Given the proliferation of scientific literature, it’s more important than ever to have tools which help researchers and clinicians quickly identify recent, relevant articles. This is especially true when it comes to identifying the randomized control trials (RCTs) which drive evidence-based practice in so many fields.
In a recent issue of the Journal of the American Medical Informatics Association (JAMIA), Marshal et al describe the creation of a tool called Trialstreamer, which promises to assist searchers in the identification of the most current RCTs on their topics of interest. What separates Trialstreamer from other resources though is its complete reliance on machine learning to identify published RCTs from PubMed and information about ongoing trials from the World Health Organization International Clinical Trials Registry Platform (ICTRP).
Considering the delay that exists from the time an article is entered into PubMed to the time that it is assigned a ‘publication type’ tag, relying on the RCT tag to identify all pertinent articles will mean missing the most recent entries. Trialstreamer’s algorithm identifies RCTs immediately upon their
inclusion in PubMed allowing searchers to locate the most current RCTs.
What makes Trialstreamer especially remarkable is that beyond simply identifying RCTs, its algorithm extracts information for analysis such as the elements of PICO (population, intervention, control, and outcomes), sample sizes, and an estimate of the risk of bias in the article.
To give Trialstreamer a try, click on the link from the Himmelfarb Library’s Electronic Database page then select Research Tools.
Marshall, I., Nye, B., et al. (2020) Trialstreamer: A living, automatically updated database of clinical trial reports. J Amer Med Informatics Assoc. 27(12): 1903-1912.
