Enhancing abstract screening classification in evidence-based medicine: incorporating domain knowledge into pre-trained models.

Ofori-Boateng, Regina; Aceves-Martins, Magaly; Wirantuga, Nirmalie; Moreno-García, Carlos Francisco

Abstract

Evidence-based medicine (EBM) represents a cornerstone in medical research, guiding policy and decision-making. However, the robust steps involved in EBM, particularly in the abstract screening stage, present significant challenges to researchers. Numerous attempts to automate this stage with pre-trained language models (PLMs) are often hindered by domain-specificity, particularly in EBMs involving animals and humans. Thus, this research introduces a state-of-the-art (SOTA) transfer learning approach to enhance abstract screening by incorporating domain knowledge into PLMs without altering their base weights. This is achieved by integrating small neural networks, referred to as knowledge layers, within the PLM architecture. These knowledge layers are trained on key domain knowledge pertinent to EBM, PICO entities, PubmedQA, and the BioASQ 7B biomedical Q\&A benchmark datasets. Furthermore, the study explores a fusion method to combine these trained knowledge layers, thereby leveraging multiple domain knowledge sources. Evaluation of the proposed method on four highly imbalanced EBM abstract screening datasets demonstrates its effectiveness in accelerating the screening process and surpassing the performance of strong baseline PLMs.

Citation

OFORI-BOATENG, R., ACEVES-MARTINS, M., WIRANTUGA, N. and MORENO-GARCIA, C.F. 2024. Enhancing abstract screening classification in evidence-based medicine: incorporating domain knowledge into pre-trained models. In Finkelstein, J., Moskovitch, R. and Parimbelli, E. (eds.) Proceedings of the 22nd Artificial intelligence in medicine international conference 2024 (AIME 2024), 9-12 July 2024, Salt Lake City, UT, USA. Lecture notes in computer science, 14844. Cham: Springer [online], part I, pages 261-272. Available from: https://doi.org/10.1007/978-3-031-66538-7_26

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 22nd Artificial intelligence in medicine international conference 2024 (AIME 2024)
• Start Date: Jul 9, 2024
• End Date: Jul 12, 2024
• Acceptance Date: Feb 29, 2024
• Online Publication Date: Jul 25, 2024
• Publication Date: Dec 31, 2024
• Deposit Date: Jul 25, 2024
• Publicly Available Date: Jul 26, 2025
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• Volume: Part 1
• Pages: 261-272
• Series Title: Lecture notes in computer science (LNAI)
• Series Number: 14844
• Series ISSN: 0302-9743; 1611-3349
• ISBN: 9783031665370
• DOI: https://doi.org/10.1007/978-3-031-66538-7_26
• Keywords: Pre-trained language models; Domain integration; Transfer learning; Evidence-based medicine; Abstract text classification
• Public URL: https://rgu-repository.worktribe.com/output/2418785
• Related Public URLs: https://rgu-repository.worktribe.com/output/2584551 (paper presented as SICSA REALLM workshop 2024 which has extended results to this output)