Cost-effectiveness of an AI-based app compared to usual care for early skin cancer detection in Belgium

This study evaluated the cost-effectiveness of SkinVision, an AI-based app for early skin cancer detection in Belgium compared to usual care. Decision-tree models were conceptualized to simulate the diagnostic pathways for melanoma, basal cell carcinoma (BCC), squamous cell carcinoma (SCC) and benign skin lesions. Analyses were performed from a healthcare sector perspective, with a time horizon limited to a confirmed diagnosis. For melanoma, the detection rate was lower with the app compared to usual care; (0.601 vs. 0.874) but so were the costs (€419 vs. €536). For BCC, the app detection rate was lower; 0.585 vs. 0.912 and cost difference was minimal (€143 vs. €145). The app detection rate for SCC was 0.706 vs. 0.922 and costs were €228 vs. €233. For benign skin lesions, the app incurred lower costs: €60 vs. €90 and a reduced detection rate of 0.820 vs. 0.937. In its current diagnostic performance the AI-based app was associated with lower skin cancer detection rates and as a consequence, lower costs. The app also led to unnecessary referrals for benign lesions, impacting the healthcare system. Within the Belgian healthcare setting, the study findings do not support implementation or reimbursement of the app for skin cancer detection.