Research reveals that machine learning and stochastic differential equations improve predictions of permeability coefficients from molecular weight data.
Key Points
This research aims to enhance predictions of chemical transport across skin by applying machine learning and stochastic modeling techniques.
Applied stochastic differential equations to model variability in experimental data.
Utilized the Euler–Maruyama algorithm and Milstein scheme among others to predict permeability.
Implemented machine learning models including gradient boost regression and LevenbergMarquardt algorithm.
Gradient boost regression achieved R2 of 0.84 and MSE of 0.45.
Stochastic models provided MSE values around 0.63 with R2 above 0.70.
The study suggests improved accuracy in predicting skin permeability coefficients.