Beyond Track-and-Trace: Behavioural AI for Pharmaceutical Supply-Chain Assurance

This presentation explores how behavioural artificial intelligence can strengthen pharmaceutical supply-chain assurance beyond conventional track-and-trace systems. Using GS1 EPCIS custody-transfer data, it explains how deterministic rules, calibrated GANomaly and traditional anomaly detectors can identify different forms of suspicious behaviour, including extreme delays, unusual routes, missing endpoints and replayed events. The presentation reports findings from a leakage-controlled, five-seed experiment and demonstrates that hybrid architectures combining explicit rules with behavioural models offer stronger and more stable detection than GANomaly alone. It also examines operational risks, governance responsibilities, model limitations and a practical 90-day pilot approach for responsible organisational adoption.