Mining Real-World Big Data to Characterize Adverse Drug Reaction Quantitatively: Mixed Methods Study.

Area of Science:

• Pharmacovigilance and Drug Safety
• Computational Toxicology
• Precision Medicine

Background:

• Adverse drug reactions (ADRs) are a significant concern in clinical medicine, impacting drug supervision and new drug discovery.
• Current drug safety evaluations are often qualitative, necessitating quantitative methods for improved accuracy.

Purpose of the Study:

• To develop a quantitative model for characterizing Adverse Drug Reaction (ADR) severity.
• To establish a robust system for evaluating drug safety and improving pharmacovigilance.

Main Methods:

• Developed the Adverse Drug Reaction Classification System (ADReCS) severity-grading model using millions of real-world adverse event reports.
• Introduced a 'Severity_score' parameter with defined boundaries for five severity grades.
• Mined medication prescriptions to calculate ADR occurrence rates in large patient populations.

Main Results:

• The ADReCS model achieved 99.22% consistency with expert grading (Common Terminology Criteria for Adverse Events).
• Quantitatively graded the severity of 6,277 ADRs for 129,407 drug-ADR pairs.
• Calculated occurrence rates for 6,272 distinct ADRs across 127,763 drug-ADR pairs.

Conclusions:

• This study provides the first comprehensive quantitative assessment of both ADR severity grades and frequencies.
• Establishes a foundation for AI-driven drug discovery, focusing on high efficacy and low toxicity.
• Marks a shift towards quantitative evaluation in clinical toxicity research.