Health Agency Uses AI to Reduce COVID-19 Citation Screening and Data Extraction By 50%

This national agency conducts systematic reviews to inform public health policies, including conducting critical COVID-19 research.

Monitoring an Exponentially-Growing Volume of COVID-19 Literature

Back in January 2020, this health agency was already closely monitoring the alarming spread of COVID-19 worldwide. Since then, there has been an exponential growth of new scientific literature and a constant influx of new information about the disease and how to treat it.

Led by an epidemiologist, the health agency’s team is tasked with monitoring the COVID-19 situation and producing extensive reports that are the backbone of their country’s response to the pandemic.

The team reviewed an average of 1200 citations per day and extracted pertinent information from approximately 50% of them per day. This is in addition to the non-COVID-related activities the team undertakes normally. Without DistillerSR’s AI capabilities, the health agency would not be able to effectively and quickly screen the large volume of incoming new citations they received daily.

Cutting Reference Screen Time by 50%

The number of citations the organization manages every day has more than doubled since COVID-19 became a pandemic.

Manually screening and extracting data from references takes a human reviewer up to 10 minutes per citation. DistillerSR, however, has reduced screening times to 3-5 minutes per citation. As a result, the health agency’s reviewers can complete their work in 40 minutes to 1.5 hours on average, when in the past it could take approximately 3 hours per day.

By leveraging multiple AI classifiers to automatically triage citations into two research streams, the health agency was able to effectively eliminate a majority of their project management and workload assignment overhead. This meant they were able to easily build and maintain a comprehensive database on COVID-19 literature, which supported numerous concurrent projects affecting national-level healthcare decisions.

DistillerSR’s AI classifiers are algorithms that learn to recognize similarities between citations placed in the same category by humans. These learnings are then used to identify references that belong in that same category.

Without a tool like DistillerSR, the organization would require more resources, including more time spent on project coordination to manage distributed teams of volunteers, and a growing volume of new citations. By leveraging DistillerSR’s AI capabilities, the team can meet urgent update request received from a variety of sources, including the media and prominent public health decision-makers.

Helping Employees Do More Reviews Faster

AI-enabled automation has made it possible for the organization to stay up-to-date with growing volumes of new COVID-19 scientific literature. In addition, it makes updating evergreen reviews or umbrella reviews much easier, while the automatic audit trail and reporting means they can keep tabs on everything that is done in the project.

The health agency’s project lead valued DistillerSR’s secure database and workflow. With DistillerSR, the health agency kept tabs on how everything is evolving among distributed reviewers, many of whom are volunteers.

To manage the sheer volume of COVID-19 citations, the public agency leveraged DistillerSR’s AI capabilities. The experience is helping the team consider a broader range of planning and best practices to implement moving forward.