Natural Language Processing for Digital Health in the Era of Large Language Models.

Area of Science:

• Biomedical Natural Language Processing (NLP)
• Artificial Intelligence in Healthcare

Background:

• Large language models (LLMs) are rapidly advancing NLP capabilities.
• The healthcare sector is exploring diverse applications for LLMs.

Purpose of the Study:

• To synthesize current research on LLMs in medical NLP.
• To summarize advancements and applications of LLMs in healthcare.

Main Methods:

• Conducted a selective literature review of recent studies on LLMs in biomedical NLP.
• Sourced literature from PubMed, ACL Anthology, IEEE Explore, and Google Scholar.
• Analyzed LLMs customized for medical texts and various data sources (medical literature, EHRs, social media).

Main Results:

• General-purpose LLMs are popular, but customized open-source LLMs for biomedical tasks are increasing.
• Applications in EHRs and biomedical literature are more prominent than social media.
• Traditional models outperform LLMs in classification/NER; LLMs excel in few-shot and generative tasks.
• Research on LLM evaluation, bias, privacy, and equitability is limited.

Conclusions:

• LLMs hold significant potential to revolutionize medical NLP tasks.
• Future research must prioritize task-oriented evaluation, bias mitigation, and equitable LLM deployment in healthcare.