University of Cambridge research arm develops AI digital twin for pharma manufacturing

The Cambridge Advanced Research and Education in Singapore (CARES), together with the A*Star Institute for Infocomm Research (A*Star), has developed a digital twin platform that applies artificial intelligence (AI) and real-time plant data to improve fault detection, system monitoring, and predictive maintenance in pharmaceutical manufacturing.

The system will be commercialized through Chemical Data Intelligence, a CARES spin-off, and deployed within the Pharma Innovation Program Singapore (PIPS) consortium, according to the organizations. CARES led the development of the platform’s ontology and physical modeling, while A*Star created an AI agent designed to detect anomalies. The agent reportedly integrates physical models with data-driven models trained on sensory data such as temperature, flow rates, and pressure to identify potential issues like mismatched flows or abnormal tank levels before they escalate.

The platform was demonstrated on a real-time testbed provided by Accelerated Materials, another CARES spin-off, and is hosted on Microsoft Azure cloud. Researchers and engineers can use the system to monitor plant performance, simulate production scenarios, and test responses to potential faults.

Lianlian Jiang, co-lead principal investigator and unit lead of Digital & Sustainable Manufacturing at A*Star, said the AI agent could also be extended to support quality monitoring, production scheduling, and resource planning.