The chemometric analysis of point and dynamic data in pharmaceutical and biotech production (PAT) — some objectives and approaches

Checking that a process is doing what it is supposed to is of critical importance in manufacturing, economics, environmental monitoring, patient monitoring, and more. Given sufficient and adequate analytical and process measurements made during the history of the well functioning process, a multivariate model of the process variation around a multivariate dynamic trajectory will, in principle, form a good basis for this checking. Such systems are often labeled process monitoring, real-time quality control (RTQC), PAT level 4, and advanced process control/fault detection and classification (APC/FDC). Here PAT stands for process analytical technology indicating the reliance on adequate and multiple data for this checking. ctice, there are many difficulties in making an RTQC/PAT-4 system work well. Starting from an industrial example, the problems of constructing and implementing a well working checking system are discussed in relation to its different parts — analytical and process data, chemometrical and other methods for their modeling and analysis, and various forms of data management to handle the data flow and synchronization, as well as storage and retrieval. The display and interpretability of diagnostics and results are emphasized.