A PASSION to provide dermatological care in LMICs

Dermatology offers many opportunities for the implementation of artificial intelligence (AI) due to its large clinical and image database. The PASSION project was designed to help diagnose skin conditions in countries with a lack of dermatology care.

Chronic skin conditions affect millions of individuals around the world. Unfortunately, dermatologists are underrepresented in low-income and middle-income countries (LMICs), which makes access to specialist care challenging. Dr Christophe Hsu (University Hospital of Basel, Switzerland) shared the objective of the PASSION project, which is to design and implement algorithms that can help diagnose skin conditions, including atopic dermatitis, impetigo, tinea, scabies, and insect bites, in the paediatric population [1]. These 5 treatable disorders account for almost 80% of skin conditions in children. The PASSION project is a non-profit initiative that combines clinical trials with AI in Madagascar and Guinea. The study will soon be extended to other countries such as Mali, Mauritania, Mexico, and Indonesia.

Clinical data of patients together with high-quality images of the skin conditions are uploaded onto a digital platform. Through machine learning, a Convoluted Neural Network algorithm is trained to recognise specific skin conditions. The algorithm will improve automatically through experience. In a testing group, diagnosis is made by both AI and a dermatologist and compared with each other.

While no results of the PASSION project itself are available yet, early results from a European dataset showed that atopic dermatitis can be diagnosed through AI with a precision of 93% and a recall of 91%, based on 20,000 images for training and 4,000 for testing.

Overall, AI could contribute to the diagnosis of paediatric patients with chronic skin conditions in areas where access to dermatology care is limited.

1. 
   
   1. Hsu C, et al. Identification of common skin disorders on dark skin using artificial intelligence. Abstract 69, ICD 2021, 10–13 November.

 

Copyright ©2021 Medicom Medical Publishers



Posted on 2 December 202128 March 2024