LP-SDA

LP-SDA propagates safety signals across a chemical-structure-based drug similarity network to enhance detection of adverse drug reactions (ADRs) by integrating pre-clinical chemical structures with post-market reports for improved pharmacovigilance of newly approved drugs.

Key Features:

• Integration of Pre-Clinical Drug Structures: Uses pre-clinical chemical structures to construct a drug similarity network that supplements post-market data.
• Enhancement of Safety Signals: Applies a label propagation framework to enhance signals computed by Proportional Reporting Ratio (PRR), Reporting Odds Ratio (ROR), Multi-item Gamma Poisson Shrinker (MGPS), and Bayesian Confidence Propagation Neural Network (BCPNN).
• Addressing Insufficient Case Reports: Mitigates sparse spontaneous reporting for newly approved drugs by propagating information across chemically similar entities and combining pre-clinical data with FDA Adverse Event Reporting System (FAERS) post-market reports.
• Early Detection Capability: Enables earlier identification of potential ADRs for newly approved drugs to support proactive pharmacovigilance.

Scientific Applications:

• Pharmacovigilance Signal Detection: Enhances the accuracy and timeliness of ADR detection by combining pre-clinical chemical structures with FAERS post-market data.
• Safety Assessment for Newly Approved Drugs: Supports assessment of potential ADRs for drugs with limited spontaneous reports to inform safety monitoring and regulatory decision-making.

Methodology:

Initial safety signals are computed using PRR, ROR, MGPS, and BCPNN; a drug similarity network is constructed from pre-clinical chemical structures and integrated with FAERS post-market reports; original safety signals are propagated through the similarity network using a label propagation framework to generate enhanced signals.