iTromsø · IBM, VC2
10 minutes per day — time spent reviewing potential stories
Independently reportedMedia & entertainment · Document processing · NLP · In production
Document processing
Bayer Consumer Health · Acodis
Vendor-reportedPharma & biotechIn productionEnterpriseNLP
Vendor-reported. The customer is named and the numbers are quoted from the source page, but the account comes from the vendor. No independent confirmation.
Reported by acodis.io (vendor self-report). Checked against the source page on 2026-09-06. 2 of 3 figures below appear on that page word for word; the rest are marked.
The problem
Raw data from historic clinical studies was trapped in scanned PDF copies, making it inaccessible for meta-analysis and further exploration.
What was deployed
- Vendor
- Acodis
- Products
- Acodis data extraction solution, Acodis table recognition solution
- Technique
- NLP
- Build or buy
- Bought off the shelf
- Deployment
- SaaS
- Data used
- scanned PDF copies of historic study reports, listings in the appendices of integrated study reports, clinical trial data
- Scale
- over 3500 pages of documents with 200 complex tables
What changed
Each row is quoted from the source. Figures we could not find on the page in those words are marked — they are kept, not deleted, so you can judge them.
over 3500Pages of documents processed
“With the Acodis solution, the team extracted data from over 3500 pages of documents with complex 200 tables in them.”
Quoted word for word from the source
200Tables processed
“Data extraction of over 200 tables, which ranged in formatting and length in over 3500 pages of documents”
Not found on the source page in these words
3 weeksProject completion time
“The speedy and easy-to-use solution meant that the entire project was done in 3 weeks.”
Quoted word for word from the source
good-quality data extractedData quality
“The use of Acodis within the project allowed for the ability to extract good-quality data that was otherwise trapped in unusable formats.”
Quoted word for word from the source
analyzableData accessibility and analysisfrom inaccessible
“As a result, the team could analyze high volume of data that was previously inaccessible, provide additional business insights”
Not found on the source page in these words
ROI as stated: Without the new technology, this would not have been possible within reasonable costs and timelines.
“By extracting raw data from scanned PDF copies of historic study reports, we were able to re-use the clinical data and conduct further exploration and analysis by combining datasets from various studies. Without the new technology, this would not have been possible within reasonable costs and timelines.”
Difficulties and limits
- much of the clinical trial data was stored in unstructured data formats
- data was not readily available or accessible for analysis
- required extraction from old, scanned copies of PDF documents
- high-quality extraction from multiple different tabular formats
- some tables had differing (or lack of) border and row structuring, and/ or a high level of skewness due to the scanning
- some scanned PDF files were of poor quality, with noisy data
- limitations with the scalability and flexibility of a solution that allowed for consistent data extraction of over 200 tables, which ranged in formatting and length
Sources
- acodis.io — vendor self-report
Similar deployments
Amplitude · Wolfia
4X — throughput of security questionnaires
Vendor-reportedTechnology & software · Document processing · AI agents · Scaled
Adyen · Rossum
70 — automation rate
Vendor-reportedBanking & finance · Document processing · NLP · Scaled · 2023
Skaylink · Moterra
dropped by around 80% — NIS2 preparation time
Vendor-reportedTechnology & software · Document processing · LLM · Scaled
OM Pharma · DrugCard
3 minutes — review time per article
Vendor-reportedPharma & biotech · Document processing · NLP · In production
AMPREP (Amalgamations Repco) · Cimplify.AI
100% — invoice automation
Vendor-reportedManufacturing · Document processing · NLP · In production