Guidelines for Trustworthy AI application in clinical trials

Abstract

Abstract Issue Innovative information technologies (IIT) like artificial intelligence (AI), big data, etc. promise to support individual patient care, and promote public health. Their use raises ethical, social and legal issues. Here we demonstrate how the guidelines for trustworthy AI, can assist to answer those ethical issues in the case of clinical trials (CT). Description In 2018 the European Commission established the High-Level Expert Group on Artificial Intelligence (AI HLEG). The group proposed Guidelines to promote Trustworthy AI, with three components, which should be met throughout the system's entire life cycle, as it should be lawful, ethical and robust. Trustworthiness is a prerequisite for people and societies to develop, and use AI systems. We used a focus group methodology to explore how the guidelines for trustworthy AI can assist to answer the ethical issues that rise by the application of AI in CTs. Results The discussion was directed to the seven requirements for trustworthy AI in CTs, by questions like: Are they relevant in CTs as a whole? Would they be applicable to the use of IIT as AI in CTs? Are you currently applying part, or all, of the proposed list? In the future, would you attach some, or all, of the proposed list? Is the administrative burden of applying the requirements justified by the effect? Lessons It was recommended that: the guidelines are relevant in the conduct of the CT; planning and implementation of CTs using IIT, should take them into account; ethical aspects and challenges are of the utmost importance; the proposed list is a very comprehensive framework; particular attention should be paid where more vulnerable groups are affected; the administrative burden is acceptable, as the effect exceeds the resources invested. Key messages IIT are becoming increasingly important in medicine, and requirements for trustworthy IIT, and AI are necessary. Appropriate instrument in the case of the CTs are the provided by AI HLEG guidelines.