Ensembles of Large Language Models for Identifying EQ-5D Studies in PubMed Based on Their Abstracts

This study presents a multi-phase framework utilizing Google's Gemini and Gemma large language models to automate the identification of EQ-5D studies in PubMed based on abstracts. The ensemble of gemini-2.5-pro, gemma-3-12b, and gemma-3-27b achieved a weighted F1-score and accuracy of 0.74, demonstrating improved precision and recall over individual models. This approach highlights the potential for ensemble-based LLMs to enhance efficiency and reliability in systematic literature reviews within biomedical research.