Subject Specific Deep Neural Network for Longitudinal Study in Pharmacokinetics and Pharmacodynamics
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
In this study we propose a subject specific deep neural network (SSDNN) model for analyzing pharmacokinetic (PK) and pharmacodynamic (PD) data. PK and PD data are obtained at subject-specific irregular time intervals, and a different number of observations are collected for each subject, based on the number of times the subject visited the hospital. The SSDNN’s performance is compared to that of the standard neural network (NN) and support vector machine (SVM) using three evaluation metrics, which are mean squared error (MSE), mean absolute error (MAE) and mean relative absolute error (MRAE). We find that the absolute values of the four measures of the proposed SSDNN are significantly lower than those of NN and SVR for PK and PD data. These findings imply that the proposed SSDNN is an appealing tool for analyzing PK and PD data.
Abstract
In this study we propose a subject specific deep neural network (SSDNN) model for analyzing pharmacokinetic (PK) and pharmacodynamic (PD) data. PK and PD data are obtained at subject-specific irregular time intervals, and a different number of observations are collected for each subject, based on the number of times the subject visited the hospital. The SSDNN’s performance is compared to that of the standard neural network (NN) and support vector machine (SVM) using three evaluation metrics, which are mean squared error (MSE), mean absolute error (MAE) and mean relative absolute error (MRAE). We find that the absolute values of the four measures of the proposed SSDNN are significantly lower than those of NN and SVR for PK and PD data. These findings imply that the proposed SSDNN is an appealing tool for analyzing PK and PD data.
참고문헌 (14)
- Gardiner JC, "Fixed effects, random effects and GEE : What are the differences", Stat Med, 28, pp.221~239 (2009)
- Beal S, "NONMEM User’s Guide Part 1", NONMEM Project Group, University of California, ~ (1992)
- Durisova M, "New mathematical methods in pharmacokinetic modeling", Basic Clin Pharmacol Toxicol, 96, pp.335~342 (2005)
- Brier ME, "Neural network predicted peak and trough gentamicin concentrations", Pharm Res, 12, pp.406~412 (1995)
- Chow H, "Application of neural networks to population pharmacokinetic data analysis", J Pharm Sci, 86, pp.840~845 (1997)
- Kang SH, "Population pharmacokinetic and pharmacodynamic models of remifentanil in healthy volunteers using artificial neural network analysis", Br J Clin Pharmacol, 64, pp.3~13 (2007)
- Seok KH, "Semiparametric mixed-effect least squares support vector machine for analyzing pharmacokinetic and pharmacodynamic data", Neurocomputing, 74, pp.3412~3419 (2011)
- Vapnik V. N., "The nature of statistical learning theory" (1995), Springer
- Kim K, "Pharmacokinetics and pharmacodynamics of propofol microemulsion and lipidemulsion after an intravenous bolus and variable rateinfusion", Anesthesiology, 106, pp.924~934 (2007)
- Molenberghs G, "Linear mixed models for longitudinal data" (2000), Springer
- 최연지, "Machine Learning Algorithms for Predicting the Returns of IPO Companies", Quantitative Bio-Science, 40(2), pp.71~75 (2021), 자연과학연구소[DOI]
- Klambauer G, "Selfnormalizing neural networks", Adv Neural Inf Process Syst (2017)
- Cerda P, "Similarity encoding for learning with dirty categorical variables", Mach Learn, 107, pp.1477~1494 (2018)
- Guo C, "Entity embeddings of categorical variables", arXiv:1604.06737, ~ (2016)