Normative and fuzzy aspects of medical AI

Abstract

Normative and fuzzy aspects of Medical AI are examined and a framework is presented for the purpose of designing architectures that can accommodate the lifelong growth of data and knowledge pertaining to an individual's health, together with relevant privileges and permissions for access, growth and utilization of such a system. Considering disease through the normativistic perspective advanced by Sadegh-Zadeh, an action-theoretical analysis is adopted suitable for Medical AI. In this approach diseases are viewed as social artifacts mapped into deontic constructs. In a sense, they are action provoking states of affairs amenable to fuzzy computational approaches that in the past have found broad applicability to industrial process control. © 2019 IEEE.