BACKGROUND
There is a great interest in using artificial intelligence (AI) to screen for skin cancer. This is fueled by a rising incidence of skin cancer and an increasing scarcity of trained dermatolo-gists. AI systems, capable of identifying melanoma, could save lives, enable immediate ac-cess to screenings, reduce unnecessary care and healthcare costs. While such AI-based sys-tems are useful from a public health perspective, past research has shown that individual patients are very hesitant about being examined by an AI system
OBJECTIVE
The aim of the present study was twofold. First, to determine how important the attributes provider (in-person physician, physician via teledermatology, AI, vs. personalized AI), costs of screening (free, 10€, 25€, vs. 40€) and waiting time (immediate, 1 day, 1 week, 4 weeks) were for patients’ choices of a particular mode of skin cancer screening. Second, to investi-gate whether sociodemographic characteristics, especially, age, were systematically related to participants’ individual choices.
METHODS
The study used choice-based conjoint-analysis to examine the acceptance of medical AI for a skin cancer screening from the patient's perspective. Participants responded to twelve choice sets, each containing three screening-variants, where each variant was described through attributes; provider, costs and waiting time. Furthermore, sociodemographic charac-teristics (age, gender, income, job status, educational background) were assessed.
RESULTS
126 (33%) respondents completed the online survey. The results from the conjoint analysis showed that the three attributes were more or less equal important for the participant’s choices, with provider being the most important. Inspecting the individual part worths showed that treatment by a physician was most preferred, followed by e-consultation with a physician and personalized AI. The three AI levels scored significantly lower. Concerning the relationship between sociodemographic characteristics and relative importances we found, that only age showed a significant positive association to the important of the attrib-ute provider (r = 0.21; p < .02). Younger participants put a lesser importance on the provider than older participants. All other correlations were not significant.
CONCLUSIONS
The present study adds to the growing body of research using choice-experiments to investi-gate the acceptance of artificial intelligence in health contexts. Future studies need to ex-plore the reasons why AI is accepted or rejected and whether sociodemographic characteris-tics are associated this decision.