Proposal for Type Classification for Building Trust in Medical Artificial Intelligence Systems

概要

Medical × AI Study Group has been organizing seminars since 2019, inviting physicians, engineers, policy makers and other stakeholders. The purpose of this study group is to promote the implementation of new technologies in the medical field by sharing useful information in the development and clinical field based on the experiences and knowledge of physicians, developers, and policymakers through seminars.

This recommendation was one of the outputs of the seminar. The original article was accepted to AAAI/ACM Conference on Artificial Intelligence, Ethics and Society and the authors made a presentation on the 8th February.

Policy Recommendation

This paper not only classifies the medical AI systems, but also makes policy recommendations for future medical AI system discussions. Therefore, we will focus on the policy proposal part of this paper and reconstruct it in Japanese as a policy proposal of Institute for Future Initiatives. The full text can be downloaded below.

Policy Recommendation

Policy recommendation 1: Utilize MA Types to Promote Shared Awareness among Stakeholders according to Purposes and Applications


The requirements for medical AI systems vary depending on the clinical purposes and applications. Depending on the magnitude of risk and ripple effects, with the classification, it will become possible to discuss the types that each clinical department and academic society should aim for.

Policy recommendation 2: Utilize MA Types to Construct Real-World Data that Can be Used Effectively


In recent years, there has been a push towards the use of “real-world data” which is data generated in clinical practice. Therefore, from the development stage of AI that supports the input of information, such as electronic medical records, secondary use of data should be promoted to improve the quality of medical care and medical management.

Policy recommendation 3: Utilize MA Types to Promote Institutional Reforms toward Patient-Centered Medicine

Patient-centered medicine is becoming main-stream worldwide as opposed to the medical institutions centered medicine. For patient information control and patient/user benefits, the introduction of the right of patients to handle their own data digitally (the right to data portability) may also be necessary. With these aspects, a system reform including preventive medical care, healthcare, and life in the scope for the patient-centered medicine should take into account the medical systems, responsibility and roles of physicians and AI systems, and benefits of patients/users.

MA Types can be useful to promote discussion regarding the purpose and application of the clinical site. Although MA Types are based on the current technologies and regulations in Japan, but that does not hinder the potential reform of the technologies and regulations. MA Types aims to facilitate discussions among physicians, healthcare workers, engineers, public/patients and policymakers on AI systems in medical practices.

 

Figure：Relationship between MA type and doctor, patient / user and medical AI system

These policy recommendations are extracted from the following proceeding:

Arisa Ema, Katsue Nagakura, and Takanori Fujita. Proposal for Type Classification for Building Trust in Medical Artificial Intelligence System, Proceedings of the 3rd AAAI/ACM Conference on Artificial Intelligence, Ethics and Society (AIES), 2020, NY, USA, pp. 251-7,
doi.org/10.1145/3375627.3375846

  • This Proceeding is published under the Creative Commons license CC-BY 4.0. (Anyone may reuse or modify the content as long as the author and original source are properly cited.)
  • This policy recommendation is also published under the Creative Commons license CC-BY 4.0.