The rapid growth of biomedical literature poses challenges for manual knowledge curation and synthesis. Biomedical Natural Language Processing (BioNLP) automates the process. While Large Language Models (LLMs) have shown promise in general domains, their effectiveness in BioNLP tasks remains unclear due to limited benchmarks and practical guidelines. We perform a systematic evaluation of four LLMs, GPT and LLaMA representatives on 12 BioNLP benchmarks across six applications. We compare their zero-shot, few-shot, and fine-tuning performance with traditional fine-tuning of BERT or BART models. We examine inconsistencies, missing information, hallucinations, and perform cost analysis. Here we show that traditional fine-tuning outperforms zero or few shot LLMs in most tasks. However, closed-source LLMs like GPT-4 excel in reasoning-related tasks such as medical question answering. Open source LLMs still require fine-tuning to close performance gaps. We find issues like missing information and hallucinations in LLM outputs. These results offer practical insights for applying LLMs in BioNLP.
@article{arxiv.2305.16326,
title = {Benchmarking large language models for biomedical natural language processing applications and recommendations},
author = {Qingyu Chen and Yan Hu and Xueqing Peng and Qianqian Xie and Qiao Jin and Aidan Gilson and Maxwell B. Singer and Xuguang Ai and Po-Ting Lai and Zhizheng Wang and Vipina Kuttichi Keloth and Kalpana Raja and Jiming Huang and Huan He and Fongci Lin and Jingcheng Du and Rui Zhang and W. Jim Zheng and Ron A. Adelman and Zhiyong Lu and Hua Xu},
journal= {arXiv preprint arXiv:2305.16326},
year = {2025}
}