Skip to main navigation Skip to search Skip to main content

EPEE: towards efficient and effective foundation models in biomedicine

  • Zaifu Zhan
  • , Shuang Zhou
  • , Huixue Zhou
  • , Zirui Liu
  • , Rui Zhang

Research output: Contribution to journal › Article › peer-review

Abstract

Foundation models, including language models, e.g., GPT, and vision models, e.g., CLIP, have significantly advanced numerous biomedical tasks. Despite these advancements, the high inference latency and the “overthinking” issues in model inference impair the efficiency and effectiveness of foundation models, thus limiting their application in real-time clinical settings. To address these challenges, we proposed EPEE (Entropy- and Patience-based Early Exiting), a novel hybrid strategy designed to improve the inference efficiency of foundation models. The core idea was to leverage the strengths of entropy-based and patience-based early exiting methods to overcome their respective weaknesses. To evaluate EPEE, we conducted experiments on three core biomedical tasks-classification, relation extraction, and event extraction-using eight foundation models (i.e., BERT, ALBERT, GPT-2, ViT, Qwen, GPT-oss, BioMistral, and Meditron3) across twelve datasets, including clinical notes and medical images. The results showed that EPEE significantly reduced inference time while maintaining or improving accuracy, demonstrating its adaptability to diverse datasets and tasks. EPEE addressed critical barriers to deploying foundation models in healthcare by balancing efficiency and effectiveness. It potentially provided a practical solution for real-time clinical decision-making with foundation models, supporting reliable and efficient workflows.

Original languageEnglish (US)
Article number30
Journalnpj Health Systems
Volume3
Issue number1
DOIs
StatePublished - Dec 2026

Bibliographical note

Publisher Copyright:
© The Author(s) 2026.

Fingerprint

Dive into the research topics of 'EPEE: towards efficient and effective foundation models in biomedicine'. Together they form a unique fingerprint.

Cite this