The rheological properties of natural polymer solutions are difficult to be modeled universally because of the strongly nonlinear relations between viscosities and the external factors and the discreteness of the rheological data owning to different molecular parameters including the molecular weights and size of clusters from different types of natural polymers and solvents. In this study, a typical natural polymer-lignin was selected and dissolved in polyethylene glycol (PEG). The rheological properties of different PEG-lignin solutions (PEG-Ls) and the molecular parameters of the pretreated lignin were tested. Subsequently, machine learning was applied to establish the generalized models considering the molecular parameters. The models were successfully developed in Newtonian and non-Newtonian regimes for PEG-Ls with correlation coefficients of 0.982 and 0.980, respectively. The models and relevant methodology can provide scenarios for further application of natural polymer solutions.

With the large-scale development of drugs, understanding the drug phase behaviors in complex systems become increasingly important. Among them, the solubility of drugs in biorelevant media needs to be urgently understood. To address this challenge, new strategies based on machine learning models are proposed. First, the strategy trains five machine learning models based on fifteen molecular descriptors of the drug molecular properties. The XGboost model was identified as the best predictive model for predicting drug solubility performance in various solvents. Next, the input feature vectors were expanded for machine learning using the MACCS chemical fingerprint coupled with the XGboost model. The MACCS chemical fingerprint coupled with the XGboost model has significantly improved the prediction accuracy of drug solubility. This finding demonstrates that the proposed strategy has solubility prediction capability, which is expected to provide valid information for drug development and drug solvent screening.

In this work, the influence of polymeric excipients and relative humidity (RH) on stability of amorphous solid dispersions (ASDs) was investigated. Irbesartan and oxaprozin of BCS Ⅱ were selected as model active pharmaceutical ingredients (APIs). PVP and PVP/VA were chosen as model excipients. The water sorption and the physical stability of amorphous solid dispersions stored at constant temperature and humidity were measured. The thermodynamic phase diagrams were constructed using Perturbed-Chain Statistical Associating Fluid Theory (PC-SAFT) and Gordon-Taylor equation. The results showed that the hygroscopicity of PVP-containing was stronger than PVP/VA-containing. The water sorption was found to significantly decrease the API solubility in the polymer as well as the glass-transition temperature of the formulation. The solubility of irbesartan and oxaprozin in PVP and PVP/VA were further predicted at 25 °C with PC-SAFT. It could guide the design of amorphous solid dispersions and selection of storage conditions.

The influence of temperature, stirring speed, and excipients on crystal growth kinetics of mesalazine and allopurinol was investigated through experiment and chemical potential gradient model. The results indicated that the Diffusion-Surface Reaction model (DSR (1,2)) showed good performance in modeling API crystal growth kinetics within the ARDs of 4%. Excipients played a crucial role in inhibiting crystal growth in all the systems. It can not only improve the API solubility, but also reduce the crystal growth rate. By comparing diffusion rate and surface-reaction rate constant within the DSR (1,2) model, it was found that the controlling step of mesalazine crystallization was surface-reaction. Allopurinol crystallization was dominated by both surface-reaction and diffusion. Meanwhile, the crystal growth kinetics of mesalazine and allopurinol were predicted successfully with the ARDs of 2.53% and 4.78%. This work provided a mechanistic understanding of polymer influence on the inhibition of API crystal growth.