Dermoscopy-Based AI Risk Scoring Enhanced Experienced Dermatologists’ Decision-Making: a Large Retrospective Reader Study
Year: 2026
Type: article
Abstract: INTRODUCTION: The integration of artificial intelligence (AI) in dermoscopy is promising for diagnostic and management decisions. OBJECTIVES: This study aimed to assess the impact of a deep learning-generated AI risk score, based on dermoscopic images and metadata, on dermatologists' diagnostic accuracy, confidence, and management strategy for lesions suspicious of skin cancer. METHODS: A neural network was developed using a database of dermoscopic images of benign and malignant skin lesions, including both proprietary and public data. In a multicenter, cross-sectional study, 104 experienced dermatologists, with a median dermoscopic experience of 10 years, evaluated batches from a test set of 922 skin lesions, including 577 benign (63%) and 345 malignant (37%) cases, resulting in a dataset comprising 9,198 observations. Each case was assessed before and after the AI-generated risk score. Key outcomes included accurate diagnosis, correct decision on malignancy, and management strategy. RESULTS: Overall accuracy for correct diagnosis increased from 74.6% before AI to 81.6% after AI (95% confidence interval (CI): 3.8%-10.2%; P<0.001). Overall sensitivity for decision on malignancy increased from 91.7% before AI, to 97.6% after AI (95% CI: 2.9%-8.8%; P<0.001). The biggest increase in sensitivity was seen for melanoma and high-grade dysplastic naevi, from 79.0% before AI to 93.8% after AI (95% CI: 5.4%-24.4%; P=0.002). AI was significantly associated with choosing the correct management strategy (53.5% to 59.0%, 95% CI: 2.5%-8.5%; P<0.001). CONCLUSIONS: AI enhanced experienced dermatologists' decision-making, particularly for melanoma and high-grade dysplastic naevi. Future prospective studies should further explore AI's integration into routine clinical settings.
Institutions: KU Leuven, Barco (Belgium), Ghent University Hospital
Language: English
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W7171964573