Mind the gap! Overcoming the lack of evidence to support the innovation journey of AI in healthcare

The rapid development of artificial intelligence (AI) applications for the healthcare setting confronts providers and practitioners with the challenge of choosing those applications that have the best chance to reduce the burden of care in their context. This is challenging due to a general lack of evaluation metrics and because the evidential claims provided by AI vendors are not always in line with the forms of evidence needed by healthcare providers and practitioners. This evidence gap currently harms the development of trust in and acceptability of AI, and thereby hampers the successful implementation and adoption of AI in healthcare. In this viewpoint, we argue that closing this evidence gap is crucial to helping AI achieve its full potential in the healthcare context and we provide practical guidance towards this objective.