Bernardo Magnini

The EDEN dataset not only fills a critical gap in Italian clinical data but also introduces a novel benchmark for structured information extraction that could transform medical NLP applications.

Unlabeled data can be transformed into powerful training examples, boosting classification performance by nearly 20 points without the need for costly annotations.

LLMs can fill out medical forms from Italian clinical notes in a zero-shot setting, but watch out for those "unknown" biases.

Small language models can not only match but significantly outperform their larger counterparts in specialized medical NLP tasks, challenging the assumption that bigger is always better.