Application of machine learning in combination with mechanistic modeling to predict plasma exposure of small molecules DOI Creative Commons
Panteleimon D. Mavroudis, Donato Teutonico, Alexandra Abós

et al.

Frontiers in Systems Biology, Journal Year: 2023, Volume and Issue: 3

Published: June 20, 2023

Prediction of a new molecule’s exposure in plasma is critical first step toward understanding its efficacy/toxicity profile and concluding whether it possible first-in-class, best-in-class candidate. For this prediction, traditional pharmacometrics use variety scaling methods that are heavily based on pre-clinical pharmacokinetic (PK) data. We here propose novel framework which preclinical prediction performed by applying machine learning (ML) tandem with mechanism-based modeling. In our proposed method, relationship initially established between molecular structure physicochemical (PC)/PK properties using ML, then the ML-driven PC/PK parameters used as input to mechanistic models ultimately predict candidates. To understand feasibility framework, we evaluated number (1-compartment, physiologically (PBPK)), PBPK distribution (Berezhkovskiy, PK-Sim standard, Poulin Theil, Rodgers Rowland, Schmidt), parameterizations (using vivo , or vitro clearance). most scenarios tested, results demonstrate PK profiles can be adequately predicted framework. Our analysis further indicates some limitations when liver microsomal intrinsic clearance (CLint) only pathway underscores necessity investigating variability emanating from different providing predictions. The suggested approach aims at earlier drug development process so decisions molecule screening, chemistry design, dose selection made early possible.

Language: Английский

Augmented allometric scaling: Predicting drug clearance in farm animals with machine learning using body weight DOI Creative Commons

David Inauen,

L.S. Lautz, Jan C.M. Hendriks

et al.

Computational Toxicology, Journal Year: 2025, Volume and Issue: unknown, P. 100341 - 100341

Published: March 1, 2025

Language: Английский

Citations

1
Multi-parameter molecular MRI quantification using physics-informed self-supervised learning DOI Creative Commons

Alex Finkelstein,

Nikita Vladimirov, Moritz Zaiß

et al.

Communications Physics, Journal Year: 2025, Volume and Issue: 8(1)

Published: April 15, 2025

Language: Английский

Citations

1
The Comparison of Machine Learning and Mechanistic In Vitro–In Vivo Extrapolation Models for the Prediction of Human Intrinsic Clearance DOI
Christopher Keefer, George Chang, Li Di

et al.

Molecular Pharmaceutics, Journal Year: 2023, Volume and Issue: 20(11), P. 5616 - 5630

Published: Oct. 9, 2023

Accurate prediction of human pharmacokinetics (PK) remains one the key objectives drug metabolism and PK (DMPK) scientists in discovery projects. This is typically performed by using vitro-in vivo extrapolation (IVIVE) based on mechanistic models. In recent years, machine learning (ML), with its ability to harness patterns from previous outcomes predict future events, has gained increased popularity application absorption, distribution, metabolism, excretion (ADME) sciences. study compares performance various ML models for IV clearance a large (645) set diverse compounds literature data, as well measured relevant vitro end points. were built multiple approaches descriptors: (1) calculated physical properties structural descriptors chemical structure alone (classical QSAR/QSPR); (2) inputs only no structure-based (ML IVIVE); (3) silico IVIVE model predictions inputs. For models, well-stirred parallel-tube liver considered without use empirical scaling factors renal clearance. The best intrinsic (CLint) was an six average absolute fold error (AAFE) 2.5. used model, resulting AAFE 2.8. corresponding full achieved 3.3. These relative performances confirmed 16 Pfizer candidates that not part original data set. Results show are comparable or superior their counterparts. We also can be derive insights into improvement prediction.

Language: Английский

Citations

16
Utility of life stage-specific chemical risk assessments based on New Approach Methodologies (NAMs) DOI
Pavani Gonnabathula,

Me-Kyoung Choi,

Miao Li

et al.

Food and Chemical Toxicology, Journal Year: 2024, Volume and Issue: 190, P. 114789 - 114789

Published: June 5, 2024

Language: Английский

Citations

6
Application of machine learning in combination with mechanistic modeling to predict plasma exposure of small molecules DOI Creative Commons
Panteleimon D. Mavroudis, Donato Teutonico, Alexandra Abós

et al.

Frontiers in Systems Biology, Journal Year: 2023, Volume and Issue: 3

Published: June 20, 2023

Prediction of a new molecule’s exposure in plasma is critical first step toward understanding its efficacy/toxicity profile and concluding whether it possible first-in-class, best-in-class candidate. For this prediction, traditional pharmacometrics use variety scaling methods that are heavily based on pre-clinical pharmacokinetic (PK) data. We here propose novel framework which preclinical prediction performed by applying machine learning (ML) tandem with mechanism-based modeling. In our proposed method, relationship initially established between molecular structure physicochemical (PC)/PK properties using ML, then the ML-driven PC/PK parameters used as input to mechanistic models ultimately predict candidates. To understand feasibility framework, we evaluated number (1-compartment, physiologically (PBPK)), PBPK distribution (Berezhkovskiy, PK-Sim standard, Poulin Theil, Rodgers Rowland, Schmidt), parameterizations (using vivo , or vitro clearance). most scenarios tested, results demonstrate PK profiles can be adequately predicted framework. Our analysis further indicates some limitations when liver microsomal intrinsic clearance (CLint) only pathway underscores necessity investigating variability emanating from different providing predictions. The suggested approach aims at earlier drug development process so decisions molecule screening, chemistry design, dose selection made early possible.

Language: Английский

Citations

13