Laure Ponthier, Julie Autmizguine, Benedicte Franck, Anders Åsberg, Philippe Ovetchkine, Alexandre Destere, Pierre Marquet, Marc Labriffe, Jean-Baptiste Woillard
{"title":"Optimization of Ganciclovir and Valganciclovir Starting Dose in Children by Machine Learning.","authors":"Laure Ponthier, Julie Autmizguine, Benedicte Franck, Anders Åsberg, Philippe Ovetchkine, Alexandre Destere, Pierre Marquet, Marc Labriffe, Jean-Baptiste Woillard","doi":"10.1007/s40262-024-01362-7","DOIUrl":null,"url":null,"abstract":"<p><strong>Background and objectives: </strong>Ganciclovir (GCV) and valganciclovir (VGCV) show large interindividual pharmacokinetic variability, particularly in children. The objectives of this study were (1) to develop machine learning (ML) algorithms trained on simulated pharmacokinetics profiles obtained by Monte Carlo simulations to estimate the best ganciclovir or valganciclovir starting dose in children and (2) to compare its performances on real-world profiles to previously published equation derived from literature population pharmacokinetic (POPPK) models achieving about 20% of profiles within the target.</p><p><strong>Materials and methods: </strong>The pharmacokinetic parameters of four literature POPPK models in addition to the World Health Organization (WHO) growth curve for children were used in the mrgsolve R package to simulate 10,800 pharmacokinetic profiles. ML algorithms were developed and benchmarked to predict the probability to reach the steady-state, area-under-the-curve target (AUC<sub>0-24</sub> within 40-60 mg × h/L) based on demographic characteristics only. The best ML algorithm was then used to calculate the starting dose maximizing the target attainment. Performances were evaluated for ML and literature formula in a test set and in an external set of 32 and 31 actual patients (GCV and VGCV, respectively).</p><p><strong>Results: </strong>A combination of Xgboost, neural network, and random forest algorithms yielded the best performances and highest target attainment in the test set (36.8% for GCV and 35.3% for the VGCV). In actual patients, the best GCV ML starting dose yielded the highest target attainment rate (25.8%) and performed equally for VGCV with the Franck model formula (35.3% for both).</p><p><strong>Conclusion: </strong>The ML algorithms exhibit good performances in comparison with previously validated models and should be evaluated prospectively.</p>","PeriodicalId":10405,"journal":{"name":"Clinical Pharmacokinetics","volume":" ","pages":"539-550"},"PeriodicalIF":4.6000,"publicationDate":"2024-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Clinical Pharmacokinetics","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1007/s40262-024-01362-7","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2024/3/16 0:00:00","PubModel":"Epub","JCR":"Q1","JCRName":"PHARMACOLOGY & PHARMACY","Score":null,"Total":0}
引用次数: 0
Abstract
Background and objectives: Ganciclovir (GCV) and valganciclovir (VGCV) show large interindividual pharmacokinetic variability, particularly in children. The objectives of this study were (1) to develop machine learning (ML) algorithms trained on simulated pharmacokinetics profiles obtained by Monte Carlo simulations to estimate the best ganciclovir or valganciclovir starting dose in children and (2) to compare its performances on real-world profiles to previously published equation derived from literature population pharmacokinetic (POPPK) models achieving about 20% of profiles within the target.
Materials and methods: The pharmacokinetic parameters of four literature POPPK models in addition to the World Health Organization (WHO) growth curve for children were used in the mrgsolve R package to simulate 10,800 pharmacokinetic profiles. ML algorithms were developed and benchmarked to predict the probability to reach the steady-state, area-under-the-curve target (AUC0-24 within 40-60 mg × h/L) based on demographic characteristics only. The best ML algorithm was then used to calculate the starting dose maximizing the target attainment. Performances were evaluated for ML and literature formula in a test set and in an external set of 32 and 31 actual patients (GCV and VGCV, respectively).
Results: A combination of Xgboost, neural network, and random forest algorithms yielded the best performances and highest target attainment in the test set (36.8% for GCV and 35.3% for the VGCV). In actual patients, the best GCV ML starting dose yielded the highest target attainment rate (25.8%) and performed equally for VGCV with the Franck model formula (35.3% for both).
Conclusion: The ML algorithms exhibit good performances in comparison with previously validated models and should be evaluated prospectively.
期刊介绍:
Clinical Pharmacokinetics promotes the continuing development of clinical pharmacokinetics and pharmacodynamics for the improvement of drug therapy, and for furthering postgraduate education in clinical pharmacology and therapeutics.
Pharmacokinetics, the study of drug disposition in the body, is an integral part of drug development and rational use. Knowledge and application of pharmacokinetic principles leads to accelerated drug development, cost effective drug use and a reduced frequency of adverse effects and drug interactions.