Public health surveillance systems depend on a structural bottleneck: physicians must observe, diagnose, and manually report disease cases. This paper proposes pharmaceutical point-of-sale (POS) data as a real-time epidemiological surveillance proxy that bypasses this bottleneck. We introduce the Pharmaceutical Signal Framework (PSF), a three-layer architecture connecting pharmacy transaction data to public health response through AI-driven anomaly detection. We demonstrate that India's GST e-invoicing infrastructure already captures every pharmacy transaction digitally — the technical infrastructure for pharmaceutical surveillance exists; the barrier is architectural and policy-oriented, not technological. We present drug-to-condition signal mappings across three confidence tiers, illustrative detection scenarios for respiratory disease, waterborne outbreaks, and influenza, and a working prototype validated against synthetic ground truth scenarios. This work extends the EPDS framework from enterprise software to public health, demonstrating that perception dependency is a general architectural pattern across information systems. Working paper with prototype implementation available at https://pharmamonitoring.streamlit.app