An artificial intelligence based app for skin cancer detection evaluated in a population based setting

Area of Science:

• Dermatology
• Health Informatics
• Artificial Intelligence

Background:

• Mobile health (mHealth) apps utilizing AI for skin lesion classification are increasingly available.
• The impact of these AI-driven mHealth tools on healthcare systems remains largely undocumented.
• A large Dutch health insurance company provided free access to an AI skin cancer detection app for 2.2 million adults.

Purpose of the Study:

• To evaluate the effect of a widely accessible mHealth app on dermatological healthcare consumption.
• To compare the rate of claims for malignant, premalignant, and benign skin lesions between app users and non-users.
• To assess the short-term cost-effectiveness of AI-driven mHealth for detecting skin (pre)malignancies.

Main Methods:

• A retrospective, population-based pragmatic study design was employed.
• 18,960 mHealth app users were matched with 56,880 non-users.
• Odds ratios were calculated to compare dermatological claims within the first year post-app access, alongside a cost-effectiveness analysis.

Main Results:

• mHealth users showed a higher incidence of claims for (pre)malignant skin lesions compared to controls (6.0% vs 4.6%, OR 1.3).
• Users had a more than threefold increased risk of claims for benign skin tumors and nevi (5.9% vs 1.7%, OR 3.7).
• The cost to detect one additional (pre)malignant skin lesion using the app was €2567.

Conclusions:

• AI-powered mHealth apps show promise in increasing the detection of cutaneous (pre)malignancies.
• The current implementation of AI in mHealth leads to a substantial increase in healthcare utilization for benign skin conditions.
• The benefits of enhanced (pre)malignancy detection must be weighed against the increased consumption of dermatological care for non-malignant findings.