Castor Catalyst: first to market to show industry-leading data extraction accuracy in a sponsor-audit-friendly approach

Castor Catalyst reads data straight from source records under human review. Reviewers accepted 99.16% in a Top-10 BioPharma study, now live in five countries.

NEW YORK, NY, UNITED STATES, September 3, 2026 /EINPresswire.com/ — Castor today announced a milestone for Castor Catalyst, its AI-driven clinical data platform that extracts data directly from source records under human review. In a recent study for a Top-10 BioPharma sponsor, reviewers accepted 99.16 percent of the values Catalyst extracted, overriding 0.78 percent.

Catalyst is live in production across five countries, in studies spanning real-world evidence, post-market clinical follow-up (PMCF), and pivotal medical device trials. Instead of teams re-keying source documents into their EDC by hand, Catalyst reads the source, extracts the data, and presents it in a reviewer-friendly interface that keeps the original context. The reviewer keeps, corrects, or rejects each value before it enters the clinical record.

Extracting data directly from source records with AI is not widely used in regulated research, where every value has to withstand audit. Catalyst uses a Google Gemini model with data residency in the United States and the European Union. Personal data is redacted through Catalyst’s proprietary localized pseudonymization before anything is processed. Castor operates under zero data retention, and no client data is used to train models. Catalyst is GxP compliant, designed to meet ICH E6(R3) GCP, 21 CFR Part 11, and EU Annex 11, with a full immutable audit trail and mandatory human review.

In that Top-10 BioPharma study, reviewers assessed 14,290 data points Catalyst extracted from 285 source files across 158 participants. They accepted 99.16 percent as extracted and overrode 0.78 percent, submitting 14,282 values to the clinical database with no upload errors. Each value was shown to a reviewer against its source before submission, so the low override rate reflects review, not unchecked pass-through. In internal EMR benchmarking of site-based chart review, Catalyst reduced the work from 39 minutes to 6 minutes per chart.

“The need for automation in clinical trials is greater than ever, with budget pressure and staff burnout more common than ever. Yet using AI in clinical trials, especially in day-to-day clinical operations, is not an easy task. With Catalyst, I believe we have found the right blend of compliant, auditable AI combined with human oversight. Seeing AI actually make an impact on real clinical operations is something we are really proud of.”
Derk Arts, Founder and CEO, Castor

“Using AI to enter data directly from source records is not widely utilized for regulated research outside of recruitment, because it has to clear a high bar. When reviewers accept more than 99 percent of what the AI extracts, across studies in five countries, it shows the approach works and holds up to the standards that clinical trials, real-world evidence, and med tech all demand. The demand we are seeing tells us research teams have been waiting for this, real speed with a human in charge.”
Lisa Charlton, Chief Product Officer, Castor

About Castor
Castor is a clinical trial data platform for medical device, biopharma, diagnostics, and academic research teams. From site-based studies to fully decentralized, direct-to-patient designs, Castor unifies EDC, eCOA and ePRO, eConsent, data management, and Catalyst AI in one compliant platform. Learn more at castoredc.com, or explore Catalyst.

Kevin Thornton
Castor
kevin.thornton@castoredc.com
Visit us on social media:
LinkedIn
YouTube

Media gallery