Development of a miniaturized, single-port automated insulin delivery system utilizing a glucose sensing catheter, ultra-concentrated insulin, and an optimized control algorithm

摘要 意义：美国有500多万接受胰岛素治疗的糖尿病患者 在可归因于糖尿病的2370亿美元直接医疗费用中，这一比例高得不成比例。对.的使用 连续血糖监测(CGM)已被证明可以降低HbA1c水平，而HbA1c已被证明是健康的预测指标 在这一人群中的结果，当CGM与CGM相结合时取得了最大的改善 持续皮下胰岛素输注(CSII)。最近CGM和胰岛素泵的融合已经 启用了第一代自动胰岛素递送(AID)系统，承诺会有更好的血糖 胰岛素治疗糖尿病的对照。然而，当前的AID系统复杂、笨重且昂贵 因为他们需要在身体上佩戴多种设备：葡萄糖传感器，胰岛素 泵和胰岛素输送导管。我们已经开发了一种葡萄糖传感导管，它可以减少 皮下组件的数量从2个减少到1个，显著降低了这些组件的大小和复杂性 系统。通过将传感导管集成到补丁泵中，我们将通过减少 占地面积，并实现了第一个完全集成的单组件AID系统。结果减少了 系统规模、复杂性和成本将提高AID的采用率，有助于提高合规性、降低 糖化血红蛋白水平，并改善1型糖尿病患者的健康状况。初步数据：我们有 证明了葡萄糖传感可以在胰岛素输送部位进行。然而，我们发现， 在分配液体(胰岛素或生理盐水)后立即出现血糖测量伪影 很可能是由周围间质液体稀释引起的。初步数据显示， 伪影的大小与药丸的体积有关。具体目标：这项提议是第一个 将两用插管与微型补丁泵和AID算法相结合的努力的阶段， 利用一种新的浓缩胰岛素来减少CGM伪影尺寸。在具体目标1中，我们将减少 使用超浓缩胰岛素对稀释伪影的影响。我们将描述 在一项猪研究中，这件文物使用了较小剂量的U500胰岛素。在特定的目标2中，我们将创建一个校准 基于卡尔曼滤波和对未来传感器值的预测模型的算法，以进一步缓解稀释 神器。在具体目标3中，我们将把U-500胰岛素的动力学集成到模型预测控制中 俄勒冈健康与科学大学(OHSU)开发的(MPC)AID算法，并对算法进行优化以 (1)结合Thermarin胰岛素的动力学模型和(2)消除任何剩余的稀释伪影 这一点可能仍然存在。我们将评估新的校准算法在数据上的性能 在目标1中收集的数据也存在于我们使用葡萄糖传感导管收集的其他人体数据中。这个 MPC算法将在Silico中使用OHSU虚拟患者群体进行评估，为完全 在本提案的第二阶段整合到Thermarin StampPump中。总而言之，这种合作努力 在PDT、OHSU和Thermarin之间将唯一发表的安培法葡萄糖传感结合在一起 导管配有容错AID算法和唯一快速、浓缩的胰岛素。简而言之，我们是唯一 该团队目前能够为统一的自动胰岛素输送设备提供这一新颖的解决方案。

ABSTRACT Significance: There are over 5 million people with insulin-treated diabetes in the United States who represent a disproportionately large share of the $237B in direct medical costs attributable to diabetes. The use of continuous glucose monitoring (CGM) has been shown to reduce HbA1c levels, a proven predictor of health outcomes within this population, with the greatest improvement achieved when CGM is coupled with continuous subcutaneous insulin infusion (CSII). The recent convergence of CGM and insulin pumps has enabled the first generation of automated insulin delivery (AID) systems, promising even better glycemic control for insulin-treated diabetes. However, current AID systems are complex, cumbersome, and expensive for the patient because they require multiple devices to be worn on the body: a glucose sensor, an insulin pump, and an insulin delivery catheter. We have developed a glucose sensing catheter that reduces the number of subcutaneous components from two to one, significantly reducing the size and complexity of these systems. By integrating the sensing catheter into a patch pump, we will further simplify the system by reducing the footprint, and enabling the first fully-integrated single component AID system. Resulting reductions in system size, complexity, and cost will increase adoption rates for AID, helping improve compliance, lower HbA1c levels, and improve health outcomes among people with type 1 diabetes. Preliminary Data: We have demonstrated glucose sensing can be performed at the site of insulin delivery. However, we have discovered that there is a glucose measurement artifact that occurs immediately after dispensing liquid (insulin or saline) from the cannula, likely caused by dilution of the surrounding interstitial fluid. Preliminary data suggest the magnitude of the artifact is related to the volume of the bolus. Specific Aims: This proposal represents the first phase of an effort to integrate the dual-use cannula with a miniaturized patch pump and an AID algorithm, taking advantage of a novel concentrated insulin to reduce CGM artifact size. In Specific Aim 1, we will reduce the impact of the dilution artifact through the use of ultraconcentrated insulin. We will characterize the impact of the artifact using smaller boluses of U500 insulin in a swine study. In Specific Aim 2, we will create a calibration algorithm based on a Kalman filter and a predictive model of future sensor values to further mitigate the dilution artifact. In Specific Aim 3, we will integrate the kinetics of the U-500 insulin into a model predictive control (MPC) AID algorithm developed by Oregon Health & Science University (OHSU) and optimize the algorithm to (1) incorporate a model of the kinetics of the Thermalin insulin and (2) eliminate any remaining dilution artifact that may still be present. We will evaluate the performance of the new calibration algorithm on the data collected in Aim 1 as well is in other human data that we have collected with the glucose sensing catheter. The MPC algorithm will be evaluated in silico using the OHSU virtual patient population in preparation for full integration into the Thermalin StampPump in phase 2 of this proposal. In summary, this collaborative effort between PDT, OHSU, and Thermalin brings together the only published amperometric glucose sensing catheter with a fault-tolerant AID algorithm and the only rapid, concentrated insulin. In short, we are the only team currently capable of providing this novel solution for a unified automated insulin delivery device.