SafeRx: A Canonical Pharmaceutical Knowledge Integration Framework for Multimodal Medication Safety

Area of Science:

• Pharmaceutical informatics
• Clinical decision support
• Patient safety research

Background:

• Medication-related harm is a significant patient safety issue due to fragmented data and limited integration of safety domains.
• Current clinical decision-support systems often focus narrowly on drug interactions, neglecting other critical safety aspects like lactation and IV compatibility.

Purpose of the Study:

• To develop and evaluate SafeRx, a platform for integrating heterogeneous medication safety data into a unified framework.
• To create a provenance-aware pharmaceutical knowledge integration platform to address challenges in pharmacovigilance.

Main Methods:

• SafeRx integrated data from regulatory summaries, drug-drug interaction repositories, FDA warnings, lactation safety records, and IV compatibility data.
• Automated extraction, ontology-aware normalization, AI semantic enrichment, and pharmacist-supervised governance were employed.
• The platform consolidated canonical substances, identifier mappings, interaction pairs, boxed warnings, lactation entries, and IV compatibility records.

Main Results:

• The platform successfully integrated data from multiple heterogeneous sources, creating a unified canonical framework.
• SafeRx consolidated 2190 canonical substances, 9932 interaction pairs, 514 boxed warnings, 1895 lactation entries, and 9088 IV compatibility records.
• The system established interoperable, explainable, and pharmacist-supervised medication safety infrastructure.

Conclusions:

• SafeRx demonstrates the feasibility of a unified medication safety infrastructure integrating diverse data domains.
• The platform supports research, pharmacovigilance, and clinical decision-support applications.
• Future research should focus on clinical implementation and impact on prescribing workflows and safety outcomes.