“Adding Passion to PASSION.derm”: Detecting Skin Problems in Higher Phototypes

Since 2018 the PASSION.derm project has being ongoing and in the last year, our lives are tossed upside down. However the objective of recognizing dermatological diseases in skin of color -and safely integrating solutions in health using artificial intelligence remains unchanged.

In many African countries, collaborations have been built so as to collect five skin conditions which make up 80% of skin problems in the pediatric population (these are eczema (atopic dermatitis) and infections (fungal*-bacterial**-infestation***). [The current medical team is available on telederm.ai].

Picture collection room in Antananarivo, Madagascar

Now that we have the data, we need to train it and AI solutions with a proven experience on recognizing skin of color are invaluable. We have teamed up with such a technical team, learning from their experience in India to recognize 40 skin diseases with a model, then deploying it. The results were published recently (February 2020) in JEADV (https://doi.org/10.1111/jdv.16967):

A machine learning‐based, decision support, mobile phone application for diagnosis of common dermatological diseases.

R. Pangti, J. Mathur, V. Chouhan, S. Kumar, L. Rajput, S. Shah, A. Gupta, A. Dixit, D. Dholakia, S. Gupta, S. Gupta, M. George, V.K. Sharma,S. Gupta

Accuracy scores for recognition of skin disorders on Indian skin (reproduced with permission)

We are hopeful that clinical and technical partners will be achieving the goals of the PASSION project, as well as increasing the sensitivity scores of the model (Specificity scores are high).

*fungal: the tinea spectrum




Scroll to Top