Diagnostic accuracy of smartphone and artificial intelligence tools for diabetic foot screening

Introduction: Diabetic foot ulcers precede most diabetes-related lower-limb amputations, and early detection depends on specialist clinical assessment that is not always available. Smartphone-based and artificial intelligence image tools could extend screening, but their diagnostic accuracy remains scattered across heterogeneous studies. Methods: A systematic review; the bibliographic databases consulted were Scopus, Pubmed, Google Scholar, Open Alex and Crossref. Results: From 963 records, 19 studies were included, 8 of smartphone-based tools and 11 of artificial intelligence models. A smartphone artificial intelligence system detected ulcers with 0.9% sensitivity and 0.9% specificity against expert judgement. Deep learning models reported accuracy between 78.3% and 99.9%, sensitivity between 95.2% and 99.9%, and specificity between 94.0% and 99.9%. Patient selection was at high risk of bias in 15 of 19 studies. Conclusions: Smartphone and artificial intelligence tools reach high diagnostic accuracy for diabetic foot screening, but the evidence derives mostly from retrospective image datasets and requires prospective external validation before clinical use.