Artificial Intelligence Meets Real-Life Dermatology: Diagnostic Accuracy Assessment in a Retrospective Case Series

Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown

Published: April 28, 2025

Background: Although artificial intelligence (AI) has shown considerable promise in dermatological diagnostics, its real-world clinical validation remains limited. This study aimed to evaluate the diagnostic accuracy and decision-support capabilities of GPT-4.5 a routine outpatient dermatology setting. Methods: A total 402 dermatologic cases from 400 patients were retrospectively analyzed at secondary-care clinic. was provided with dermoscopic images, along brief metadata (e.g., age, lesion location, duration), generate differential diagnoses management suggestions. Model outputs compared dermatologist assessments. Performance metrics included accuracy, sensitivity, specificity, precision, F1 score. Misclassification patterns also reviewed. Results: achieved an overall 89.3% correctly identified primary diagnosis as top-ranked suggestion 71.9% cases. Sensitivity specificity 89.7% 91.4%, respectively, score 94.3%. Clinical guidance recommendations concordant physician decisions 91.0% Diagnostic higher non-biopsied (96.0%) those requiring histopathological confirmation (84.2%). Highest performance observed infectious (94.3%) inflammatory (96.2%) dermatoses. Misclassifications most common pigmented neoplasms morphologically similar disorders. Conclusion: demonstrated high strong alignment dermatology, especially for visually distinct conditions. However, declined diagnostically complex or ambiguous These findings support potential supplementary tool, while underscoring need multimodal inputs, oversight, broader prospective prior integration.

Language: Английский

The Role of Artificial Intelligence in Diagnosis and Management of Cutaneous Infections

Current Dermatology Reports, Journal Year: 2025, Volume and Issue: 14(1)

Published: May 14, 2025

Language: Английский