Newsletter Spring 2019 - Prasad Palthur, Artificial Pancreas Device Systems – An Evolving Approach and Research Pathway

Artificial Pancreas Device Systems – An Evolving Approach and Research Pathway 

M Prasad Palthur, Ph.D. 

Co-founder & VP, Design & Development, Innoneo Health Technologies, Inc. 



Artificial Pancreas (AP) 

Artificial Pancreas Device Systems (APDS) 

Continuous Subcutaneous Insulin Delivery (CSII) 

Diabetes Mellitus (DM) 

Juvenile Diabetes Cure Alliance (JDCA) 

Juvenile Diabetes Research Foundation (JDRF) 

Real-time Continuous Glucose Monitoring (rtCGM) 

Type 1 Diabetes Mellitus(T1DM) 

U.S. Food and Drug Administration (FDA) 


Biomimetics is an interdisciplinary field thatcomprises the adaptations and derivations of biological functions, structures, and principles of various objects from nature, as well as the designs and fabrications of materials and devices that useartificial mechanisms tomimic natural ones.1 One such biological function explored in biomimetic drug delivery is the homeostasis of blood glucose, in which principles from engineering, information technology, chemistry, and biology are applied to the genesis of device systems thatartificially mimic the biological processes of blood glucose homeostasis, more specifically glucose-stimulated insulin secretion.  


However,mimicking any biological processes is not simple.Biological processes in the human body are generally complex and non-linear. The maintenance of glucose homeostasis within a narrow physiological range is an essential component of human metabolism, and the dynamics of blood glucose homeostasis are considerably complex at both the cellular and systemic level. 


In a healthy body, glucose homeostasis is a highly sophisticated networkthat includes, but is not limited to, various signaling molecules, signaling pathways, the central nervous system, and hormones and neuropeptides released from the brain, pancreas, liver, intestine, adipose tissue, and muscle tissue. Our knowledge of these complex physiological mechanisms is still expanding.Within this network, the pancreas playsa key roleby secreting the hormones that regulate blood sugar.2  


The pancreas is a complex organ with exocrine and endocrine components.3The endocrine cells are groupedto form the islets of Langerhans and account for only 1–2% of the entire organ. Through its various endocrine hormones, particularly insulin and its antagonistglucagon, the pancreas maintains blood glucose levels within a narrow range. The glucose-triggered stimulus-secretion coupling is an established paradigm of insulin secretion from β-cells in the pancreas and includes a variety of modulators that trigger, potentiate, or inhibit glucose-stimulated insulin secretion.2 


Disturbances in blood glucose homeostasis mechanisms may lead to improper human metabolism, notably diabetes mellitus (DM). DM is a multifactorial disease affecting increasing numbers of patients worldwide. DM is not a disease of improper blood sugars, but a disease of improper metabolism.In 2017, 451 million people (aged 18-99 years) around the world were estimated to haveDM. This numberisexpected to increase to 693 million by 2045.4   


Several subtypes of diabetes are identified, of which the most prevalent are referred to as type 2 DM (T2DM) and type 1 DM (T1DM). Most diabetic patients have T2DM (90 percent of cases of diabetes), resulting from a combination of impaired insulin action and possibly insufficient β-cell function. T1DM, also known as insulin-dependent DM (5-10 percent of cases ofdiabetes),is an auto-immune disorder in which the immune system targets and irreversibly destroys the insulin-producing β-cells in the islets of Langerhans of the pancreas.5Progression to insulin-dependent DM is characterized by the loss or dysfunction of pancreatic β-cells, but the pathomechanisms underlying β-cell failure are still poorly defined.6 Much remains to explore the causes and triggers of T1DM in humans.  


