least 18 years (n = 232) used the app to send pictures of their suspicious lesions for a dermatologic assessment and were invited to undergo a free full-body examination by a different dermatologist.

Online assessment classified 56 lesions as suspicious. After direct clinical examination, 14 (25%) were confirmed as suspicious. One lesion (0.5%) classified as nonsuspicious online was considered suspicious on direct clinical examination, according to the study.

Lesions that were classified as suspicious on clinical examination were assessed histologically. Six lesions (2.6%) confirmed as melanomas were also classified as suspicious online.

Researchers determined the diagnostic accuracy of the suspicious lesions via online teledermatology assessment compared with direct clinical examination was 81% (95% CI, 75.4-85.9), sensitivity was 92.9% (95% CI, 66.1-99.8), specificity was 80.3% (95% CI, 74.4-85.3), positive predictive value was 23.2% (95% CI, 13-36.4) and negative predictive value was 99.4% (95% CI, 96.9-100).

The researchers added that teledermatology may help promote patient empowerment. Those who used the app reported that their main motivators were a suspicious lesion (67.2%) and the ease of access to the system (50.5%). About 70% of participants would not have seen a dermatologist without the program, according to the study.

The sample was biased toward those who were more concerned about their health and also familiar with technology.

“Our system, which involved the use of remote clinicians to judge lesions, was expensive and time-consuming,” Anna Di Landro, MD, from the Study Center of the Italian Group for Epidemiologic Research in Dermatology, and colleagues wrote. “In the near future, machine-learning systems connected with an app could revolutionize melanoma early diagnosis.” – by Abigail Sutton

 

Disclosures: The authors report no relevant financial disclosures.