Adding precision and speed to your reviews
We’ve been exploring how Machine Learning and Automation can help us alleviate pain points in the Systematic Review journey whilst boosting productivity and improving user experience.
We are approaching machine learning in a considered, tested and measured way to implement automation appropriately throughout our tool whilst ensuring our users remain in full control.
One of the enhancements we are making with this methodology is to use automation to remove references not reporting on randomized controlled trials (RCT) before manual screening takes place. This is targeted specifically at research teams completing a review in the medical and health science research area, who want to only include RCTs in their review.
We’ve taken this step into championing Machine Learning and Automation to help users save time by reducing the volume of references to manually screen by up to 45%.
Our new feature automatically removes references considered as not reporting on RCTs by utilizing the Cochrane RCT classifier. The same classifier is currently being used in the Covidence product to automatically tag whether references are possibly an RCT.
Keeping you in control
We want to ensure that users can take full advantage of our Machine Learning enhancements to boost productivity whilst remaining in complete control of their experience. As such, the new automation feature is optional and is disabled by default. Users are given the opportunity to review the references automatically removed, and move references back into the manual screening process should they wish. Users can disable the feature at any time, resulting in all references moving into manual screening.
You can find out more information about this new feature by watching our demo video below:
Quality and Efficiency will remain uncompromised
Ensuring the highest levels of quality and efficiency during the systematic review process is paramount for us at Covidence. By using the Cochrane RCT classifier, we are using a tool with a recall performance of 99.5% and is endorsed by Cochrane, the world’s leading and most trusted organization in systematic reviews. The Cochrane RCT classifier has been vigorously evaluated and tested, with the published results showing the classifier can reduce manual study identification workload for medical and health science reviews with a very low and acceptable risk of missing eligible RCTs. Paired with the users’ ability to review references automatically removed, our new feature sets an incredibly high standard for quality and effectiveness.
Our solution is also aligned with PRISMA 2020, showing automatically excluded references correctly in the PRISMA diagram.
How can you take advantage early?
We’ve made it as easy as possible for you to take advantage of this new feature. If you have an account with us, simply sign in and create a new review (or edit an existing review). Then, on the ‘Start a new review’ setup page, simply scroll down to ‘Automation options’ and tick the box to enable the feature.
You can also enable this feature by heading to the ‘Edit Review Settings’ page and selecting the feature in the ‘Automation Options’ section underneath ‘Area of research’.
Please note this feature will only appear if you have indicated that you are in the medical and health science research field.
To get started, sign in here.
Or if you’re new to Covidence, click here to try it for free!
