(652a) Model Based Multivariate Monitoring of Pharmaceutical Processes

Multivariate statistical process control (MSPC) was introduced in the early 90’s as a tool to monitor a process that “routinely collect hundreds to thousands of pieces of data from a multitude of plant sensors every few seconds” [1]. Since its introduction, the available software technology to perform MSPC has matured from in-house custom code to available over the shelf.

In pharmaceutical manufacturing the use of MSPC has flourished in monitoring large molecule operations such as bioreactors and downstream purification processes. These systems are mostly implemented as “preventive engineering controls” and used by operations personnel to take early action before an event can cause a deviation in a batch. In small molecule and drug product operations however, these proven tools are not as commonly found. This talk explores the barriers found to the adoption of MSPC in drug product pharmaceutical operations and provides some alternatives to potentially remove these barriers and find a justifiable business case to their implementation.

[1] Kresta, J.V., Macgregor, J.F. and Marlin, T.E., 1991. Multivariate statistical monitoring of process operating performance. The Canadian journal of chemical engineering, 69(1), pp.35-47.