A major blockade to support the evidence-based clinical decision-making is accurately and efficiently recognizing appropriate and scientifically rigorous studies in the biomedical literature. We trained a multi-layer perceptron (MLP) model on a dataset with two textual features, title and The dataset consisting of 7958 PubMed citations classified in two classes: scientific rigor and non-rigor, is used to train the proposed model. We compare our model with other promising machine learning models such as Support Vector Machine (SVM), Decision Tree, Random Forest, and Gradient Boosted Tree (GBT) approaches. Based on the higher cumulative score, deep learning was chosen and was tested on test datasets obtained by running a set of domain-specific queries. On the training dataset, the proposed deep learning model obtained significantly higher accuracy and AUC of 97.3% and 0.993, respectively, than the competitors, but was slightly lower in the recall of 95.1% as compared to GBT. The trained model sustained the performance of testing datasets. Unlike previous approaches, the proposed model does not require a human expert to create fresh annotated data; instead, we used studies cited in Cochrane reviews as a surrogate for quality studies in a clinical topic. We learn that deep learning methods are beneficial to use for biomedical literature classification. Not only do such methods minimize the workload in feature engineering, but they also show better performance on large and noisy data.
To support evidence-based precision medicine and clinical decision-making, we need to identify accurate, appropriate, and clinically relevant studies from voluminous biomedical literature. To address the issue of accurate identification of high impact relevant articles, we propose a novel approach of attention-based deep learning for finding and ranking relevant studies against a topic of interest. For learning the proposed model, we collect data consisting of 240,324 clinical articles from the 2018 Precision Medicine track in Text REtrieval Conference (TREC) to identify and rank relevant documents matched with the user query. We built a BERT (Bidirectional Encoder Representations from Transformers) based classification model to classify high and low impact articles. We contextualized word embedding to create vectors of the documents, and user queries combined with genetic information to find contextual similarity for determining the relevancy score to rank the articles. We compare our proposed model results with existing approaches and obtain a higher accuracy of 95.44% as compared to 94.57% (the next best performer) and get a higher precision by about 14% at P@5 (precision at 5) and about 12% at P@10 (precision at 10). The contextually viable and competitive outcomes of the proposed model confirm the suitability of our proposed model for use in domains like evidence-based precision medicine.