Back to all cases How Haleon makes clinical research data accessible with AI
Haleon has extensive current and historical clinical study data, but it was fragmented and stored inconsistently. Arie Weeren, a consultant at BigData Republic, used AI to standardise the data and consolidate it in Databricks. The result is Cliniverse: one central environment that makes research data accessible to multiple researchers.
The challenge
Haleon's historical and current clinical study data was stored in fragmented SAS files. Inconsistencies between studies made cross-comparisons virtually impossible, while conversion to a single unified Common Data Model (CDM) was essential for R&D.
The solution
All SAS files were converted into tables in a Databricks Unity Catalog. Since manual conversion of all this data was unfeasible, processing was automated using AI technology such as Large Language Models and vector search.
The result
- Cliniverse was successfully delivered as an internal R&D product
- All studies and metadata are consolidated in Databricks Unity Catalog and linked to the Sonar registry
- An initial dashboard tracking participant diversity is live
- Turnaround time for setting up overview studies is significantly shorter