China
Africa
Explore chapters and articles related to this topic
Leveraging Plasma Insulin Estimates and Wearable Technologies to Develop an Automated Insulin Delivery System in Type 1 Diabetes
Published in Emmanuel Opara, Controlled Drug Delivery Systems, 2020
Iman Hajizadeh, Mudassir Rashid, Sediqeh Samadi, Mert Sevil, Nicole Hobbs, Rachel Brandt, Mohammad Reza Askari, Ali Cinar
Accurate estimates of PIC can be obtained by using CGM measurements with adaptive estimators designed for simultaneous state and parameter estimation based on reliable glucose–insulin models [18–22]. In our previous studies, the design of adaptive and personalized PIC estimators that directly take into account the inter- and intrasubject variabilities in glucose–insulin dynamics is investigated using three different estimation techniques, including continuous-discrete extended Kalman filter (CDEKF), unscented Kalman filter (UKF), and moving horizon estimation (MHE) [21]. The results are based on clinical experiments conducted with adolescents at the Yale Children’s Diabetes Clinic (New Haven, CT) involving 13 datasets from subjects with T1DM. We further analyzed the performance of the proposed individualized PIC estimation algorithm using 20 clinical datasets from closed-loop experiments conducted continuously over 60 h involving young adults with T1DM at the Kovler Diabetes Center, University of Chicago Medical Center, Chicago, IL [22]. The diversity of the subjects and the length of the clinical experiments allowed for a more comprehensive and critical evaluation of the performance of the PIC estimation method that will be incorporated into the AP system. The significant variability in data for various subjects is due to the different meals (amount and type of carbohydrate intake), varied daily basal rates, varied bolus insulin infusions, physical activity (PA) levels, and sleep characteristics. The discrete sampling nature of the CGM measurement output, the lack of knowledge on the exact time and amount of meals, the time-varying nature of human physiology, the unmeasured disturbances caused by exercise and sleep, the intersubject variability, the constraints on the state variables, and the rate of change of the model parameters are some of challenges addressed in this work. The PIC estimator is individualized using readily available demographic information, such as body weight, height, body mass index (BMI), and total daily insulin dose. The PIC estimation results are compared against those obtained through the conventional IOB curves to demonstrate the merits of the proposed individualized PIC estimator [21,22].
A homotopy-based moving horizon estimation
Published in International Journal of Control, 2019
Mohammad Abdollahpouri, Rien Quirynen, Mark Haring, Tor Arne Johansen, Gergely Takács, Moritz Diehl, Boris Rohaľ-Ilkiv
Moving horizon estimation (MHE), as one of the candidates for online constrained state estimation, can deal with nonlinear dynamics explicitly and directly incorporate the physical or logical constraints in its design. The MHE problem formulation is an approximation of a full information-constrained estimation technique. It considers a fixed number of measurements inside a moving time window and this makes it tractable in practice, which is almost impossible for a full information estimation approach. The arrival cost term is typically introduced in the MHE formulation to represent the truncated data; see the explanation by Rao, Rawlings, and Mayne (2003) for more details on MHE. Its increased performance compared to classical techniques, such as Kalman filtering, has been studied extensively on several case studies; for example, see the results on chemical engineering systems by Haseltine and Rawlings (2005) and on vibration systems by Abdollahpouri et al. (2016) and Abdollahpouri, Takács, and Rohal’-Ilkiv (2017). There are several factors that make MHE quite practical in industry compared to the traditional nonlinear estimation techniques (Johansen, 2011), including N samples of measurements in the moving time window and incorporating the physical and logical constraints in a dynamic optimisation framework. One of the challenges that needs to be addressed from an optimisation point of view, is to find a good local solution (depending on the case study) or ultimately to obtain a global solution to a non-convex problem reliably. Most available algorithms for treating non-convex optimisation problems are gradient based and end up with a local minimum. Suboptimal solutions not only degrade the quality of estimates, they can be misleading as well, e.g. in a structural health monitoring system, by utilising the incorrect state and parameter estimates.