Exploring the Latest Highlights in Medical Natural Language Processing across Multiple Languages: A Survey

Area of Science:

• Biomedical Informatics
• Computational Linguistics
• Natural Language Processing

Background:

• Biomedical and clinical Natural Language Processing (NLP) research is increasingly expanding beyond English.
• Multilingual NLP presents unique challenges in data availability and model development.

Purpose of the Study:

• To survey the current state of biomedical and clinical NLP in languages other than English (LoE).
• To focus on data resources, language models, and common NLP tasks in LoE.
• To identify research gaps and future opportunities in multilingual medical NLP.

Main Methods:

• Literature review of clinical and biomedical NLP publications from 2020-2022.
• Focus on multilinguality and LoE challenges.
• Database queries and manual selection of relevant studies, supplemented by existing review papers.

Main Results:

• Transformer-based language models are increasingly used for medical NLP tasks in LoE.
• Annotated datasets for clinical NLP in LoE, especially European languages, have grown.
• Common tasks include information extraction, named entity recognition, and negation detection.
• Need for tailored datasets and models for low-resource languages remains.

Conclusions:

• Significant progress in medical NLP for languages other than English is evident.
• Opportunities exist for developing specialized resources for underrepresented languages.
• Further research is needed to address the unique challenges of multilingual medical text processing.