Supervised Extended Iterative Optimization Technology for Estimation of Powder Compositions in Pharmaceutical Applications: Method and Lifecycle Management

Further enhancements to the extended iterative optimization technology (EIOT) method are presented. These are meant to include the effects of nonchemical interferences of known origin in the EIOT parameters and to manage the lifecycle of the method in a pharmaceutical manufacturing application. A derivation of EIOT using projection to latent structures (PLS) is also presented and shown to be mathematically equivalent to the EIOT model. A structured methodology to apply the method is proposed and illustrated with a real pharmaceutical case study. Two diagnostics are proposed to diagnose the performance of an EIOT model in realtime. The M and the SSR are described and shown to be mathematical equivalences to the established Hotelling's T-2 and the orthogonal distance to the model plane (or SPE), used in a traditional solution based on a PLS model.