(584g) Diffusion Coefficient Prediction Model for Sustained Release From Silk Films

The overall goal of the research project is to develop a prediction model set that can accurately estimate the diffusion coefficient for small molecule release from silk films by accounting for the physiochemical properties of the chemical. This type of a predictive model would benefit the design of silk fibroin sustained release devices for drug delivery by simulating release and reducing experimentation. Also, a robust predictive model will help to identify material properties that can be manipulated as design variables to achieve a target release profile. In order to develop this predictive model set the diffusional release mechanism was confirmed by fitting an analytical diffusion model to the experimental release data. A group of synthetic dyes varying in molecular weight, solubility, partition coefficient, pK_a, and pK_b were chosen for release from silk fibroin films to construct the predictive model. Synthetic dyes were chosen for the experimental design because they are relatively stable, their release into solution can be easily detected by spectroscopy, and they exhibit a wide range of physiochemical factors that can impact sustained release kinetics. Also there exist many classes of synthetic dyes that are derived from the same precursors and share similar structures. This allows the information to be approached as classes of pseudo drug libraries. Once several diffusion coefficient estimates were obtained for the dyes, empirical relationships were constructed between the diffusion coefficient estimates and the physiochemical properties of the dyes with multivariate regression analysis. Cross validation of the predictive model with a separate data was used to judge its interpolative and extrapolative predictive power in simulation. The developed model set can be updated for material properties, so that release kinetic relationships for more complex systems can be determined, in order to expand the predictive capabilities of the model set.