Cost-Effectiveness of an AI-Based Smartphone App for the Early Detection of Skin Cancer: A Decision Tree Analysis
Author(s)
Annick Meertens, MSc1, Ruben Paul Willems, MSc, PhD2, Lieve Brochez, Md, PhD, Prof3, Isabelle Hoorens, MD, PhD4, Nick Verhaeghe, MSc, DrPH, PhD5.
1PhD student, Ghent university, Ghent, Belgium, 2Ghent University, Gent, Belgium, 3Ghent University Hospital, Ghent, Belgium, 4Ghent University Hospital, Gent, Belgium, 5Ghent University, Ghent, Belgium.
1PhD student, Ghent university, Ghent, Belgium, 2Ghent University, Gent, Belgium, 3Ghent University Hospital, Ghent, Belgium, 4Ghent University Hospital, Gent, Belgium, 5Ghent University, Ghent, Belgium.
OBJECTIVES: To evaluate the incremental cost-effectiveness ratio (ICER) of an AI-based smartphone app for the early detection of skin cancer in the general population (≥18 years) in Belgium, compared to the standard clinical pathway without app assessment.
METHODS: Three decision-tree models were conceptualized alongside the ARTIS-trial (NCT05246163) to simulate the diagnostic process for the three most common types of skin cancer: melanoma, basal cell carcinoma (BCC) and squamous cell carcinoma (SCC). The trial included 1458 patients with worrisome skin lesions at the Department of Dermatology of Ghent University Hospital. The analyses were performed from a healthcare sector perspective. One-way and probabilistic sensitivity analysis (PSA) were used to account for parameter uncertainty.
RESULTS: The smartphone app was less effective in detecting melanoma, BCC and SCC compared to the standard of care, while costs differences varied by cancer type. For melanoma, the cost per detect case was €419 (app) vs. €536 (standard of care) with a detection rate of 0.601 vs. 0.874, yielding an ICER of €430. Most PSA simulations (98.9%) were located in the south-west quadrant. For BCC, costs were €143 vs. €145, with detection rates of 0.585 vs. 0.912 (ICER: €6). 52.4% of the PSA simulations were located in the south-west quadrant and 47.6% in the north-west. For SCC, costs were €228 vs. €233, with detection rates of 0.706 vs. 0.922 (ICER €20). 55.9% of the PSA simulations were located in the south-west quadrant and 44.1% in the north-west.
CONCLUSIONS: Although the smartphone app showed potential cost savings, it was associated with lower detection rates for skin cancer. This may relate to the app’s sensitivity and specificity. A long-term analysis is necessary to evaluate the trade-off between missed or delayed diagnoses and the potential benefits of earlier detection of skin cancer through the app.
METHODS: Three decision-tree models were conceptualized alongside the ARTIS-trial (NCT05246163) to simulate the diagnostic process for the three most common types of skin cancer: melanoma, basal cell carcinoma (BCC) and squamous cell carcinoma (SCC). The trial included 1458 patients with worrisome skin lesions at the Department of Dermatology of Ghent University Hospital. The analyses were performed from a healthcare sector perspective. One-way and probabilistic sensitivity analysis (PSA) were used to account for parameter uncertainty.
RESULTS: The smartphone app was less effective in detecting melanoma, BCC and SCC compared to the standard of care, while costs differences varied by cancer type. For melanoma, the cost per detect case was €419 (app) vs. €536 (standard of care) with a detection rate of 0.601 vs. 0.874, yielding an ICER of €430. Most PSA simulations (98.9%) were located in the south-west quadrant. For BCC, costs were €143 vs. €145, with detection rates of 0.585 vs. 0.912 (ICER: €6). 52.4% of the PSA simulations were located in the south-west quadrant and 47.6% in the north-west. For SCC, costs were €228 vs. €233, with detection rates of 0.706 vs. 0.922 (ICER €20). 55.9% of the PSA simulations were located in the south-west quadrant and 44.1% in the north-west.
CONCLUSIONS: Although the smartphone app showed potential cost savings, it was associated with lower detection rates for skin cancer. This may relate to the app’s sensitivity and specificity. A long-term analysis is necessary to evaluate the trade-off between missed or delayed diagnoses and the potential benefits of earlier detection of skin cancer through the app.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
EE234
Topic
Economic Evaluation
Disease
Oncology