AI-Based Prognostic Assessment Shows Promising Results in Early-Stage Melanoma

A recent study published in the Journal of the European Academy of Dermatology and Venereology (JEADV) presents a comparative evaluation of an artificial intelligence (AI)-based prognostic assay, DiaSurv™ Melanoma, and a gene expression profiling (GEP) test in early-stage melanoma.

Accurate risk stratification in stage I–II melanoma remains a significant clinical challenge due to the heterogeneity of patient outcomes. This study investigates whether AI-driven analysis of standard histopathology slides can provide clinically meaningful prognostic insights.

Study Overview
The analysis was conducted on a retrospective cohort of 50 patients. The AI-based model, developed using deep learning on whole slide images (WSI), was compared to a commercially available GEP assay.

Key Findings

• 74% overall concordance between AI and GEP in risk classification
• All relapse events were assigned to the high-risk group by the AI-based assay
• A majority of deaths occurred in patients classified as high-risk by the AI model
• Statistically significant separation in relapse-free survival (RFS) using the AI approach, not observed with GEP

These results indicate that AI-based histopathological analysis may improve prognostic stratification compared to existing molecular approaches.

Clinical Relevance
The DiaSurv™ Melanoma assay is based solely on standard H&E-stained slides and does not require additional molecular testing, offering a rapid and scalable solution for routine clinical workflows.

By improving risk stratification, such tools may support more informed clinical decision-making, including patient follow-up and selection for adjuvant therapies.

About the Study
The study was conducted using data from the French RicMel multicentric database and involved collaboration between pathology, dermatology, and AI research teams.