Leveraging Unstructured Data in Electronic Health Records to Detect Adverse Events from Pediatric Drug Use: A Scoping...

Area of Science:

• Pharmacovigilance
• Health Informatics
• Pediatric Drug Safety

Background:

• Adverse drug events (ADEs) in children present significant public health concerns.
• Existing research on detecting and monitoring pediatric ADEs is notably limited.
• Electronic health records (EHRs) contain valuable unstructured data for pharmacovigilance.

Purpose of the Study:

• To conduct a scoping review on the utilization of unstructured EHR data for identifying ADEs in pediatric populations.
• To assess the current state of research and identify gaps in pediatric pharmacovigilance using EHR data.

Main Methods:

• Systematic search of six major databases (MEDLINE, Embase, IEEE Xplore, etc.) conducted in September 2024.
• Inclusion criteria applied to 984 retrieved records, resulting in the selection of nine relevant studies.
• Analysis focused on methods for extracting ADE information from unstructured EHR text.

Main Results:

• A significant research gap exists concerning the identification of ADEs in children.
• Unstructured EHR data holds potential for enhancing pediatric pharmacovigilance.
• Current natural language processing (NLP) methods for ADE extraction from EHRs show limitations in generalizability and contextual interpretation.

Conclusions:

• Despite limited current use, unstructured EHR data is valuable for pediatric pharmacovigilance.
• Advances in transformer-based models and large language models (LLMs) can overcome existing NLP challenges.
• Future research should leverage advanced AI to utilize EHR data at scale for improved pediatric drug safety.