Medical Artificial Intelligence Standard System: History and Current Status

The standardization of medical artificial intelligence (AI) is currently in its infancy and falls short of meeting the needs for the development, deployment, control, assessment, and guidance of medical AI products. This not only makes it difficult to standardize the research and development process and therefore increase the cost and affect the quality of the products, but also leads to challenges in achieving unified interaction, comparison, and evaluation of AI products. It may result in incorrect estimation and evaluation of products, thus misguiding the direction of medical AI development. Consequently, establishing a mature and unified standard system for medical AI has become an urgent priority. To facilitate the advancement of the medical AI standard system from its nascent stage to maturity, we conduct an in-depth analysis of the development history of medical AI standards from four aspects: medical data standards, standard datasets, benchmarks, and norms/guidelines. By revealing the problems in the current medical AI standards, we aim to provide a reference for related research.