A Multi-Method Comparison of Machine Learning in Predicting Pharmacokinetic Parameters - A Simulation Study

Aug 5, 2026

This simulation study evaluates whether machine learning methods can replace the Lasso covariate selection approach in nonlinear mixed-effects models for pharmacokinetic (PK) parameter prediction. Lasso, support vector regression (SVR), and random forest (RF) are compared by mean absolute prediction error after simulating PK data from a one-compartment oral model. Covariates are generated with controlled correlation, and true covariates affect clearance at varying magnitudes. SVR shows the strongest performance in small datasets, especially under high covariate correlation, while Lasso yields higher error.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Abstract

Introduction

Methods

Results

Conclusion

Journal of Biostatistics and Epidemiology J BiostatEpidemiol. 2024;10(1): 98-110 Original Article A Multi-Method Comparison of Machine Learning in Predicting Pharmacokinetic Parameters: A Simulation Study Marziyeh Doostfatemeh, Kamal Amini, Elham Haem* Department of Biostatistics, Shiraz University of Medical Sciences, Shiraz, Iran. | ARTICLE INFO | ABSTRACT | | --- | --- | Received 08.12.2023 Revised 26.01.2024 Accepted 20.02.2024 Published 15.03.2024 Keywords: Population pharmacokinetics; Machine learning; lasso; Random forest; Support vector regression Introduction: One important aim of population pharmacokinetics (PK) and pharmacodynamics (PD) is the identification and quantification of the relationships between the parameter and covariates to improve the predictive performance of the population PK/PD modeling. Several new mathematical methods have been developed in pharmacokinetics in recent years which indicated that the machine learning-based methods are an appealing tool for analyzing PK/PD data. Methods: This simulation-base study aims to determine whether machine learning methods, including support vector regression (SVR) and Random forest (RF) which are specifically designed for the prediction of blood serum concentration or clearance, could be an effective replacement for the Lasso covariate selection method in nonlinear mixed effect models. Accordingly, the predictive performance of penalized regression Lasso, SVR, and RF regression was compared to detect the associations between clearance and model covariates. PK data was simulated from a one-compartment model with oral administration. Covariates were created by sampling from a multivariate standard normal distribution with different levels of correlation. The true covariates influenced only clearance at different magnitudes. Lasso, RF, and SVR were compared in terms of mean absolute prediction error (MAE) . Results: The results show that SVR performed the best in small data sets, even in those in which a high correlation existed between covariates. This makes SVR a promising method for covariate selection in nonlinear mixed-effect models. Conclusion: The Lasso method offered a higher MAE, making it less promising than RF and SVR, especially when dealing with a high correlation between covariates and a low number of individuals. Introduction Population pharmacokinetics and pharmacodynamics (PK/PD) models have been extensively used to identify how individual * .[Corresponding Author:](Corresponding Author: Elhamhaem@gmail.com)[ ](Corresponding Author: Elhamhaem@gmail.com)[Elhamhaem@gmail.com](Corresponding Author: Elhamhaem@gmail.com) factors such as demographics, genotype, phenotypic disease, and environmental factors such as medications or alcohol consumption affect patient exposure to the drug and their further response. The structure of data in PK/ Copyright © 2024 Tehran University of Medical Sciences. Published by Tehran University of Medical Sciences. This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International license ([https://creativecommons.org/licenses/by-nc/4.0/](https://creativecommons.org/licenses/by-nc/4.0/)) . Noncommercial uses of the work are permitted, provided the original work is properly cited. A Multi-Method Comparison of Machine Learning in predicting ... PD studies includes subject demographics, drug administration details, and measurements of drug concentration and effects and they are essential in understanding how drugs behave in the body and their effects on biological systems. Nonlinear mixed effects modeling, as implemented by the popular software, NONMEM, has been widely used in the analysis of PK/ PD data. The most important part of population PK/PD modeling is the evaluation of the relationships between model parameters and covariates. In this regard, the selection of a subset of the covariate relations is often performed via the stepwise method; however,

FAQs

What is the primary goal of the simulation study?

To determine whether machine learning methods (SVR and random forest) can effectively replace Lasso covariate selection in nonlinear mixed-effects models for predicting PK parameters.

How was the PK dataset generated for the comparison?

PK data were simulated from a one-compartment model with oral administration, and covariates were sampled from a multivariate standard normal distribution with different correlation levels.

Which method performed best and under what conditions?

Support vector regression performed best in small data sets, including cases with high correlation between covariates.

Why is the Lasso method considered less promising in the study?

Lasso produced higher mean absolute prediction error than RF and SVR, particularly when covariates were highly correlated and the number of individuals was low.

Aug 5, 2026

Upload by Theodore