No abstract

No abstract

The development, validation, and maintenance of spectral libraries for raw materials identification based on near‐infrared spectroscopy (NIRS) are presented in this article. A pharmaceutical example is used to describe concepts and general criteria about the whole lifecycle of these classification methods. The process is sequentially presented in two parts. The first one involves the development and validation of a first version of a library, using an initial number of raw materials and samples. This section exposes the preparation of the data sets, the selection of the discrimination method, the configuration of the data pretreatment, and the evaluation of the acceptance criteria along with internal and external validation processes. The second part describes the sequential updating and revalidation of the library. It is based on a second and a third versions of the method, which are generated after the introduction of new raw materials and new samples of original materials over two different periods. This part of the article includes further considerations that must be taken into account to complete a proper maintenance of the library. In this section, the qualification through the construction of cascading libraries and the renewal of the model by means of calibration sets update are presented. A final scheme of the general strategy attempts to summarize the main steps during the calculation, validation, and maintenance of this qualitative method. Current recommendations of the European Medicine Agency (EMA) and the European Pharmacopoeia (Ph. Eur.) about this topic are considered throughout the entire article.