Overcoming Solubility Challenges PDF Download

Solid dosage forms, such as tablets, capsules, and powder, are one of the most widely used administration methods for drug delivery. However, scientists and researchers are facing various challenges when developing them. During the formulation development process, critical quality attributes (CQAs) of a drug product must be carefully controlled, and they are typically influenced by critical material attributes (CMAs) and critical processing parameters (CPPs). A conventional development pathway for solid dosage forms utilizes a trial-and-error approach, which requires a large amount of time and a laborious workload. Several emerging technologies, including Artificial Intelligence (AI) and 3D printing, have been extensively applied to develop solid dosage forms. As a subfield of AI, machine learning has gained more attention in pharmaceutical industries and exhibits numerous benefits to overcome the challenges during formulation development. In Chapter 1, the applications of different machine learning techniques in solid dosage forms were reviewed, which provides general guidance to formulation scientists. In Chapter 2, we investigated the applications of machine learning to design inhaled dry powder prepared by thin-film-freezing (TFF) technology. Aerosol performance, which can be indicated by fine particle fraction and mass median aerodynamic diameter, is one of the most important factors when developing inhaled dry powder. In this chapter, we first obtained a tabular dataset containing formulation information and processing condition, and scanning electron microscopy (SEM) images by literature mining and in-house experiments. Then, we applied multiple machine learning algorithms to predict aerosol performance. Poor water solubility is another critical challenge for drugs in development pipelines and commercial products, and it may lead to low absorption and bioavailability. Amorphization is a intermolecular modification method to convert crystalline drug substances into amorphous states. In Chapter 3, we developed a machine learning-based method to predict the glass forming ability of pharmaceutical compounds, which may potentially facilitate the in-silico screening process of amorphous drugs. In addition, hot-melt extrusion (HME) is one of the most widely used methods to prepare amorphous solid dispersions (ASDs), but the energy input, including thermal energy and specific mechanical energy, needs to be carefully controlled to prevent residual crystallinity and chemical degradation. In Chapter 4, we investigated the feasibility of predicting the forming of chemically stable ASDs by using machine learning modeling. Moreover, the trained models achieved high accuracies and relatively good interpretability