AI-Assisted Post-Market Surveillance and Signal Detection: What Regulators Will Still Expect Humans to Defend

Post-market surveillance programs are increasingly incorporating AI-assisted tools to support complaint analysis, signal detection, trend identification, adverse event review, and large-scale data evaluation. While these technologies may help organizations process growing volumes of information more efficiently, regulatory responsibility for safety decisions, escalation activities, risk assessments, MDR reporting, and post-market actions continues to remain firmly with the organization. Questions are increasingly emerging around how AI-supported outputs are interpreted, challenged, verified, and ultimately defended when product safety concerns arise within FDA, EU MDR, and other regulated post-market environments. This session examines the practical realities of using AI-assisted approaches within post-market surveillance activities while maintaining defensible regulatory oversight. The discussion will focus on signal evaluation, complaint review processes, documentation practices, escalation decisions, and the role of human judgment when automated tools contribute to safety-related assessments. Participants will better understand where organizations remain accountable, how oversight expectations may be applied in practice, and what regulators will still expect teams to explain, justify, and defend. WHY YOU SHOULD ATTEND Finding potential signals is rarely the difficult part of post-market surveillance. Organizations routinely collect complaint information, adverse event data, MDR submissions, service records, and other sources of post-market evidence. The difficult decisions begin when teams must determine whether a pattern is meaningful, whether additional investigation is justified, whether escalation is warranted, or whether available evidence supports reporting, field action, or continued monitoring. As AI-assisted tools become more capable of identifying patterns and anomalies, the volume of information requiring interpretation may increase significantly while responsibility for the resulting decisions remains unchanged. FDA and EU MDR requirements do not evaluate whether an algorithm identified a signal. They evaluate how the organization interpreted the information, what actions were taken, what evidence supported those actions, and how conclusions were ultimately reached. This session focuses on the judgment, rationale, and decision-making processes that remain essential within post-market surveillance programs regardless of how sophisticated the underlying technology becomes. Code: FDB1417