Development of software for predictive modeling of polymeric nanoparticles of pharmaceutical interest
nanoprecipitation; polymeric nanocapsule; data mining; data science; machine learning
Nanoparticles, designed on the order of billionths of a meter, may have a crucial therapeutic function for patient survival under certain pathological conditions. However, some obstacles in reproducing or controlling their production variables prevent their systematic use. Data science, which involves methods and processes to study data in order to gain new information and discoveries relevant to business and research, has been used to create models in the field of nanotechnology, but with limited success. With this in mind, this thesis aims to discuss state-of-the-art data science methods and the state of the art of machine learning in the context of polymeric nanoparticles and to develop predictive software for modeling nanoparticles of pharmaceutical interest. Of the various published studies, only a few used a machine learning approach, and even then, methodological and data interpretation errors were found in these studies. Most of the problems are related to the fact that appropriate analytical steps were not performed, such as cross-validation, data matching or testing of alternative models. Another recurring problem that is difficult to solve is the insufficient number of experiments related to the high investment required. In this way, the appropriate use of study protocols in data science can actually help to obtain reliable models that can be used to solve real-world problems.