Back

PubChat: Evaluating the Effectiveness of Semantic Search in Establishing Drug Safety in Children

Smith, A. N.; Dunbar, E.; Crowder, C.; Might, M.

2025-10-03 pediatrics
10.1101/2025.10.01.25335910 medRxiv
Show abstract

Children often respond differently to pharmaceuticals differently than adults, necessitating dedicated safety labeling. Although pediatric safety labeling is available through the FDA Online Label Repository, updates often lag newly published research, and manual searches of drug lists lack efficiency. To address these limitations, we developed PubChat, a tool that leverages semantic search and language models to extract, summarize, and categorize pediatric safety information from scientific literature, enabling rapid comparison with FDA-approved labeling. In a random sample of 80 drugs, PubChat agreed with FDA labeling in 41.25% of cases. In 23.75% of cases, it found supporting evidence that the FDA age recommendation was potentially too conservative. By transforming unstructured literature into structured, actionable insights, PubChat establishes a foundation for enhanced surveillance of pediatric drug safety and offers a complementary tool to bridge the gap between evolving research and static regulatory labeling.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.