Biomedical Natural Language Processing in the Era of Large Language Models

Area of Science:

• Biomedical informatics
• Artificial intelligence
• Natural language processing

Background:

• Biomedicine has undergone significant digitization, from genomic sequencing to electronic medical records.
• Large language models (LLMs) are now driving a generative artificial intelligence (AI) revolution in natural language processing (NLP).

Purpose of the Study:

• To review the challenges and opportunities in biomedical NLP.
• To provide historical context and survey the current state of the art.
• To explore future frontiers for AI researchers and biomedical practitioners.

Main Methods:

• Review of current literature and emerging trends in biomedical NLP.
• Analysis of the impact of LLMs on healthcare and biomedical discovery.
• Discussion of challenges including hallucinations, omissions, compliance, and safety.

Main Results:

• Biomedical NLP automates tasks like knowledge extraction and medical abstraction, boosting productivity.
• Emerging AI approaches promise creative gains and uncovering new capabilities using large-scale data.
• Integrating diverse data modalities like imaging and genomics is crucial for comprehensive solutions.

Conclusions:

• LLMs offer unprecedented possibilities for optimizing patient care and accelerating biomedical discovery.
• Ensuring LLM compliance, safety, and addressing limitations like hallucinations are critical.
• The integration of AI, particularly LLMs, is poised to transform biomedical research and practice.