In recent years, artificial intelligence (AI) systems including machine learning, are expected to be used to reduce the burden on doctors and healthcare workers. However, there are many challenges in clinical implementation. In order to facilitate discussions on medical AI software systems among healthcare professionals, technology developers, policy makers, and public/patients, this paper proposed a type classification for medical AI systems (MA Type). In addition to technical requirements, we have developed a classification system that includes the perspectives of user interface, institutional design, and the role and impact on health professionals and patients/users. In developing and implementing medical AI systems, we hope that MA types will be used to share awareness among healthcare professionals and technology developers. The following three recommendations are made regarding the use of MA types and their future development.
発行年
2020-04-14
権利
CC BY 4.0
著者版フラグ
publisher
出版者
Institute for Future Initiatives, The University of Tokyo
Proposal for Type Classification for Building Trust in Medical Artificial Intelligence Systems
policy_recommendation_tg_20200213.pdf
(169.84KB)
