Bio-Behavioral Monitoring and Control of IDDM

在过去的几年里，我们在不增加严重低血糖(SH)风险的情况下，对优化血糖控制的临床问题进行了数学描述，并成功地开发了特定于糖尿病的数学方法，量化了1型糖尿病(T1 DM)的控制方面。我们的成就包括：1)低血糖指数--迄今为止预测SH的最佳指标，占随后6个月SH发作的46%；2)预测50%的即将到来的中/重度低血糖；3)胰岛素-血糖动力学网络模型，解释高胰岛素钳夹期间血糖波动的95%。我们发表了15篇科学论文。基于这些进展，我们提出了一项为期四年的研究，将利用现有数据并进行新的数据收集，目的是创建理论模型和先进的信息处理方法，在几个水平上评估患者的代谢控制。阶段1(1-2年)将量化T1 DM患者的行为和生理之间的相互作用。我们将设计、测试和验证一种新的生物行为模型，并伴随着Palm Pilot类型计算机的行为评估协议，该模型将收集有关低血糖/高血糖的意识和检测以及自我治疗判断/行为的现场数据。第二阶段(第2-3年)是一项实地研究，将利用自我监测血糖(SMBG)数据开发和验证三种算法，以估计HbA1c、SH的长期风险和即将发生的(24小时内)低血糖的风险。我们将设计和测试一种新的增强型SMBG采样程序，基于SMBG的效用不仅取决于其频率，而且取决于测量时间的思想。第三阶段(第3-4年)是在普通临床研究中心进行的一项研究，将通过网络模型调查高胰岛素钳夹和受控医院条件下的血糖动态。具体地说，我们将研究胰岛素和葡萄糖、体力活动和心率之间的功能关系，以及心率变异性和激素逆向调节。在本项目结束后，我们将开发出理论模型和量化方法，将：1.提供导致T1 DM控制不佳的特殊行为的评估(如高HbA1c和/或SH风险增加)，从而促进T1 DM患者的行为训练；2.在两个基本水平上处理和解释SMBG数据，估计HbAt和长期低血糖风险；3.每当检测到高血糖风险时，提供增强的SMBG和对即将到来的低BG的监测；4.提供连续的BG监测数据的动态解释，目标是在30分钟内预测BG波动，特别是低血糖风险。

During the last the years, we have formulated mathematically the clinical problem of optimization of glycemic control without increasing the risk of severe hypoglycemia (SH), and successfully developed diabetes- specific mathematical methods quantifying aspects of the control of Type 1 Diabetes (T1DM). Our accomplishments include: 1) the Low BG Index - the best predictor of SH to date, accounting for 46% of SH episodes in the subsequent 6 months; 2) prediction of 50% of upcoming moderate/severe hypoglycemia, and 3) a network model of insulin- glucose dynamics accounting for 95% of blood glucose (BG) fluctuations during a hyperinsulinemic clamp. We published 15 scientific papers. Based on these advances we propose a four-year study that will utilize existing data and will perform new data collections with the goal to create theoretical models and advanced information processing methods evaluating patients' metabolic control at several levels. Phase 1 (Years 1- 2) will quantify interactions between behavior and physiology in T1DM. We will design, test, and validate a new bio-behavioral model, accompanied by a behavior-assessment protocol for Palm Pilot-type computers, that will collect field data on awareness and detection of hypoglycemia/hyperglycemia, and self-treatment judgment/behavior. Phase 2 (Years 2-3) is a field study that will develop and validate three algorithms using self-monitoring BG (SMBG) data to estimate HbA1c, long-term risk for SH, and risk for imminent (within 24 hours) hypoglycemia. We will design and test a new Enhanced SMBG sampling procedure, based on the idea that the utility of SMBG greatly depends not only on its frequency, but also on the timing of measurement. Phase 3 (Years 3-4) is a study, performed at the General Clinical Research Center that will investigate, through a network model, BG dynamics during both hyperinsulinemic clamp and controlled hospital conditions. Specifically, we will investigate functional relationships between insulin and glucose, physical activity and heart rate, and heart rate variability and hormonal counterregulation. Upon the conclusion of this project, we will have developed theoretical models and quantitative methods that will: 1. Provide assessment of idiosyncratic behaviors leading to poor control of T1DM (as represented by high HbA1c and/or increased risk for SH), thus facilitating behavioral training of patients with T1DM; 2. Process and interpret SMBG data at two basic levels, estimation of HbAt, and long-term risk of hypoglycemia; 3. Offer Enhanced SMBG and monitoring of upcoming low BG, whenever high risk of hypoglycemia is detected, and 4. Provide dynamic interpretation of continuous BG monitoring data with the goal to forecast the BG fluctuations, and especially the risk for hypoglycemia, within 30 minutes.