T1DM is an absolute deficiency of insulin secretion andrequires daily (or continuous) external insulin injections to maintain carbohydrate metabolismExternal insulin replacement, through multiple daily injections (MDI) or continuous subcutaneous insulin delivery (CSII) using insulin pumps, is a therapeutic norm in T1DM management.However, glycemic control through exogenous insulin replacement or supplementation is not as efficient as endogenous insulin secretion. Data from the T1D Exchange registry demonstrate that only a minority of adults and youth with T1D in the United States achieve ADA goals for HbA1c.7Consequently, people with T1DM face a life-long optimization problem to maintain strict glycemic control and reduce hyperglycemia without increasing their risk for hypoglycemia. Blood glucose level is both the measurable result of this optimization and the principal feedback signal to the patient incontrolling theirDM 


This understanding of diabetes optimization objectives led to a quantitative description of the glucose-insulin control network, modeling, and simulation, and, ultimately, to bioengineering the control of diabetes.8Controlled drug delivery systems can improve the efficacy and safety of therapeutics by optimizing the duration and kinetics of release. Among these systems, closed-loop delivery strategies, also known as self-regulated administration, have proven to be a practical tool for homeostatic regulation by tuning drug release as a function of bio-signals relevant to physiological processes.9 Closed-loop drug delivery promises autonomous control of pharmacotherapy through the continuous monitoring of biomarker levels. For decades, researchers have strived for portable closed-loop systems capable of treating ambulatory patients with chronic conditions,includingDM. Closed-loop glucose control and continuous glucose monitoring biosensors make up the overwhelming majority of published closed-loop drug delivery research.10  


Inthe past four decades, advances in closed-loop delivery strategies and overalldevelopment of diabetes technologies have progressed remarkably through CSII, mathematical models and computer simulation of the human metabolic system, real-time continuous glucose monitoring (rtCGM), and control algorithms driving closed-loop control systems known as the “artificial pancreas” (AP).8 




While the appellation “artificial pancreas” may imply an implanted pancreassubstitute that functions like a biological pancreas, the developmental approach of “artificial pancreas” in its current statedoes not encompass the comprehensive exocrine and endocrine functions of the pancreas. Various terms representing AP cited in literature include: “bionic pancreas,” “closed-loop insulin delivery system,” “hybrid closed-loop insulin delivery system,” “closed-loop control of glucose levels,” hybrid closed-loop system at home,” “automated insulin delivery (AID) systems, and “ambulatory pancreas devices.”11 The term “artificial pancreaseis an over-encompassing misnomer for insulin delivery systems whose control algorithms may or may not deliver all requisite insulin.12 


The current concept of AP represents more the development of a series of increasingly sophisticated cyber-electromechanical systems that will automate the delivery of insulin and possibly other pancreatic hormones forpeople with T1D. Several types of AP are being developed using various approaches, including electromechanical and bioengineered approaches. The bioengineering approach targets to implant encapsulated α- and β-cells in a matrix of bioengineered tissue that would react to body glucose levels and release insulin. An electromechanical APis essentially an automated closed-loop controlled drug delivery system comprised of a glucose sensor, pump to deliver exogenous pancreatic hormones, and an embedded controller to calculate the required doses. 


Consensus on the common terminology and vocabulary that pertain to the functionality of such systems should be implemented before various terms describing AP concepts are routinely used. 11Harmonization and clarification of terminologies are required to distinguish systems that automate insulinor future anticipated multi-hormonesto different degrees as we move toward the development of device systems that fully and automatically control all aspects of delivery per closed-loop algorithms.10 As industry and academic research move forward, and as we hope to have several new systems become available shortly, it is important for regulators, academia, and industry partners to standardize terminologies so that the end users have a clear understanding of the context of different AP systems in the market.13Significant progress has been made, and the safety and feasibility of AP systems have been demonstrated in the clinical research center and more recently in outpatient “real-world” environments.14 


For brevity, this article elaboratesonly on electromechanical AP systems. In today’sapproach, the concept of electromechanical APis broadly used to represent any device system that automates the calculation and administration of insulin delivery based on sensor glucose sensing, ranging from hybrid closed-loop system to anticipated fully automated closed-loop, bi-hormonal (insulin-glucagon), and triple-hormonal (insulin-amylin-glucagon) delivery systems. As it stands today, the concept of electromechanical APrepresents a medical device approach.The medical device approach involves an integrated system by combiningrtCGM and an implanted/wearable pump for CSII that can function together with a computer-controlled algorithm to enable an automated closed-loop controlled drug delivery system for blood glucose monitoring and pancreatic hormone(s) delivery and to better mimic the physiological mechanisms of glucose homeostasis (i.e., glucose-stimulated insulin secretion). Thus, by definition, the approach of electromechanical AP will, theoretically, free patients with diabetes from the responsibility of self-management of their condition.15   


Various “closed-loop systems encompassing varied automated approaches (hybrid-to-fully automated), device systems, hormonal delivery approaches (mono-to-triple hormones), and methodologies and algorithms that control hormone release are being developed.These systems will evolve to increase automation and sophistication. 


