Instituto Politécnico Nacional, Centro de Investigación y Desarrollo de Tecnología Digital (CITEDI-IPN), Av. Instituto Politécnico Nacional, No. 1310, Col. Nueva Tijuana, 22435 Tijuana, BC, Mexico.
Facultad de Ciencias Químicas e Ingeniería, Universidad Autónoma de Baja California, Calzada Universidad 14418, Parque Industrial Internacional Tijuana, 22390 Tijuana, BC, Mexico.
Comput Intell Neurosci. 2018 Nov 1;2018:1983897. doi: 10.1155/2018/1983897. eCollection 2018.
Population pharmacokinetic (PopPK) models allow researchers to predict and analyze drug behavior in a population of individuals and to quantify the different sources of variability among these individuals. In the development of PopPK models, the most frequently used method is the nonlinear mixed effect model (NLME). However, once the PopPK model has been developed, it is necessary to determine if the selected model is the best one of the developed models during the population pharmacokinetic study, and this sometimes becomes a multiple criteria decision making (MCDM) problem, and frequently, researchers use statistical evaluation criteria to choose the final PopPK model. The used evaluation criteria mentioned above entail big problems since the selection of the best model becomes susceptible to the human error mainly by misinterpretation of the results. To solve the previous problems, we introduce the development of a software robot that can automate the task of selecting the best PopPK model considering the knowledge of human expertise. The software robot is a fuzzy expert system that provides a method to systematically perform evaluations on a set of candidate PopPK models of commonly used statistical criteria. The presented results strengthen our hypothesis that the software robot can be successfully used to evaluate PopPK models ensuring the selection of the best PopPK model.
群体药代动力学（PopPK）模型允许研究人员预测和分析个体人群中的药物行为，并量化这些个体之间的不同变异来源。在 PopPK 模型的开发中，最常用的方法是非线性混合效应模型（NLME）。然而，一旦开发了 PopPK 模型，就有必要确定在群体药代动力学研究中，所选模型是否是已开发模型中最好的一个，这有时会成为多准则决策问题（MCDM），并且研究人员经常使用统计评估标准来选择最终的 PopPK 模型。上述使用的评估标准存在很大问题，因为选择最佳模型容易受到人为错误的影响，主要是由于对结果的误解。为了解决上述问题，我们引入了软件机器人的开发，该机器人可以在考虑人类专业知识的情况下，自动选择最佳 PopPK 模型。软件机器人是一个模糊专家系统，它提供了一种方法，可以系统地对一组常用统计标准的候选 PopPK 模型进行评估。所呈现的结果证实了我们的假设，即软件机器人可以成功地用于评估 PopPK 模型，从而确保选择最佳的 PopPK 模型。
