Natural language processing in medical text processing: A scoping literature review.

Area of Science:

• Medical Informatics
• Computational Linguistics

Background:

• Digitized medical data is growing exponentially, increasing complexity for healthcare professionals.
• Natural Language Processing (NLP), particularly Named Entity Recognition (NER), is vital for extracting information from clinical texts.
• Transformer-based models like BERT have revolutionized medical data interpretation.

Purpose of the Study:

• To review and analyze recent NLP approaches for medical text processing.
• To examine techniques, performance metrics, and advancements in NLP for healthcare.
• To assess NLP applications across different languages and healthcare contexts.

Main Methods:

• Scoping literature search using PRISMA methodology in Scopus and PubMed (2019-2024).
• Included studies focused on language model fine-tuning and information extraction in healthcare.
• A specific search query targeted relevant NLP techniques.

Main Results:

• 31 studies were included, with BERT-based approaches, neural networks, and CRF/LSTM techniques dominating.
• These methods consistently achieved F1-scores above 85%.
• Studies covered multiple languages, with a prevalence of English and Chinese, and addressed data privacy and limited data.

Conclusions:

• Modern NLP, especially BERT and hybrid approaches, shows significant promise for medical text processing.
• Challenges in cross-lingual adaptation and data availability persist.
• These technologies have the potential to enhance medical data interpretation and analysis.