For this article, the term “artificial pancreas device systems” (APDS) is used to generalize the context of integrated systems and technologies of electromechanicalAP approaches.The overarching goal of APDS is the development of innovative technologies and systems that enable an integrated, wearable/implantable, and more accurate glucose-regulated closed-loop insulin/pancreatic hormone delivery systems to achieve and sustain daily euglycemia management andprevent acute and chronic complications in a personalized fashion, ultimately relieving patients of the burden of diabetes self-management. In the near term, these APDS will reduce hypoglycemia through the reduction or cessation of insulin delivery, begin to automatically dose insulin to target ranges (hybrid closed-loop systems, hyperglycemia/hypoglycemia-minimizing systems, and semiautomated insulin delivery systems), and eventually dose additional hormones, such as glucagon and/or amylin (dual-hormone AP, and multi-hormone AP).14 



Devices designed to mimic pancreatic endocrine function have been under development since the 1970s.11The history of AP is closely associated with the history of CSII and CGM. Additionally, closed-loop AP systems have been in development for several years. Advances in microelectronics and glucose sensor technology and the FDA’s decision to make the development of closed-loop systems a priority has resulted in rapid progression in this field. There are several AP systems currently under development in both academic and commercial endeavors. Each system offers unique features in the configuration of pumps, glucose sensors, algorithms, single- or dual-hormone delivery functionality, user interface, and data management.  


The Juvenile Diabetes Cure Alliance (JDCA) considers APDS research as one of the four broad research pathways in development that could yield a practical cure for T1D patients. However, APDS should be an exceptionally reliable closed-loop system that is adaptive to each individual.16In 2018, JDCA reports 8.3% of active T1D projects in human trials was related to APs.17 As of September 2018, nine AP device solutions and AP combinations are being evaluated in human trials.17At the time of writing this article,18clinical studies are found to be recruiting patients for AP research.18  


AP systems uniformly improved glucose control in outpatient settings, despite heterogeneous clinical and technical factors.19Recently published meta-analysis suggest that AP systems are superior to the standard sensor-augmented pump treatment of T1D in non-adult people with diabetes.20This hybrid closed-loop technology does not represent a cure for diabetes; however, it holds the promise to allow those living with diabetes to achieve more targeted glycaemic control, thereby reducing the risk of long-term complications.21 Randomized outpatient clinical trials over the past five years have demonstrated the feasibility, safety, and efficacy of the approach, and the FDA approval of the first single hormone closed-loop system(Medtronic 670G automated insulin delivery system11) establishes a new standard of care for people with T1D.22The hybrid closed-loop technology and AP systems remain a feasible therapeutic option pending a biological cure, which would require either regeneration or transplantation of normal β-cells with long-term success. 


FDA is helping to advance the development of an APDS by prioritizing the review of research protocol studies, providing clear guidelines to industry, setting performance and safety standards, fostering discussions between government and private researchers, co-sponsoring public forums, and finding ways to shorten study and review time.23 Meanwhile, government funding bodies in the United States and across the globe, as well as many medical device and digital health companies, have started spending millions of dollars to encourage the development of AP systems and combinations.24Additionally, the Juvenile Diabetes Research Foundation (JDRF), the world’s most prominent nonprofit funder of T1D research,is funding and focusing on AP programsthatcreate miniaturized devices, enable open systems, and drive algorithm advancements.25,26 


The global APDS market is expected to reach $2 billion in 2025, following ~$482 million in 2017. The global APDS market is estimated to grow with a compound annual growth rate of 20.9% from 2016-2025.27Key players involved in the funding and commercial development of APDS include Medtronic Minimed, Inc., Dexcom, Inc., Sensorics, Inc., Abbott Diabetes Care, Inc., Insulet Corporation, Beta Bionics Corporation, Tandem Diabetes Care, Inc., Bigfoot Biomedical, Inc., Tidepool, SFC Fluidics, Inc.,Ypsomed AG,Seventh Sense Biosystems, Xeris Pharmaceuticals, Inc.,Cellnovo Limited,TypeZero Technologies, Inc.,and Admetsys Corporation. 


