Can AI-driven teledermatology increase access to healthcare in rural African settings?

Artificial intelligence (AI)-driven teledermatology may offer a solution to the limited access to dermatologic care in some parts of Africa. Since access to smartphones is high and internet connections are generally good, healthcare workers may improve their clinical decision-making skills with respect to skin disease through smartphone-guided teledermatology.

“The access to dermatologic care in [some parts of] Africa is very limited,” emphasised Mr Philippe Gottfrois (Basel University, Switzerland) [1]. “For example, to provide annual coverage for the entire population of Madagascar, dermatologists would have to see 5,200 patients per day.” Mr Gottfrois added that most healthcare workers are not trained in dermatology. “Also, dermatologists are mostly based in larger cities, leaving the rural areas uncovered. Fortunately, 57% of the population has a smartphone and the internet connection is generally good,” he said.

“Our AI-driven teledermatology project uses the high access to smartphones to offer teledermatology, in order to educate medical workers in managing common skin conditions,” said Mr Gottfrois. He explained that 5 conditions account for 80% of the consultations in Africa: dermatophytosis, insect bites, atopic dermatitis, scabies, and impetigo. Therapies are available for these conditions, and swift interventions can heal patients with these diseases efficiently.

The AI-teledermatology platform that is developed by Mr Gottfrois and colleagues begins with data collection by first healthcare providers. They take pictures of the skin condition and collect additional relevant patient information, which are then transferred via WhatsApp to a web-based app. This app will run an AI-driven diagnostic decision-making tool, informing the healthcare provider. He or she then decides whether there is a need to contact a dermatologist, who may provide feedback through telecommunication or whether the provided diagnosis by the app is sufficient.

Thus far, 1,400 cases of common skin diseases have been collected across Madagascar, Guinea, Tanzania, and Malawi, with an equal distribution of Fitzpatrick skin types 3 to 6. Pictures and additional data are added to the web app to create an AI-driven case atlas. The first results indicate that this approach has a sensitivity of 80% and a specificity of 85% with regard to detecting eczema. Further results are expected after the initiation of a clinical study in October 2023.

1. Gottfrois P, et al. Passion project: AI driven teledermatology for common skin conditions in Africa. Session Teledermatology, WCD 2023, 3-8 July, Singapore.