Weave Bio and Takeda have expanded their collaboration to deploy AI-based regulatory automation across 14 active drug development programs in multiple global markets, targeting administrative bottlenecks associated with regulatory dossier preparation.
The collaboration initially focused on automated Investigational New Drug application module drafting and has since expanded to Health Authority Question (HAQ) intake and Response to Question (RTQ) workflows. The platform extracts information, cross-references source materials and produces initial drafts while allowing regulatory professionals to verify claims, make edits and approve final submissions.
A study co-authored by researchers from Weave Bio and Takeda and published on arXiv reported several operational findings from the deployment. Drafting time for nonclinical written summaries in eCTD Module 2 decreased from approximately 100 hours to three hours, representing a 97% reduction in initial drafting time. This contributed to a reported 60% compression in overall preparation timelines.
The platform also achieved 95% fidelity to source nonclinical study reports in eCTD Module 4. Across an evaluation dataset containing more than 100 regulatory questions, the system recorded no data extraction errors and achieved an average answer quality score of 68%, compared with an internal target range of 50% to 70%.
Weave Bio’s system is designed to provide traceability between generated regulatory content and underlying evidence. Each generated sentence and numerical table can be mapped directly to its source paragraph in study reports, enabling regulatory teams to verify references during review and maintain data lineage.
The deployment comes as regulators increase their focus on the governance of artificial intelligence in drug development. FDA draft guidance and joint FDA and European Medicines Agency principles have emphasized human accountability when AI is incorporated into drug development and regulatory processes.
Within the Weave Bio and Takeda workflow, AI therefore functions as an assistive technology rather than an autonomous decision-maker. Regulatory teams remain responsible for reviewing extracted information, validating generated content and providing final approval before submissions are delivered to health authorities.
Click here for the original news story.
