Dataset for: User Preferences and Willingness to Pay for physician vs. AI-delivered healthcare sys-tems: A Systematic Review of Discrete Choice Experiments

Objectives: As AI-powered systems increasingly shape healthcare, understanding user preferences and willingness to pay (WTP) for physician-led versus AI-led care is crucial for successful integration. This systematic review summarizes discrete choice experiments (DCEs) that investigated user preferences for AI-powered healthcare systems and WTP for physician-led compared to AI-led care. Methods: A systematic literature search was conducted in PubMed and PsycINFO in January 2025 (PROSPERO: CRD42024627150). The review included DCEs published in English between 2000 and 2025, focusing on AI-powered systems in healthcare and examining preferences among patients, healthcare professionals or the general public. Results: A total of 16 DCEs met the inclusion criteria, with most focusing on AI-powered diagnostics, particularly skin cancer screening. The general public was the primary study population (n=13), while physicians were analysed in six studies. The most frequently examined attribute was provider (n=15), followed by accuracy (n=13) and cost (n=11). Physicians ranked accuracy as the most important attribute, while the general public prioritized provider (physician vs. AI). Regarding WTP, the general public showed positive WTP for physician-led care, i.e. payed extra to keep physicians in the loop, whereas physicians exhibited negative WTP, i.e. payed extra for autonomous systems without physicians. Conclusions: In realistic choice conditions participants prefer physician involvement and are willing to pay extra for it. This review highlights the importance of accuracy and provider for adoption of AI-powered systems. So far only a handful of choice-based studies were conducted to study behaviour. While far from perfect, they offer a more realistic view than questionnaires.