The application of feature selection to the development of Gaussian process models for percutaneous absorption

Abstract

The use of a nonlinear GPR method produced models with significantly improved predictivity, compared with SLN or QSPR models. Feature selection methods were able to provide important mechanistic information. However, it was also shown that significant synergy existed between certain parameters, and as such it was possible to interchange certain descriptors (i.e. molecular weight and melting point) without incurring a loss of model quality. Such synergy suggested that a model constructed from discrete terms in an equation may not be the most appropriate way of representing mechanistic understandings of skin absorption.

The application and limitations of mathematical modelling in the prediction of permeability across mammalian skin and polydimethylsiloxane membranes

“…However, Roberts and others [82–87] have shown that other parameters, particularly hydrogen bonding, are of importance. This has also been shown recently [16,20,88] . Indeed, Lam et al [16] .…”

“…Nevertheless, a major limitation to in‐vitro diffusion studies through excised human skin is the availability of suitable tissue. This has resulted in the widespread use of synthetic membranes and various animal models in place of excised human skin, and may impact upon the ability to develop representative models based on in‐vitro experimental data [16] . Studies comparing drug penetration across the skin of various animals with drug penetration across human skin have indicated that the skin of several animals provide excellent models for human skin.…”

“…Analysis of human skin permeability has been shown to be non‐linear when mathematically related to the physicochemical parameters of penetrants [16,20,21] . These studies have also shown that non‐linear methods, such as Gaussian processes (GPs), have statistically outperformed other methods, such as QPSRs.…”

“…These studies have also shown that non‐linear methods, such as Gaussian processes (GPs), have statistically outperformed other methods, such as QPSRs. Further, they have shown that GP models provided more accurate predictions of permeability than QSPR models [16,21] . Therefore, the aims of this study were twofold; firstly, to apply and evaluate GP methods, by comparison with QPSR models, to datasets for membranes other than human skin, which are commonly used as reasonable replacements.…”

Gaussian process regression with functional covariates and multivariate response

“…See for example the seminal book [3] and the references therein for details. In chemometrics and related areas, GPR has been applied to a range of problems, such as calibration of spectroscopic analysers [4,5], response surface modelling [6], system identification [7], ensemble learning [5,8], prediction of transmembrane pressure [9], and prediction of percutaneous absorption [10,11], among others.…”