Announcing Deep Dynamics’ Extract: literature reviews, powered by machine learning
With Extract, reference screening is reduced with up to 85.9% of its original size. Machine learning enables to classify references (with a score) from highly relevant to not relevant. Reviewers are always in control as they decide whether to exclude or include a reference.
First, screen a part of the references manually (average 43.9%), to train the software. Relevant references are prioritized. Then, check the results and classify the remaining references manually, in bulk. In this way, the reviewer is always in control. Only highly irrelevant references can be classified in bulk and no single reference is automatically excluded or classified.
Reviewing with Extract, means higher quality. Human retrieval rates (recall) are 92% (Edwards et al.). With Extract, 98% recall was found (SD 2,41%). Tested on Cochrane Diagnostic Test Accuracy Reviews (Cormack et al.). Considered the most challenging to classify with software.
Screening references with Extract, enables higher quality than fully manual reviews (single reviewer)
The reviewer is always in control, as he/she decides how to classify each reference, in the results section
Machine learning is here. Be one of the first to try this exciting technology, without compromising quality
Use as a second reviewer, and save on average two days on twelve days abstract screening.
Small number of users (1-5), low monthly cost. High flexibility and effectiveness.
Higher number of users (6-24), very reasonable cost structure. Great flexibility with low fixed costs.
Highest number of users (25 or more). Highest discount. Low fixed cost and high usage still possible.
Send an email to request a free trial (two weeks). Or plan a live demonstration.
Be among the first to harness the power of machine learning and gain real advantage. Finish literature reviews earlier and enjoy a competitive advantage on execution speed. Perform literature reviews with lower cost, and charge a lower price.