China
Africa
Explore chapters and articles related to this topic
Statistical Approaches in the Development of Digital Therapeutics
Published in Oleksandr Sverdlov, Joris van Dam, Digital Therapeutics, 2023
Oleksandr Sverdlov, Yevgen Ryeznik, Sergei Leonov, Valerii Fedorov
A hallmark of precision medicine is the development of various artificial organs to replace, recover, and enhance the functioning of the human body. One example is the artificial pancreas system for patients with type 1 diabetes (Elleri et al., 2011). The system performs continuous glucose monitoring (CGM), autonomously calculates and administers the appropriate insulin dose for the patient. In September 2016, the FDA approved the first fully autonomous artificial pancreas system to manage type 1 diabetes in patients 14 years or older.8 More such systems are being developed and tested in clinical trials (Trevitt et al., 2016; Benhamou et al., 2019; Brown et al., 2019). The artificial pancreas is an example of an autonomous DTx for cases where disease mechanisms are well established. In the above example, control of blood glucose level for a given patient can be achieved by delivering insulin at the right time and at the right dose, calculated using advanced mechanistic models based on the person's CGM data and the respective computationally effective algorithms. The rapidly evolving open- and closed-loop (autonomous) treatments provide promising results in other areas, such as HIV/AIDS disease management (Costanza et al., 2009).
AI/ML in Medical Research and Drug Development
Published in Wei Zhang, Fangrong Yan, Feng Chen, Shein-Chung Chow, Advanced Statistics in Regulatory Critical Clinical Initiatives, 2022
Type 1 Diabetes Mellitus (T1DM) is another form of chronic disease that can potentially benefit from RL approaches. In particular, the concept of Artificial Pancreas (AP) [Cobelli et al., 2011, Albisser et al., 1974] that is used in blood glucose control by computing and administrating appropriate doses of insulin, which includes a Continuous Glucose Monitoring (CGM) device and an insulin pump, can be thought of as a closed-loop control system that fits well into a reinforcement learning algorithm. The blood glucose management with minimizing hypoglycemia events (an event caused by deficiency of blood glucose in blood stream that can result in dizziness, shakiness or even passing out in severe cases) involves a high between and within patient variability. Moreover, there are various parameters such as lifestyle, eating behavior, amount of exercise and many other factors that play an important role in this control problem. Therefore, a well defined model to describe the problem or a one rule fits all type of approach is almost impossible in this case and rather it a model free approach, in statistical term, tailored to each patient is what is required here. An RL-based algorithm can provide different optimal policies for different patients and be a potential solution to this problem.
Meal Detection Module in an Artificial Pancreas System for People with Type 1 Diabetes
Published in Emmanuel C. Opara, Sam Dagogo-Jack, Nutrition and Diabetes, 2019
S. Samadi, K. Turksoy, I. Hajizadeh, J. Feng, M. Sevil, C. Lazaro, N. Hobbs, R. Brandt, J. Kilkus, E. Littlejohn, A. Cinar
Artificial pancreas (AP) control systems aim to achieve automated regulation of blood glucose concentration in people with type 1 diabetes (T1D). While in open-loop insulin administration, the individual, with physician supervision, makes all the decisions regarding the insulin adjustments. AP is a closed-loop system that automatically manipulates the insulin doses and infusion [1–6]. An AP system consists of a minimum of four components: (1) a person with T1D, (2) a continuous glucose monitor (CGM) that measures and reports the interstitial glucose concentration, (3) a control algorithm that makes decisions regarding the insulin infusion rate, and (4) an insulin pump that administers the insulin to the patient. The AP may utilize additional measuring devices in order to feed more comprehensive information to the controller about the subject’s physiological condition [7]. The physiological conditions, such as physical activity or exercise, stress, and sleep, are assessed by the interpretation of physiological variables reported by wearable devices, such as heart rate, energy expenditure, etc. [8–10]. Using the measured physiological variables in addition to CGM data enables a multivariable AP with a controller to manage the effect of otherwise unknown (unannounced) disturbances [11]. In a fully automated AP (no manual entries), while the measured physiological variables relieve the need for manual entries about the physiological conditions, integration of meal detection and meal size estimation components into AP systems is needed to compensate for meal announcements.
Emerging drugs for the treatment of type 1 diabetes mellitus: a review of phase 2 clinical trials
Published in Expert Opinion on Emerging Drugs, 2023
Future care of established T1DM would include precision insulin therapy, non-injectable insulin formulations, and fully automated insulin pumps. Phase 3 clinical trials of once-weekly analog insulins as well as oral insulin capsule are currently ongoing, and their availability for clinical use ison the horizon. Once-weekly ultra-long-acting insulin would reduce the burden associated with MDI therapy, while an oral insulin capsule would bring a significant impact for those with needle phobia. Fully-automated insulin pump therapy is also in development and iLet Bionic insulin pump (artificial pancreas) has received breakthrough device status. Artificial pancreas has potential to maintain tight glycemic control without the expanse of hypoglycemia. In summary, T1DM has proved to be more resistant to therapeutic intervention either conventional or experimental approach, whether the therapeutic goal is disease prevention or reversal. Results of ongoing phase 3 clinical trials of teplizumab and ustekinumab are eagerly awaited. Regarding therapy for established diabetes, substantial advances have been made in modern insulin therapy and technologies using continuous blood glucose monitoring and insulin pump to create artificial pancreas.
Artificial pancreas systems: experiences from concept to commercialisation
Published in Expert Review of Medical Devices, 2022
David L. Rodríguez-Sarmiento, Fabian León-Vargas, Maira García-Jaramillo
Recently, automated insulin delivery (AID) systems known as artificial pancreas or closed-loop glucose control systems have been developed to safely and automatically control glycemia owing to the integration of the following components: a subcutaneous insulin infusion pump, subcutaneous continuous glucose monitor (CGM), and control algorithm that continuously computes the amount of insulin to be infused into the patient. Using the subcutaneous route for both glucose sensing and insulin delivery is preferred over the intravenous (IV) and intraperitoneal (IP) routes, owing to the convenience and safety of outpatient use. The IV route can lead to frequent catheter problems such as intravenous migrations and fibrin clot obstructions, while the IP route needs to be implanted and may increase the production of anti-insulin antibodies [2].
MiniMed 670G hybrid closed loop artificial pancreas system for the treatment of type 1 diabetes mellitus: overview of its safety and efficacy
Published in Expert Review of Medical Devices, 2019
Aria Saunders, Laurel H. Messer, Gregory P. Forlenza
The idea of an artificial pancreas or closed-loop system for use in diabetes management has been an exciting concept for many years. However, it is important to consider that there may be a large gap between a patient’s expectation versus reality of the current technology [44]. Many patients and even clinicians have a concept of what an artificial pancreas should be and, while researchers are working hard to make those wishes become reality, a fully closed-loop artificial pancreas system has still not been fully attained. The MiniMed 670G system, along with other hybrid closed-loop systems that are currently undergoing clinical trials, still requires the user to input meal boluses, as neither the sensor nor the insulin is able to react quickly enough to counteract the rapid rise in blood glucose caused by carbohydrate intake. Additionally, hybrid closed-loop systems will not keep patients in the ideal blood glucose range 100% of the time. While this is the ideal, occurrences of both hypoglycemia and hyperglycemia are possible and somewhat inevitable, requiring users to intervene in order to return to normal glucose levels.