Model Choice Significantly Impacts Causal Inference in Pharmacovigilance

2026-06-17

A comparative analysis within the InferBERT framework shows that domain-specific pre-training is crucial for accurate causal adverse drug event detection. BioBERT outperformed larger models, highlighting the importance of specialized models for pharmacovigilance.

Source: arXiv · cs.LG

Reported by VERA Newswire.