Home » Key Scientific Articles » Modeling and Prediction of Solvent Effect on Human Skin Permeability using Support Vector Regression and Random Forest

Modeling and Prediction of Solvent Effect on Human Skin Permeability using Support Vector Regression and Random Forest

Journal Reference

Pharm Res. 2015 Nov;32(11):3604-17.

Baba H1,2, Takahara J3, Yamashita F4, Hashida M4,5

Show Affiliations
  1. Kyoto R&D Center, Maruho Co., Ltd., 93 Awata-cho, Chudoji, Shimogyo-ku, 600-8815, Kyoto, Japan. [email protected]
  2. Department of Drug Delivery Research, Graduate School of Pharmaceutical Sciences, Kyoto University, 46-29, Yoshida-shimoadachicho, Sakyo-ku, Kyoto, 606-8501, Japan. [email protected]
  3. Kyoto R&D Center, Maruho Co., Ltd., 93 Awata-cho, Chudoji, Shimogyo-ku, 600-8815, Kyoto, Japan.
  4. Department of Drug Delivery Research, Graduate School of Pharmaceutical Sciences, Kyoto University, 46-29, Yoshida-shimoadachicho, Sakyo-ku, Kyoto, 606-8501, Japan.
  5. Institute for Integrated Cell-Material Sciences, Kyoto University, 46-29, Yoshida-shimoadachicho, Sakyo-ku, Kyoto, 606-8501, Japan.

Abstract

PURPOSE:

The solvent effect on skin permeability is important for assessing the effectiveness and toxicological risk of new dermatological formulations in pharmaceuticals and cosmetics development. The solvent effect occurs by diverse mechanisms, which could be elucidated by efficient and reliable prediction models. However, such prediction models have been hampered by the small variety of permeants and mixture components archived in databases and by low predictive performance. Here, we propose a solution to both problems.

METHODS:

We first compiled a novel large database of 412 samples from 261 structurally diverse permeants and 31 solvents reported in the literature. The data were carefully screened to ensure their collection under consistent experimental conditions. To construct a high-performance predictive model, we then applied   support vector regression (SVR) and random forest (RF) with greedy stepwise descriptor selection to our database. The models were internally and externally validated.

RESULTS:

The support vector regression achieved higher performance statistics than random forest. The (externally validated) determination coefficient, root mean square error, and mean absolute error of support vector regression were 0.899, 0.351, and 0.268, respectively. Moreover, because all descriptors are fully computational, our method can predict as-yet unsynthesized compounds.

CONCLUSION:

Our high-performance prediction model offers an attractive alternative to permeability experiments for pharmaceutical and cosmetic candidate screening and optimizing skin-permeable topical formulations.

Go To Pharm Res