Precision revealed: diagnostic performance by consumer artificial intelligence powered app in recognizing skin conditions in primary care settings

Background In recent years, the use of generative artificial intelligence (AI) in health care has been gaining popularity in various medical domains, including dermatology. Recent advances in AI have led to the development of several AI-based diagnostic tools that use machine learning algorithms to rapidly analyze images of skin lesions, interpret visual data, and assist dermatologists in the diagnosis of skin conditions. Objective To investigate the diagnostic accuracy of Google Lens in the diagnosis of dermatological conditions in primary care settings, with a focus on its strengths, limitations, and potential impact on clinical practice. Patients and methods This prospective observational diagnostic accuracy study was conducted on 136 adult patients who were attending their first consultation for skin conditions consultations at family medicine and dermatology outpatient clinics. For every skin condition, the diagnoses that was predicted by the app was compared with the clinical diagnosis provided by the dermatologist and the family physician. Results Most participants were female (61.7%) and presented with localized lesions (80.9%). The most frequently photographed skin regions were the hand, back, and face. Compared with dermatologist’s diagnosis, Google Lens achieved a diagnostic accuracy of 53.7% (95% CI: 45.2; 62.2) when considering the first option provided and 71.3% (95% CI: 66.6; 79.1) when considering the first three options. Conclusion Google Lens application has demonstrated promising results for diagnosing skin conditions that can support institutional missions of medical knowledge and practice improvement.