Clinical named entity recognition in the Portuguese language: a benchmark of modern BERT models and LLMs

TL;DR

This study evaluates BERT-based models and LLMs for Portuguese clinical NER, demonstrating mmBERT's superior performance and the effectiveness of data balancing strategies.

Contribution

It provides a comprehensive benchmark of modern BERT models and LLMs for Portuguese clinical NER, highlighting mmBERT's effectiveness and strategies for class imbalance.

Findings

mmBERT-base achieved the highest micro F1 of 0.76.

Iterative stratification improved class balance and model performance.

Multilingual BERT models perform well for Portuguese clinical NER.

Abstract

Clinical notes contain valuable unstructured information. Named entity recognition (NER) enables the automatic extraction of medical concepts; however, benchmarks for Portuguese remain scarce. In this study, we aimed to evaluate BERT-based models and large language models (LLMs) for clinical NER in Portuguese and to test strategies for addressing multilabel imbalance. We compared BioBERTpt, BERTimbau, ModernBERT, and mmBERT with LLMs such as GPT-5 and Gemini-2.5, using the public SemClinBr corpus and a private breast cancer dataset. Models were trained under identical conditions and evaluated using precision, recall, and F1-score. Iterative stratification, weighted loss, and oversampling were explored to mitigate class imbalance. The mmBERT-base model achieved the best performance (micro F1 = 0.76), outperforming all other models. Iterative stratification improved class balance and overall performance. Multilingual BERT models, particularly mmBERT, perform strongly for Portuguese clinical NER and can run locally with limited computational resources. Balanced data-splitting strategies further enhance performance.