Large Language Models for World Health Organization–Uppsala Monitoring Centre Drug–Adverse Event Causality Assessment Using Food and Drug Administration Adverse Event Reporting System Cases: Comparative Performance Study - Scorecard - MDSpire
Advertisement
Large Language Models for World Health Organization–Uppsala Monitoring Centre Drug–Adverse Event Causality Assessment Using Food and Drug Administration Adverse Event Reporting System Cases: Comparative Performance Study
Clinical Scorecard: Evaluation of Large Language Models for Assessing Drug-Adverse Event Causality in WHO-Uppsala Monitoring Centre Data Using FDA Adverse Event Reporting System Cases: A Comparative Analysis
At a Glance
Category
Detail
Condition
Drug-Adverse Event Causality Assessment
Key Mechanisms
Utilization of large language models for analyzing spontaneous adverse event reports.
Target Population
Patients with reported adverse drug reactions.
Care Setting
Postmarketing pharmacovigilance
Key Highlights
Causality assessment is essential for signal detection and patient safety.
WHO-UMC system is widely used but relies on expert interpretation.
Large language models may enhance the efficiency of causality assessments.
Study evaluates multiple LLM configurations and prompt engineering strategies.
Internal consistency of LLM outputs is analyzed across repeated assessments.
Guideline-Based Recommendations
Diagnosis
Utilize structured analytic cases for drug-AE causality assessment.
Management
Incorporate LLMs to support expert judgment in causality assessments.
Monitoring & Follow-up
Assess the performance of LLMs using quantitative agreement metrics.
Risks
Consider interrater variability and subjectivity in expert-driven assessments.
Patient & Prescribing Data
Patients with adverse drug reactions reported in FAERS.
Causality assessments require detailed clinical information and structured data.
Clinical Best Practices
Implement standardized criteria for causality assessment.
Ensure comprehensive data extraction from adverse event reports.
Utilize expert consensus as a reference standard for evaluation.
