Skip to Content (custom)

Case Study

Domestic energy corporation reduces discovery cost and time through the use of AI-based Continuous Active Learning (CAL)

  • Energy

client need

Reduce the cost and time required to comply with an urgent discovery request

A North American energy company was faced with a quick turn review that consisted of 369,374 documents after search terms and date filters.

why Epiq?

Epiq’s experienced Advanced Technologies and Document Review Services teams had previously successfully deployed Continuous Active Learning on numerous projects yielding significant cost and time savings while accurately identifying relevant documents for production. Continuous Active Learning is a predictive coding solution that allows for immediate review of documents most likely to be responsive while continually learning and updating the model as new documents are coded.

Epiq solution

AI-based Continuous Active Learning review prioritization and culling

Utilizing a bespoke Continuous Active Learning workflow and embedded NexLP AI technology, Epiq was able to identify 20,652 documents responsive to the discovery request while only reviewing 18.43% of the total review population. The prioritized workflow achieved a cumulative responsive rate of 31.13%, nearly 4 times greater than the estimated prevalence of the entire population (8.38%).

Validation testing resulted in an elusion rate between 0.00% to 0.74% with a 95% confidence level from of the null set of documents culled from review. This confirmed that there were very few documents left unreviewed that were responsive to the discovery request. With this data the client was comfortable ending the project, saving weeks of review and hundreds of thousands of dollars.

Back