Evaluation of a prototype machine learning tool to semi-automate data extraction for systematic literature reviews

Abstract

Background: Evidence-based medicine requires synthesis of research through rigorous and time-intensive systematic literature reviews (SLRs), with significant resource expenditure for data extraction from scientific publications. Machine learning may enable the timely completion of SLRs and reduce errors by automating data identification and extraction.

Methods: We evaluated the use of machine learning to extract data from publications related to SLRs in oncology (SLR 1) and Fabry disease (SLR 2). SLR 1 predominantly contained interventional studies and SLR 2 observational studies. Predefined key terms and data were manually annotated to train and test bidirectional encoder representations from transformers (BERT) and bidirectional long-short-term memory machine learning models. Using human annotation as a reference, we assessed the ability of the models to identify biomedical terms of interest (entities) and their relations. We also pretrained BERT on a corpus of 100,000 open access clinical publications and/or enhanced context-dependent entity classification with a conditional random field (CRF) model. Performance was measured using the F₁ score, a metric that combines precision and recall. We defined successful matches as partial overlap of entities of the same type.

Results: For entity recognition, the pretrained BERT+CRF model had the best performance, with an F₁ score of 73% in SLR 1 and 70% in SLR 2. Entity types identified with the highest accuracy were metrics for progression-free survival (SLR 1, F₁ score 88%) or for patient age (SLR 2, F₁ score 82%). Treatment arm dosage was identified less successfully (F₁ scores 60% [SLR 1] and 49% [SLR 2]). The best-performing model for relation extraction, pretrained BERT relation classification, exhibited F₁ scores higher than 90% in cases with at least 80 relation examples for a pair of related entity types.

Conclusions: The performance of BERT is enhanced by pretraining with biomedical literature and by combining with a CRF model. With refinement, machine learning may assist with manual data extraction for SLRs.

Keywords: Evidence-based practice; Information science; Information storage and retrieval; Methods; Systematic reviews as topic.

Fig. 1

Our development process for refining language models to perform entity recognition and relation extraction. BERT bidirectional encoder representations from transformers, BiLSTM bidirectional long-short-term memory, CRF conditional random field, SLR, systematic literature review

Fig. 2

Performance of the pretrained BERT+CRF model across entity types. Panel A presents the relaxed F₁ scores and panel B compares actual and predicted entity labels using confusion matrices. In B, some lines do not sum to 100% owing to rounding. BERT bidirectional encoder representations from transformers, CRF conditional random field, eGFR estimated glomerular filtration rate, PFS progression-free survival SLR systematic literature review