AeronauticsARTIFICIAL INTELLIGENCE
37 days a year, given back: AI accelerator ends manual re-annotation of clinical trial documents
Aixial Group, an ALTEN company, built an AI-powered accelerator for a leading pharmaceutical company, automating the transfer of annotations between versions of Case Report Forms (CRFs) and cutting a recurring clinical documentation task from days to hours.
In clinical research, Case Report Forms are among the key documents required for submissions to regulatory agencies such as the U.S. Food and Drug Administration (FDA). CRF captures how patient data is collected and how drug side effects are tracked, and contains annotations following precise, regulation-driven codes including specific colours, line types, arrows and positions. The annotated CRF is itself a regulatory document: every version must be submitted to health authorities, making accuracy and consistency a compliance obligation, not simply good practice.
Pharmaceutical companies typically choose to annotate CRFs early in the lifecycle of their clinical studies. However, every time a CRF is revised, even for minor changes in a document spanning dozens of pages, teams have traditionally had to re-read the entire document, identify what changed, and manually reapply all annotations from the previous version before adding new ones. To address this recurring and time-consuming challenge, a leading pharmaceutical company engaged Aixial Group for support.
Challenge: Eliminate the need to systematically manually re-read and re-annotate entire Case Report Forms every time a new version is issued, even when only a small number of changes have been made.
Solutions: An AI-powered accelerator that compares CRF versions, automatically transfers existing annotations to their correct position in the new version, and leaves only genuinely new content open for annotation.
Benefits:
- Cost, quality and delay:
- 90% cut in processing time for reusing annotations from previous CRF versions.
- Ressources: effort reduced from two reviewers to one
- Very limited risk on quality due to human error as copy/paste operations are automated
- Rapid innovation, with the accelerator built and delivered in a single day.
- Immediate ROI, developed at zero additional cost with no internal billing.
- Scalable and adaptable to CRFs of different formats and lengths, across all trial phases and therapeutic areas.
- Deployable across organisations of any size, from large pharmaceutical companies to small biotechs and institutional players.
- The accelerator can be reused by any Aixial team in their day-to-day work, extending its value well beyond a single client engagement.
From re-annotating everything to annotating only what’s new
Aixial’s accelerator compares each new CRF version against the previous one, automatically detecting what has changed and what has stayed the same. Where content is unchanged, it transfers the corresponding annotations directly into the new document, placing them precisely, respecting available space and preserving the original annotation typology, whether that means specific colours, line styles, arrows or positions. Reviewers are then left to focus solely on annotating what is genuinely new, rather than re-annotating an entire document from scratch. Because the transfer logic proved reliable and repeatable, the client no longer needs a second reviewer to check and correct the work, reducing validation effort while maintaining annotation quality.
Time saved, better spent
By automating the reuse of annotations across CRF versions, the accelerator reduced the client’s annual workload on this activity from 41 days to just 4, a saving of 37 days a year, while cutting processing time for version updates by 90%. Built by Aixial’s team in a single day at no additional cost, the solution delivered a return on investment from day one. This gain in speed also changes the nature of the work itself: because each version update now takes hours rather than days, CRFs can be revised as often as protocol changes require, without the cost of each iteration weighing on timelines. The time this frees up isn’t just saved, it’s reinvested in higher-value clinical work rather than repetitive re-annotation.
Its success has also opened the door to further gains: the same approach is now being extended to trial design automation, using DocInsight, and soon Aixial’s own language model, to help synthesise clinical trial protocols, cutting that task’s turnaround from around three days to as little as two to five hours per study while improving accuracy and consistency in data extraction.
The power of AI has enabled this pharmaceutical company to redirect nearly six working weeks a year from repetitive re-annotation to higher-value clinical documentation work.
