PubMedQA: A Dataset for Biomedical Research Question Answering

By Qiao Jin, Bhuwan Dhingra, Zhengping Liu, William W. Cohen, Xinghua Lu. 2019

1,772 citations recorded by the source corpus.

We introduce PubMedQA, a novel biomedical question answering (QA) dataset collected from PubMed abstracts. The task of PubMedQA is to answer research questions with yes/no/maybe (e.g.: Do preoperative statins reduce atrial fibrillation after coronary artery bypass grafting?) using the corresponding abstracts. PubMedQA has 1k expert-annotated, 61.2k unlabeled and 211.3k artificially generated QA instances. Each PubMedQA instance is composed of (1) a question which is either an existing research article title or derived from one, (2) a context which is the corresponding abstract without its conclusion, (3) a long answer, which is the conclusion of the abstract and, presumably, answers the research question, and (4) a yes/no/maybe answer which summarizes the conclusion. PubMedQA is the first QA dataset where reasoning over biomedical research texts, especially their quantitative contents, is required to answer the questions. Our best performing model, multi-phase fine-tuning of BioBERT with long answer bag-of-word statistics as additional supervision, achieves 68.1% accuracy, compared to single human performance of 78.0% accuracy and majority-baseline of 55.2% accuracy, leaving much room for improvement. PubMedQA is publicly available at https://pubmedqa.github.io.

Subfields: Artificial Intelligence, Molecular Biology, Natural Language Processing, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Molecular Biology

In Natural Language Processing

In Evaluation & Benchmarks

Citations in the corpus

The RefinedWeb Dataset for Falcon LLM: Outperforming Curated Corpora with Web Data, and Web Data Only (2023, background): “…(Amini et al., 2019) Multiple-choice question answering 20.0 ✓ PROST (Aroca-Ouellette et al., 2021) Paraphrase identification 50.0 ✓ PubMedQA (Jin et al., 2019) Multiple-choice question answering 50.0 ✓ SciQ (Welbl et al., 2017) Multiple-choice question answering 25.0 ✓ ✓ biggest clusters in…”
Galactica: A Large Language Model for Science (2022, neutral): “(Jin et al., 2019; Pal et al., 2022).”
Large language models encode clinical knowledge (2022, methodology): “These include datasets that assess professional medical knowledge such as medical exam questions [33, 64], questions that require medical research comprehension skills [34], and questions that require the ability to assess user intent and provide helpful answers to their medical information needs [1”
MedAgents: Large Language Models as Collaborators for Zero-shot Medical Reasoning (2023, methodology): “The number k of options is 4 except for PubMedQA (3).”
In-BoXBART: Get Instructions into Biomedical Multi-Task Learning (2022, neutral): “(Uzuner et al., 2007) to perform this task.”
A study of generative large language model for medical research and healthcare (2023, positive): “We compared GatorTronGPT with six existing transformer models using three widely used benchmark dataset including PubMedQA 47 — a biomedical question answering dataset collected from PubMed abstracts, which requires answering questions with ‘ yes/no/maybe ’ ; MedMCQA 48 — a large-scale multi-choice…”
BioGPT: Generative Pre-trained Transformer for Biomedical Text Generation and Mining (2022, methodology): “…KD-DTI [14], BC5CDR [13], DDI [15] Question Answering QA-Net [26], LUKE [27], BERT [2], PubMedBERT [9], BioELECTRA [28], LinkBERT [29] PubMedQA [16], BioASQ [30, 31] Document Classiﬁcation BERT [2], BlueBERT [8], SciBERT [18], SPECTER [32], PubMedBERT [9], BioELECTRA [28], LinkBERT [29] HoC…”
LMFlow: An Extensible Toolkit for Finetuning and Inference of Large Foundation Models (2023, background): “The evaluations on three medical datasets revealed significant enhancements in both in-domain (PubMedQA [20], MedMCQA [25]) and out-of-domain (MedQA-USMLE [19]) datasets.”
Can Generalist Foundation Models Outcompete Special-Purpose Tuning? Case Study in Medicine (2023, background): “• PubMedQA [15] contains tests requiring a yes, no, or maybe answer to biomedical research questions when given context provided from PubMed abstracts.”
A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers (2021, background): “(Jin et al., 2019) derives Yes/No/Maybe questions from PubMed paper titles answered from the conclusion sections of the corresponding abstracts.”
Capabilities of GPT-4 on Medical Challenge Problems (2023, background): “• PubMedQA : PubMedQA contains tests requiring a model to answer yes, no, or maybe to biomedical research questions when given context provided from PubMed abstracts [JDL + 19].”
MEDITRON-70B: Scaling Medical Pretraining for Large Language Models (2023, methodology): “We assess the downstream performance of the trial models by evaluating the finetuned models on the training sets of PubMedQA, MedMCQA, and MedQA.”
M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation (2024, positive): “For English, we incorporate 8 datasets, including HotpotQA (Yang et al., 2018), Trivi-aQA (Joshi et al., 2017), NQ (Kwiatkowski et al., 2019), MS MARCO (Nguyen et al., 2016), COL-IEE (Kim et al., 2023), PubMedQA (Jin et al., 2019), SQuAD (Rajpurkar et al., 2016), and NLI data from SimCSE (Gao et al.”