Publications and Presentations

Prediction of In Vivo Pharmacokinetic Parameters and Time–Exposure Curves in Rats Using Machine Learning from the Chemical Structure

Prediction of In Vivo Pharmacokinetic Parameters and Time–Exposure Curves in Rats Using Machine Learning from the Chemical Structure

O. Obrezanova, A. Martinsson, T. Whitehead, S. Mahmoud, A. Bender, F. Miljković, P. Grabowski, B. Irwin, I. Oprisiu, G. Conduit, M. Segall, G. F. Smith, B. Williamson, S. Winiwarter, and N. Greene Mol. Pharmaceutics 2022, 19, 5, 1488–1504

DOI: 10.1021/acs.molpharmaceut.2c00027

This article is a collaboration with Intellegens, the University of Cambridge and AstraZeneca. It provides a proof-of-concept study in which Cerella™ is used to predict rat in vivo pharmacokinetic (PK) parameters and concentration–time PK profiles.

Prediction of In Vivo Pharmacokinetic Parameters and Time–Exposure Curves in Rats Using Machine Learning from the Chemical Structure screenshot

Abstract

Animal pharmacokinetic (PK) data as well as human and animal in vitro systems are utilized in drug discovery to define the rate and route of drug elimination. Accurate prediction and mechanistic understanding of drug clearance and disposition in animals provide a degree of confidence for extrapolation to humans. In addition, in vivo property predictions can be used to improve design during drug discovery, help select compounds with better properties, and reduce the number of in vivo experiments.

In this study, we generated machine learning models able to predict rat in vivo PK parameters and concentration–time PK profiles based on the molecular chemical structure and either measured or predicted in vitro parameters. We trained the models on internal in vivo rat PK data for over 3000 diverse compounds from multiple projects and therapeutic areas. The predicted endpoints include clearance and oral bioavailability. We compared the performance of various traditional machine learning algorithms and deep learning approaches, including graph convolutional neural networks. The best models for PK parameters achieved R2 = 0.63 [root mean squared error (RMSE) = 0.26] for clearance and R2 = 0.55 (RMSE = 0.46) for bioavailability. The models provide a fast and cost-efficient way to guide the design of molecules with optimal PK profiles. This enables the prediction of virtual compounds at the point of design, and drives the prioritization of compounds for in vivo assays.

You can download this article from the journal webpage via the button below. Additionally, for further detail on how Cerella can be used to predict PK parameters and curves, watch our webinar. It features three of the authors of this paper, Matt Segall (Optibrium), Tom Whitehead (Intellegens), Nigel Greene (AstraZeneca), discussing their work around Cerella and its applications.

INTERESTED IN AI FOR DRUG DISCOVERY?

Discover Cerella™

Cerella™ is a unique artificial intelligence platform. It supports medicinal chemists and other discovery scientists, escalating success rates and advancing small molecule drug discovery.

Cerella’s AI platform is proven to overcome limitations in drug discovery data, confidently delivering results and seamlessly integrating with your med chem software platforms.