Health Information Research Unit AutoML 2024

The data included in this dataset include lists of PubMed IDs and classification of articles as meeting or not meeting methodological criteria for rigor and clinical relevance. The articles cover a broad range of topics from about 125 clinical journals, and include original studies, reviews, evidence-based guidelines, and other types of studies. Each article has been manually classified for rigor by expert research associates and reviewed by clinicians with research methods expertise. Articles that meet rigor are then reviewed by practicing clinicians on the relevance to practice within their clinical domain. The methodological criteria are described here https://hiruweb.mcmaster.ca/hkr/what-we-do/methodologic-criteria/. The data was used in a study to train a machine learning algorithm to classify articles for methodological rigor using automated machine learning approaches.