AI
LusoClin: A Synthetic Clinical Note Dataset for Medical Information Retrieval in European Portuguese
Publication: cs.AI updates on arXiv.org
Author: Carla Teixeira Lopes, Daniel F\'elix
Source published: 19/11/2026, 05:00:00 UTC
Briefing created: 19/11/2026, 05:08:20 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 19/11/2026, 05:03:24 UTC.
Briefings
Researchers released LusoClin, a synthetic dataset of clinical notes in European Portuguese designed to advance medical information retrieval while respecting privacy. Generated by fine-tuning Llama 3.2 (3B) with Quantized Low-Rank Adaptation and BERT-based entity annotation, the dataset captures relevant language patterns but contains some medical inconsistencies and is distinguishable from real notes.
Key points
- LusoClin is a publicly released dataset of synthetic clinical notes in European Portuguese generated via Llama 3.2 (3B) fine-tuning and BERT-based entity annotation.
- The dataset supports downstream IR applications like clinical document retrieval and domain-specific model adaptation, though annotators can often distinguish synthetic from real notes.
Context
This work addresses the critical need for large-scale clinical text collections in European Portuguese, which is limited by privacy regulations. The dataset enables privacy-preserving research on clinical text retrieval and supports domain-adapted retrieval models.
Limitations
The dataset contains some medical inconsistencies and annotators can often distinguish synthetic from real notes. The evaluation focuses on utility, fidelity, diversity, and privacy using automatic metrics and human evaluation, with no specific performance benchmarks provided.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 19/11/2026, 05:08:20 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeiefo4ljsm7kxd4ze6ekwwu2uvg7aol6ua6ntckf7ilmsmw6br7yw4