Estimating reaction model parameter uncertainty with Markov Chain Monte Carlo

被引:15
作者
Albrecht, Jacob [1 ]
机构
[1] Bristol Myers Squibb Co, New Brunswick, NJ 08901 USA
关键词
Markov Chain Monte Carlo (MCMC); Non-parametric statistics; Chemical kinetic modeling; Nonlinear regression; ICH Q8 DEFINITION; CONFIDENCE-INTERVALS; DESIGN; REGIONS;
D O I
10.1016/j.compchemeng.2012.07.011
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
Predicting the performance of chemical reactions with a mechanistic model is desired during the development of pharmaceutical and other high value chemical syntheses. Model parameters usually must be regressed to experimental observations. However, experimental error may not follow conventional distributions and the validity of common statistical assumptions used for regression should be examined when fitting mechanistic models. This paper compares different techniques to estimate parameter confidence for reaction models encountered in pharmaceutical manufacturing, simulated with either normally distributed or experimentally measured noise. Confidence intervals were calculated following standard linear approaches and two Markov Chain Monte Carlo algorithms utilizing a Bayesian approach to parameter estimation: one assuming a normal error distribution, and a new non-parametric likelihood function. While standard frequentist approaches work well for simpler nonlinear models and normal distributions, only MCMC accurately estimates uncertainty when the system is highly nonlinear, and can account for any measurement bias via customized likelihood functions. (C) 2012 Elsevier Ltd. All rights reserved.
引用
收藏
页码:14 / 28
页数:15
相关论文
共 40 条
[1]   STANDARD ERRORS AND CONFIDENCE-INTERVALS IN NONLINEAR-REGRESSION - COMPARISON OF MONTE-CARLO AND PARAMETRIC STATISTICS [J].
ALPER, JS ;
GELB, RI .
JOURNAL OF PHYSICAL CHEMISTRY, 1990, 94 (11) :4747-4751
[2]  
[Anonymous], 2021, Bayesian data analysis
[3]  
[Anonymous], Philosophical Transactions of the Royal Society of London for, DOI DOI 10.1098/RSTL.1763.0053
[4]   PARAMETER TRANSFORMATIONS FOR IMPROVED APPROXIMATE CONFIDENCE-REGIONS IN NON-LINEAR LEAST-SQUARES [J].
BATES, DM ;
WATTS, DG .
ANNALS OF STATISTICS, 1981, 9 (06) :1152-1167
[5]  
Beers KJ, 2007, NUMERICAL METHODS FOR CHEMICAL ENGINEERING: APPLICATIONS IN MATLAB, P1
[6]   High fidelity mathematical model building with experimental data: A Bayesian approach [J].
Blau, Gary ;
Lasinski, Michael ;
Orcun, Seza ;
Hsu, Shuo-Huan ;
Caruthers, Jim ;
Delgass, Nicholas ;
Venkatasubramanian, Venkat .
COMPUTERS & CHEMICAL ENGINEERING, 2008, 32 (4-5) :971-989
[7]  
Bowman AW., 1997, Applied Smoothing Techniques for Data Analysis: The Kernel Approach with S-Plus Illustrations, Vvol. 18
[8]  
Box G.E., 2011, Bayesian inference in statistical analysis
[9]  
Box G. E., 1978, Statistics for experimenters, V664
[10]   Evaluation of realistic 95% confidence intervals for the activation energy calculated by the iterative linear integral isoconversional method [J].
Cai, Junmeng ;
Han, Dong ;
Chen, Yong ;
Chen, Siyu .
CHEMICAL ENGINEERING SCIENCE, 2011, 66 (12) :2879-2882