Machine learning and artificial intelligence in physiologically based pharmacokinetic modeling

Abstract

Physiologically based pharmacokinetic (PBPK) models are useful tools in drug development and risk assessment of environmental chemicals. PBPK model development requires the collection of species-specific physiological, and chemical-specific absorption, distribution, metabolism, and excretion (ADME) parameters, which can be a time-consuming and expensive process. This raises a need to create computational models capable of predicting input parameter values for PBPK models, especially for new compounds. In this review, we summarize an emerging paradigm for integrating PBPK modeling with machine learning (ML) or artificial intelligence (AI)-based computational methods. This paradigm includes 3 steps (1) obtain time-concentration PK data and/or ADME parameters from publicly available databases, (2) develop ML/AI-based approaches to predict ADME parameters, and (3) incorporate the ML/AI models into PBPK models to predict PK summary statistics (eg, area under the curve and maximum plasma concentration). We also discuss a neural network architecture "neural ordinary differential equation (Neural-ODE)" that is capable of providing better predictive capabilities than other ML methods when used to directly predict time-series PK profiles. In order to support applications of ML/AI methods for PBPK model development, several challenges should be addressed (1) as more data become available, it is important to expand the training set by including the structural diversity of compounds to improve the prediction accuracy of ML/AI models; (2) due to the black box nature of many ML models, lack of sufficient interpretability is a limitation; (3) Neural-ODE has great potential to be used to generate time-series PK profiles for new compounds with limited ADME information, but its application remains to be explored. Despite existing challenges, ML/AI approaches will continue to facilitate the efficient development of robust PBPK models for a large number of chemicals.

Keywords: in vitro to in vivo extrapolation (IVIVE); artificial intelligence; machine learning; pharmacometrics; physiologically based pharmacokinetic (PBPK) modeling; risk assessment.

Figure 1.

An emerging research paradigm to integrate machine learning and artificial intelligence approaches with physiologically based pharmacokinetic modeling. A, A database consisting of in vivo PK profiles and in vitro ADME assays (eg, permeability) is obtained from the literature, which is then used for the establishment of ML/DL-based model. B, ML/DL algorithms (eg, SVM, RF, and DNN) can be used to estimate ADME parameters by training with chemical descriptors and the properties of the molecules. These ADME parameters can be used as input parameters (eg, F, Cl, fu) for the development of a generic PBPK model. C, The ML-generic PBPK model can be used to generate the secondary PK parameters including AUC, C_max, and Vd and subsequently be evaluated with in vivo PK data. Once a PBPK model is evaluated to be adequate or acceptable at Step C, it can then be used to generate simulated time-concentration data, which in turn can be incorporated into existing databases or become a new database for Step A. Abbreviations: ADME, absorption, distribution, metabolism, and excretion; AUC, area under the curve; Cl, clearance; C_max, maximum plasma concentration; DL, deep learning; DNN, deep neural network; F, bioavailability; ML, machine learning; PBPK, physiologically based pharmacokinetic; PK, pharmacokinetic; RF, random forest; SVM, support vector machine; Vd, volume of distribution.

Figure 2.

Schematic of the neural ordinary differential equation (Neural-ODE) model. The Neural-ODE model consists of encoder, ODE solver, and decoder parts to predict the time-series PK profiles. The TFDS, TIME, CYCL, AMT, and PK cycle 1 observation are used as the input features in the RNN encoder part. Then ODE solvers are used to incorporate dosing information into the time sequence before the decoder generates the predictions. Abbreviations: AMT, the dosing amount in milligrams; CYCL, the current dosing cycle number; PK, pharmacokinetic; ODE, ordinary differential equation; RNN, recurrent neural network; TFDS, the time in hours between each dose; TIME, the time in hours since the start of the treatment. This figure was adapted based on Lu et al. (2021a) with permission from the publisher.