Abstract
Smartphone applications are increasingly used in medicine, including for the detection of skin cancer. In essence, these applications use artificial intelligence to analyse photographs of skin lesions. When reviewing the literature, only one app (SkinVision) has been evaluated in multiple clinical studies, with a pooled sensitivity of 88% (95%-CI: 70-98%) and specificity of 79% (95%-CI: 70-86%). When used in an unselected adult population, this translates to negative and positive predictive values of > 99% and 1.9%, respectively. This means that a high proportion of app users will be falsely warned of possible skin cancer and recommended to seek medical care. We believe that prior to making this technology available for unrestricted use in a community-based setting, further study is warranted to evaluate the ramifications of this application on health care usage, possible benefits (such as time to diagnosis compared with usual care), as well as unwanted side effects, such as medicalization and anxiety.
| Translated title of the contribution | Smartphone apps for self-diagnosis of skin cancer |
|---|---|
| Original language | Dutch |
| Journal | Nederlands Tijdschrift voor Geneeskunde |
| Volume | 166 |
| Publication status | Published - 19 Jan 2022 |