The Strange case of Dr. Watson: liability implications of evidence-based decision support systems in the health care

This paper investigates some legal issues emerging from the adoption of Watson and similar AI systems in the medical area. In so doing, we explore a set of questions whose answers may heavily affect the liability allocation in misdiagnosis and/or improper treatment scenarios. Section 2 explores what are the distinctive features of new generation CDSS compared to the traditional ones. Section 3 investigates whether and to what extent such features pose questions with regard to the source of the decision making authority. Section 4 deals with the legal qualification and the conformity assessment procedure of these new AI CDSS, under the European discipline of Medical Device Software. In particular, we shall evaluate whether additional criterium for the classification of those systems are needed and how they can influence the certification procedures and the medical liability. Finally, section 5 explores how and to what extent the level of automation may affect the liability allocation. To this end, in section 6 we design some scenarios, which provide variations on the possible causes of failure in the decision making process and the consequent liability assessment.