Supervised AI Implementation in Pharmacovigilance

This case study details the implementation of supervised AI in pharmacovigilance by a global pharmaceutical manufacturer. The objective was to enhance patient support through AI-powered chatbots and voicebots while ensuring compliance with regulatory policies that mandate monitoring of AI interactions for potential safety events. The study outlines the challenges faced due to the increasing volume of interactions, which overwhelmed the capacity of trained agents to manually review transcripts. A critical insight revealed that manual reviews did not significantly improve patient safety, as agents labeled safety events incorrectly nearly 40% of the time. The introduction of supervised AI, combined with human validation, allowed for automated detection of adverse events, significantly reducing manual review efforts by approximately 90%. The results included improved accuracy, favorable regulatory audit findings, and a scalable model that maintained safety and compliance, demonstrating the effectiveness of integrating AI with human oversight in pharmacovigilance processes.