Parameter identifiability of power-law biochemical system models
Parameter identifiability of power-law biochemical system models
复制标题
DOI：
10.1016/j.jbiotec.2010.02.019
复制
发表时间：
2010-09-01
影响因子：
4.1
通讯作者：
Gunawan, Rudiyanto
中科院分区：
文献类型：
--
作者：
Srinath, Sridharan;Gunawan, Rudiyanto
数学建模已成为生物技术中不可或缺的组成部分，其中这些模型经常用于设计和优化生物过程。典型模型，如生化系统理论中的幂律，提供了许多数学和数值优势，包括模拟一般非线性行为的内置灵活性。这种模型的构建依赖于通过实验数据拟合（也称为逆建模）来估计未知的病例特异性模型参数。尽管有大量的出版物在这个主题上，这个任务仍然是生化系统的规范建模的瓶颈。本文的重点关注的问题，从动态数据的幂律模型的可识别性，即参数值是否可以唯一和准确地识别从时间序列数据。将现有的和新发展的参数可辨识性方法应用于生化系统的两个幂律模型，结果表明参数可辨识性的缺乏是逆建模困难的根本原因。尽管重点是幂律模型，分析和结论是可扩展的其他典型模型，和参数可识别性的问题，预计将是一个共同的问题，在生化系统建模。(C)2010 Elsevier BV保留所有权利。
Mathematical modeling has become an integral component in biotechnology, in which these models are frequently used to design and optimize bioprocesses. Canonical models, like power-laws within the Biochemical Systems Theory, offer numerous mathematical and numerical advantages, including built-in flexibility to simulate general nonlinear behavior. The construction of such models relies on the estimation of unknown case-specific model parameters by way of experimental data fitting, also known as inverse modeling. Despite the large number of publications on this topic, this task remains the bottleneck in canonical modeling of biochemical systems. The focus of this paper concerns with the question of identifiability of power-law models from dynamic data, that is, whether the parameter values can be uniquely and accurately identified from time-series data. Existing and newly developed parameter identifiability methods were applied to two power-law models of biochemical systems, and the results pointed to the lack of parametric identifiability as the root cause of the difficulty faced in the inverse modeling. Despite the focus on power-law models, the analyses and conclusions are extendable to other canonical models, and the issue of parameter identifiability is expected to be a common problem in biochemical system modeling. (C) 2010 Elsevier B.V. All rights reserved.