TY - JOUR
T1 - Dermatologist-like explainable AI enhances melanoma diagnosis accuracy
T2 - eye-tracking study
AU - Chanda, Tirtha
AU - Haggenmueller, Sarah
AU - Bucher, Tabea Clara
AU - Holland-Letz, Tim
AU - Kittler, Harald
AU - Tschandl, Philipp
AU - Heppt, Markus V.
AU - Berking, Carola
AU - Utikal, Jochen S.
AU - Schilling, Bastian
AU - Buerger, Claudia
AU - Navarrete-Dechent, Cristian
AU - Goebeler, Matthias
AU - Kather, Jakob Nikolas
AU - Schneider, Carolin V.
AU - Durani, Benjamin
AU - Durani, Hendrike
AU - Jansen, Martin
AU - Wacker, Juliane
AU - Wacker, Joerg
AU - Brinker, Titus J
AU - Reader Study Consortium
A2 - Bondare-Ansberga, Vanda
A2 - Balcere, Alise
N1 - Publisher Copyright:
© The Author(s) 2025.
PY - 2025/12
Y1 - 2025/12
N2 - Artificial intelligence (AI) systems substantially improve dermatologists’ diagnostic accuracy for melanoma, with explainable AI (XAI) systems further enhancing their confidence and trust in AI-driven decisions. Despite these advancements, there remains a critical need for objective evaluation of how dermatologists engage with both AI and XAI tools. In this study, 76 dermatologists participate in a reader study, diagnosing 16 dermoscopic images of melanomas and nevi using an XAI system that provides detailed, domain-specific explanations, while eye-tracking technology assesses their interactions. Diagnostic performance is compared with that of a standard AI system lacking explanatory features. Here we show that XAI significantly improves dermatologists’ diagnostic balanced accuracy by 2.8 percentage points compared to standard AI. Moreover, diagnostic disagreements with AI/XAI systems and complex lesions are associated with elevated cognitive load, as evidenced by increased ocular fixations. These insights have significant implications for the design of AI/XAI tools for visual tasks in dermatology and the broader development of XAI in medical diagnostics.
AB - Artificial intelligence (AI) systems substantially improve dermatologists’ diagnostic accuracy for melanoma, with explainable AI (XAI) systems further enhancing their confidence and trust in AI-driven decisions. Despite these advancements, there remains a critical need for objective evaluation of how dermatologists engage with both AI and XAI tools. In this study, 76 dermatologists participate in a reader study, diagnosing 16 dermoscopic images of melanomas and nevi using an XAI system that provides detailed, domain-specific explanations, while eye-tracking technology assesses their interactions. Diagnostic performance is compared with that of a standard AI system lacking explanatory features. Here we show that XAI significantly improves dermatologists’ diagnostic balanced accuracy by 2.8 percentage points compared to standard AI. Moreover, diagnostic disagreements with AI/XAI systems and complex lesions are associated with elevated cognitive load, as evidenced by increased ocular fixations. These insights have significant implications for the design of AI/XAI tools for visual tasks in dermatology and the broader development of XAI in medical diagnostics.
UR - http://www.scopus.com/inward/record.url?scp=105005979720&partnerID=8YFLogxK
U2 - 10.1038/s41467-025-59532-5
DO - 10.1038/s41467-025-59532-5
M3 - Article
C2 - 40399272
AN - SCOPUS:105005979720
SN - 2041-1723
VL - 16
JO - Nature Communications
JF - Nature Communications
IS - 1
M1 - 4739
ER -