EQ-5D Classification Using Biomedical Entity-Enriched Pre-trained Language Models and Multiple Instance Learning

The EQ-5D (EuroQol 5-Dimensions) is a standardized instrument for the evaluation of health-related quality of life. In health economics, systematic literature reviews (SLRs) depend on the correct identification of publications that use the EQ-5D, but manual screening of large volumes of scientific literature is time-consuming, error-prone, and inconsistent. In this study, we investigate fine-tuning of general-purpose (BERT) and domain-specific (SciBERT, BioBERT) pre-trained language models (PLMs), enriched with biomedical entity information extracted through scispaCy models for each statement, to improve EQ-5D detection from abstracts. We conduct nine experimental setups, including combining three scispaCy models with three PLMs, and evaluate their performance at both the sentence and study levels. Furthermore, we explore a Multiple Instance Learning (MIL) approach with attention pooling to aggregate sentence-level information into study-level predictions, where each abstract is represented as a bag of enriched sentences (by scispaCy). The findings indicate consistent improvements in F1-scores (reaching 0.82) and nearly perfect recall at the study-level, significantly exceeding classical bag-of-words baselines and recently reported PLM baselines. These results show that entity enrichment significantly improves domain adaptation and model generalization, enabling more accurate automated screening in systematic reviews.

💡 Research Summary

The paper addresses the labor‑intensive task of identifying PubMed abstracts that report the use of the EuroQol 5‑Dimension (EQ‑5D) instrument, a common requirement in health‑economic systematic literature reviews (SLRs). Manual screening of thousands of papers is time‑consuming, error‑prone, and suffers from low inter‑rater consistency. To automate this process, the authors explore fine‑tuning of three pre‑trained transformer language models—general‑purpose BERT, scientific‑domain SciBERT, and biomedical‑domain BioBERT—augmented with biomedical entity information extracted by three scispaCy pipelines (en_core_sci_sm, en_core_sci_md, en_core_sci_scibert).

The dataset consists of 200 PubMed abstracts previously curated by Kertész et al., with 121 positive (EQ‑5D mentioned) and 79 negative instances. Each abstract is split into sentences using scispaCy; for each sentence, named entities are identified and appended in a simple “