The Continuous Active Learning (CAL®) Tool is designed to help researchers efficiently screen large numbers of citations from PubMed searches. It uses an AI-driven process where the tool learns from your inclusion and exclusion decisions to prioritize the most relevant articles for review.
No, the CAL® Tool does not perform literature searches. You must first complete your search on PubMed, export your results as a PMID .txt file, and then upload this file to the CAL® Tool.
The tool applies Continuous Active Learning, a form of machine learning that iteratively learns from your screening decisions. As you mark articles as relevant or irrelevant, the AI reprioritizes the remaining citations, bringing the most likely relevant studies to your attention sooner.
CAL® is a tool that best compliments rapid evidence synthesis projects such as rapid reviews or rapid scoping reviews. We recommend using CAL® for searches that return over >5000 citations.
Two people can screen citations within the same project, however there may be a small risk of over-writing one reviewers responses for the same citation. A single, final response would be saved in the .csv output file.
We suggest using simple yes/no questions that follows a standard framework e.g., Patient or Problem, Intervention, Comparison and Outcome (PICO), or Population, Concept, and Context (PCC).
However, you can set up any questions you want to support your decision to include or exclude a reference. You can also set up a “flagging” question such as “Flag this reference to review later?”
We recommend screening 100 initial citations and then noting your exclusion numbers and when you have reached 100 successive excludes, it is highly likely that all relevant documents meeting your criteria have been identifies. For more information, please refer to: Cormack GV, Grossman MR. Technology-assisted review in empirical medicine: Waterloo participation in CLEF eHealth, 2018. Available: here.
Yes. Your uploaded files are processed temporarily during your session and are not stored permanently. We do not share your data with third parties without your explicit consent. For more information, please see our Privacy Policy.
At this time, the tool is optimized specifically for PubMed PMID .txt files. Expanding support for other databases is a potential future enhancement.
If you experience any issues with uploading files, navigating the tool, or other technical challenges, please contact our support team via the contact form.