Feasibility study of an AI-powered mobile app to support cutaneous leishmaniasis diagnosis in the Brazilian Amazon

Author summary Cutaneous leishmaniasis is a major public health challenge in the Brazilian Amazon. Delays in diagnosis increase illness, stigma, and costs, especially in remote communities with limited connectivity. Community health workers and primary care teams are often the first to see suspicious skin lesions, but they lack point-of-care tools to support early triage. We developed and evaluated the feasibility of a mobile application that works entirely offline and uses artificial intelligence to assist with early triage from smartphone photographs of skin lesions. The model was trained on clinical images collected by partner institutions in the Amazon region and integrated into an app optimized for low-resource settings. In a multicenter assessment that included both more standardized (“ideal-world”) images and routine field conditions, the tool identified most cutaneous leishmaniasis cases, returned a result in under a minute, and was well accepted by users. Specificity was moderate, which is appropriate for a triage tool designed to prompt timely referral rather than to replace clinical and laboratory diagnosis. This feasibility study shows that an offline, AI-assisted approach can help accelerate referral and access to care in underserved regions. Next steps include larger, prospective clinical validation, usability refinements based on field feedback, and regulatory evaluation aligned with Brazilian requirements (ANVISA). The same strategy could be adapted to other dermatologic conditions and neglected tropical diseases in similar settings.