Journal: Nature communications
This study presents MelanoMAP, a multimodal AI model that integrates tumor microenvironment (TME)-derived digital biomarkers with clinicopathological features to improve prognostication in localized cutaneous melanoma (CM).
By analyzing over 3,500 histology slides, MelanoMAP significantly outperformed traditional AJCC staging systems, achieving a C-index of 0.82 compared to 0.66.
The results were consistent across six international cohorts.
Key predictors included:
- TME biomarkers
- Established factors such as age, mitotic count, and Breslow depth
This model highlights the potential of AI-driven digital biomarkers to enhance personalized risk assessment and guide clinical decision-making in CM.