Репозиторий Университета

© 2018 Counterfeit medicines of ‘high quality’ are the most difficult to detect as they have the same chemical composition as the genuine ones, but they are produced by underground manufacturers who violate technological regulations. Our approach is to consider a remedy as a whole object, taking into account the complex composition of APIs, excipients and manufacturing conditions. For rapid testing, the Near Infrared (NIR)-based approach is applied. It entails the acquisition of NIR spectra and processing of the collected data using a modern one-class classifier method called data driven soft independent modeling by class analogy (DD-SIMCA). We present an exemplary analysis of the suspected drugs, which have the same designation and a very similar chemical composition to the brand of a widely used medication used to treat allergies. We recognized the counterfeits using a model that had been previously developed and stored in a library for everyday monitoring in drugstores. We also describe the steps taken in development and validation of DD-SIMCA library models. In the case under consideration, the NIR-based analysis reveals 100% of counterfeits, and this result surpasses the results of the routine compedial tests. Additionally, we present a new instrument, VisCam, that is used in visual analysis of the primary and secondary packages. This instrument combines a tenfold web-camera with different light sources. It is shown that VisCam helps to reveal hidden violations in the primary and secondary packages.

© 2018 The Author(s). The rapid development of graphics processing units (GPU) allowed to use in practice resource-intensive methods of artificial intelligence such as neural networkS. At present, the authors got preliminary results on detection of some forms of leukocyteS. The article presents the first results of the work.