INTRODUCTION:
sustainable and equitable healthcare systems require effective drug price regulation, a task challenged by heterogeneous data, complex policies, and manual oversight limitations.

METHODS:
an ai-driven framework was developed integrating a three-layer architecture. first, a drug-centered knowledge graph standardized heterogeneous drug attributes, normalizing 214 irregular dosage form descriptions using nlp and rule-based parsing. second, pricing policies were formalized into executable logic; content-based comparison used a defined formulation (k = α log₂x). third, a hybrid anomaly detection framework combined rule-based triggers with a supervised machine learning classifier, assisted by large language models for complex scenarios.

RESULTS:
the system was evaluated on over 250000 drug entries. standardization achieved accuracies of 95.4% for tablets and 93.3% for capsules. applied to 11191 records from beijing, the anomaly detector identified 33.8% of entries as severe anomalies and 15.6% as warnings. representative results for drugs like tablets and injections showed distinct alert distributions (compliant, moderate, severe) and coefficient of variation values.

CONCLUSIONS:
this informatics engineering approach provides a scalable solution for automated, policy-consistent price monitoring, enhancing transparency and supporting equitable pharmaceutical governance. future work will integrate real-world evidence and advance human-ai collaboration for adaptive pricing regulation.