Search for: [Abstract = "Recommendations of the US Food and Drug Administration regarding Process Analytical Technology, Quality by Design as well as recently introduced model\-informed drug development program, indicate the need to better understanding of factors affecting final product quality. All initiatives could be realized by the development of proper models, which could be applied at product development stage ass well as quality control at the production stage. Scientific assumptions of the doctoral dissertation concern the development of tools supporting research and development of inhaled drugs. They were implemented by constructing empirical models, capable of predicting the size of particles composed lactic and glycolic acid copolymers and particles deposition in vitro of carrier\-based formulations intended for inhalation. Additionally, models were applied to optimize the features of carriers particles in\-silico. The databases were compiled on the basis of literature data and the results of laboratory experiments carried out at the School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore, which were delivered as part of the international research project “Delivery of Protein and peptide drugs through dry powder inhalation”, 2\/3\/POL\-SIN\/2012. Active pharmaceutical ingredients, as well as excipients included in the formulations, were encoded using molecular descriptors calculated using Marvin program developed by ChemAxon. Besides, an analysis of scanning electron microscope images was used to represent the carrier particles surface roughness. For this purpose, ImageJ program with an additional SurfCharJ package was applied. In total, three databases were constructed for the needs of further computational work. The computational work was carried out using the R environment. The first stage of data analysis was a selection of crucial variables and a reduction of the input vector, which was made using the fscaret package. The following work included models construction and testing, including artificial neural networks, Random Forest, fuzzy systems, rule\-based systems, and mathematical equations created with genetic programming methods. Among the developed models, the best predictions were obtained using genetic programming methods. The predictive capabilities of the best performing models were additionally tested on new experimental data obtained from the literature and research center in Singapore. The model for predicting particles deposition in vitro was also applied to optimize carrier particle features using own\-written computer programme. Proposed directions of carrier modifications were verified by the research center in Singapore."]

Number of results: 0