Artificial Intelligence as a Decision Tool for Skin Lesions Triage in Primary Care: A Narrative Review

Background: Skin lesions are one of the common reasons for consultations in primary care (PC). Among them, skin cancers are a key clinical challenge, where early detection is crucial for the patient’s prognosis. Primary care physicians (PCPs) are responsible for the initial classification of skin conditions, determining which cases require referral to a specialist. In recent years, solutions using artificial intelligence (AI) have been gaining growing interest in medicine, offering new possibilities for diagnostic support, including in the field of dermatology. Objectives: The aim of this narrative review is to analyze the literature on whether and how artificial intelligence (AI) can support primary care physicians in the initial classification of skin lesions. Methods: A comprehensive review of the literature regarding the application of artificial intelligence in skin cancer diagnosis was conducted. The analysis covered results from clinical trials and reviews identifying implementation barriers. Key findings: Results from recent multicenter clinical trials demonstrate that Artificial Intelligence algorithms achieve high diagnostic sensitivity for melanoma, as well as robust accuracy for non-melanoma skin cancers. Comparative studies indicate that while AI performance is comparable to that of board-certified dermatologists, it statistically outperforms less experienced physicians. Conclusion: The conclusions from the study indicate that artificial intelligence can support primary care physicians by increasing their diagnostic effectiveness in the triage of skin lesions. As a support tool, it has the potential to reduce the risk of missing cancers while reducing the number of unreasonable referrals to specialists. However, a key condition for introducing these technologies into everyday clinical practice is to resolve identified ethical and technological barriers.