Abstract
BACKGROUND: Mobile health (mHealth) applications (apps) integrated with artificial intelligence for skin cancer triage are increasingly available to the general public. Nevertheless, their actual uptake is limited. Although endorsement by healthcare providers (HCPs) is one of the perceived facilitators for using this technology, the perceptions of key HCPs in skin cancer triage towards those apps have not been studied. OBJECTIVES: To explore key HCPs' perceived risks, benefits, and preconditions for endorsement of mHealth apps for skin cancer triage in the general population. METHODS: An in-depth qualitative online focus group (FG) study was conducted consisting of six focus groups: three with dermatologists and three with general practitioners (GPs). Dutch dermatologists and GPs were selected using purposive sampling based on age, knowledge and previous experience with AI. A total of sixteen dermatologists and seventeen GPs attended 90-minute FGs. Data were analyzed by a multidisciplinary team in a thorough thematic content analysis using multiple phases of coding derived from Grounded Theory. RESULTS: A total of four main risks, three main benefits, and four main preconditions for endorsement were identified. Risks perceived by HCPs concerned incorrect diagnoses, exclusion of subpopulations, and loss of GP autonomy in clinical decision making and diagnostic experience. Perceived benefits were increased skin cancer awareness, facilitation of the early detection of skin cancer, and a streamlined patient journey. Preconditions for endorsement were evidence-based verification of accuracy, integration in clinical practice, clarity about liability in case of adverse events, and accessible and inclusive app design. CONCLUSIONS: Although HCPs perceive pivotal risks related to the implementation of mHealth apps, they also foresee important benefits when implemented successfully. In order for HCPs to endorse those apps, emphasis must be placed on integrating accurate mHealth apps with accessible and inclusive design and functionality into clinical practice, factors that currently appear to be largely unmet.