how to use


The system is designed as a shared space for humans and machines to work together. If you are a human, the best way to start with the system is to select one of the examples of systematic reviews on the home page. These are examples that are already in the system and for which we are relatively confident that we have identified all of the registered trials that were included in the systematic review. If you are a machine, then you probably already know why you are here.

If you haven’t already registered as a user, then email us, answer a few questions, and we may provide you with access to the full functionality of the website and access to the underlying data. Remember that if you register with the system and start to add or confirm trials that are included in systematic reviews or are relevant to their updates, then you are helping us to train our software agents to be smarter and better at predicting which new trials may be relevant to already-published systematic reviews.

For registered systematic reviewers…

If you want to monitor the accumulation of evidence for a systematic review you have published, then you can simply copy and paste the PubMed identifier for your systematic review into the search box at the top of this page. If it is not already in our database, then the system will add it to the database and then look for the set of trials that were included in the review and have registrations in ClinicalTrials.gov.

Check through the list of included studies (on the left) to see if the system has missed any (or incorrectly added published trials that were not included), and confirm that they are all present. As you add trials, other software agents will start to vote up or down on other trials with registrations in ClinicalTrials.gov to recommend trials that might be included in an update.

Once you are satisfied that all of the trials included in the published systematic review are present, you can look through the details of the similar trials (on the right) and vote up or down on those that are likely or unlikely to be relevant. If there are trials that you can’t immediately see in the list but you know are likely to be relevant, you can add them manually.

After voting on relevant trials recommended by the system, you can ask the system to add up the number of participants across included and relevant trials to check how much of the currently (or soon-to-be) available evidence the published systematic review covers.

For registered researchers...

We also provide up-to-date access to all available links between published systematic reviews and included trial registrations, and scores on trials that may be relevant to an update—all as a single large sparse matrix. While the matrix is incomplete and imperfect right now, we hope it will function as a shared resource for researchers in the area to help us all improve machine learning tools used in trial screening as well as tools used to support the prioritisation of systematic review updates where there is a risk of a change in conclusion.

For clinicians and consumers…

The trial2rev system is also designed to help you quickly estimate if a published review might be outdated. We have included an early version of a tool that estimates the risk that a systematic review would change its conclusions if it was updated based on evidence from trials that have recently been completed.

Please note that the trial2rev system is provided as-is and we make no guarantees about the accuracy of the data or any of the predictions. It is not intended to replace expert guidance on the current value or credibility of any systematic review.

If you are interested in how the interface works, want to examine one or more of the software agents we use to fetch new data, or want to improve on the machine learning methods we use to train our other intelligent agents, check the information page or visit our GitHub project page.