Empirical Representation of Blood Glucose Variability in a Compartmental Model
Empirical Representation of Blood Glucose Variability in a Compartmental Model
复制标题
DOI：
10.1007/978-3-319-25913-0_8
复制
发表时间：
2016-01-01
期刊：
影响因子：
--
通讯作者：
Breton, Marc D.
中科院分区：
文献类型：
--
作者：
Patek, Stephen D.;Lv, Dayu;Breton, Marc D.
饮食和运动行为对1型糖尿病的血糖结果有重要影响，但这些影响在现实生活中很难评估。虽然现有的数学模型忠实地表示（i）口服碳水化合物摄入和（ii）身体各个隔室中葡萄糖和胰岛素转运/作用的动态关系，但是需要精确的进餐和运动行为模型来真实地捕获在该领域中观察到的血糖的广泛波动，并且这已经成为包括人工胰腺在内的先进系统的临床前计算机评价的瓶颈。这项工作提出了一种使用连续葡萄糖监测和胰岛素泵数据来提取由口服碳水化合物净效应表示的BG变异性特征的方法，该特征可以被“反馈”回数学模型，以（i）从胰岛素输送的原始记录再现原始BG时间序列，以及（ii）用于近似胰岛素输送的修改时间表的效果。我们提供的基本方法的细节，并说明使用弗吉尼亚/帕多瓦1型模拟器和人类受试者的数据在现场研究的方法。
Eating and exercise behaviors have an important effect on glycemic outcomes in type 1 diabetes, yet these influences are difficult to assess in real-life settings. While existing mathematical models faithfully represent the dynamic relationships of (i) oral carbohydrate ingestion and (ii) glucose and insulin transport/ action in various compartments of the body, accurate models of meal and exercise behaviors are needed to realistically capture the wide excursions of blood glucose observed in the field, and this has been a bottleneck in preclinical in silico evaluation of advanced systems including the artificial pancreas. This work presents a method of using continuous glucose monitoring and insulin pump data to extract a BG variability signature represented by oral carbohydrate net effect, which can be "fed" back into the mathematical model to (i) reproduce the original BG time series from the original record of insulin delivery and (ii) be used to approximate the effect of a modified schedule of insulin delivery. We provide details of the basic method and illustrate the approach using both the Virginia/Padova Type 1 Simulator and human-subject data collected in a field study.