Problem
Core equipment in injectable drug manufacturing generates process data continuously and at volume, but there was no system to collect and use it in real time. Data accumulated without being analyzed, making early detection of equipment anomalies difficult.
Data collected
High-volume process data from core equipment was collected in real time. Given the nature of pharmaceutical processes, the collection path was designed so that when and where data was captured remains traceable.
Modules applied
The B²LAB big data platform provided the collection, storage and monitoring foundation, with an anomaly detection model built on top.
Result
A system for collecting and storing high-volume process data in real time is in place, and anomalies in core equipment can be detected early.
Constraints specific to pharmaceutical processes
Regulatory fit comes first when introducing a system into a pharmaceutical site. It must be possible to record when data was collected, by whom, and how it changed.
Adding an AI model does not soften that. If anything, because you must be able to explain where the model’s input came from and how it was processed, designing for data lineage matters more.
Handling high-volume data
Data from core equipment is large in both volume and velocity. If the collection and storage structure is shaky, the analysis above it means little. So B²LAB’s collection, storage and monitoring layers were established first, and only once the pipeline ran reliably was the anomaly detection model added.
This case is documented in terms of scope and applied technology. Quantitative results will be added once client disclosure is confirmed.