What happened
Biomet faced the claims of hundreds of plaintiffs over its M2a Magnum hip implant, gathered in one federal docket in Indiana. Keyword culling cut 19.5 million documents to 3.9 million, and removing duplicates left 2.5 million. Biomet then used a predictive coding service that the order calls Axelerate, with eight contract attorneys, to choose documents to produce from the 2.5 million. The plaintiffs' committee asked the court to make Biomet restart with predictive coding over all 19.5 million documents. On 18 April 2013 Judge Robert L. Miller Jr. found that Biomet's process complied fully with Rules 26(b) and 34(b)(2). He wrote that the question before him was not whether predictive coding beats keyword searching first. He held that the likely benefit of a restart, at a cost in the low seven figures, did not equal its burden. Biomet's e-discovery costs stood at about $1.07 million and would total $2 million to $3.25 million. Biomet's sampling tests concern the documents that keyword culling left out, and the court did not test the predictive coding.
Sources
-
Court document, by authors with no tie to the system.
What Biomet has done complies fully with the requirements of Federal Rules of Civil Procedure 26(b) and 34(b)(2). I don't see anything inconsistent with the Seventh Circuit Principles Relating to the Discovery of Electronically Stored Information.
Details
AI system
- Workflow
-
Keyword culling cut 19.5 million documents to 3.9 million, and removing duplicates left 2.5 million. Eight contract attorneys worked with the predictive coding service on a sample of the 2.5 million documents in one round of find-more-like-this training. Contract attorneys then reviewed the chosen documents for relevancy, confidentiality and privilege.
Follow-up
- Conditions for replication
-
Random samples show that few responsive documents remain among those that keyword culling left out. Biomet stays open to more targeted search terms and produces the non-privileged documents in its statistical sample. The requesting party pays for any restarted processing and predictive coding that it demands.
Entry details
- Sector
-
Legal services / e-discovery.
- Purpose
-
Recommind built Axcelerate, which the order spells Axelerate, to search and review large document collections with predictive coding that learns from reviewers' choices.
31 more questions not yet checked
model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Court upholds Biomet's predictive coding review under discovery rules (AIBD-0982). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-0982.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)