In addition to the patient, APDS have five fundamental functional components: 

  1. Continuous Glucose Monitoring (CGM) Component 

  1. Continuous Subcutaneous Infusion (CSI) Pump Component 

  1. Control Algorithm (CA) Component  

  1. Communication Pathway (CP) Component 

  1. User Interface (UI) Component 


Figure 1 represents these components. Additional accessories, software, and devices (e.g., power cords, wireless controller, data management platform) may be required to perform specificAPDSfunctions. 



Figure 1.APDS fundamental functional components 

  1. Continuous Glucose Monitoring (CGM) Functional Component 

Blood glucose monitoring is considered an integral component of effective therapy for T1D management. The CGM functional component represents the sensing arm of APDS and performs continuous or repeated measuring of the patient’s blood glucose levels.28CGM data serve as the conditional input for insulin delivery automation devices.29The application of CGM for closed-loop control is advancing rapidly, witha well-documented reduction in risk of hypoglycemia.  


The advent and progress of ambulatory glucose sensor technology have enabled rtCGM and its integration with insulin therapy. Minimally invasive rtCGMhas become the T1D standard of care and includes factory-calibrated subcutaneous glucose monitoring and long-term implantable glucose sensing.30rtCGM uses a small subcutaneously inserted sensor to measure glucose levels in the interstitial fluid,which has been found to correlate with blood glucose levels every few minutes, 24 hours aday. rtCGM provides patients with real-time and trend data, as well as alerts when the blood glucose level is rapidly rising or falling, allowing for better self-management throughout the day.31rtCGM systems are safe and effective in both T1D and T2D and can improve quality of glycemic control, reduce the risk of hypoglycemia, and permit selection of lower target levels for mean glucose and HbA1c.32  


A 2016 consensus statement by the American Association of Clinical Endocrinologists and American College of Endocrinology advocates expanding rtCGM use for patients with diabetes, because it has been found to improve glucose control and reduce the occurrence of hypoglycemic events.33A study comparing standalone rtCGM to self-monitoring of blood glucose (SMBG) in patients with T1DM have found that rtCGM is associated with lower glycated hemoglobin (A1C) levels and does not increase the risk of severe hypoglycemia.31 


Although different techniques for subcutaneous glucose measurement were introduced, only electrochemical transcutaneous CGM systems are currently available to patients. Transcutaneous CGM systems consist of a wired sensor containing glucose-sensing enzymes, a transmitter, and a display device. The wired sensor is placed below the skin in the subcutaneous fat and is continuous with the transmitter base. The transmitter is set in the transmitter base and sends data wirelessly to a display device, such as a dedicated receiver or a smartphone.34 Implantable CGM systems may provide additional ease of use over transcutaneous CGM and as an alternative solution to current transcutaneous CGM, since frequent sensor insertions through the skin is not needed and the transmitter can be removed easily without the need for sensor replacement.34 The most commonly reported obstacles to CGM and rtCGM are cost, number of alarms, inaccuracy, body image concerns related to wearing devices, discomfort, and dermatological complications in adults.35,36 


Integrating CGM data in smartphone applications instead of dedicated devices might further improve the user-friendliness of CGM. Innovation in sensor technology, overall accuracy, and point precision of CGM systems have improved insulin dosing decisions based on CGM data. Improvement in CGM signal filtering levels and calibration algorithms to account for random signal noise and calibration errors have resulted in a significant increase of sensor performance.29 


Most companies have converted their offline CGM downloads to the “cloud,” allowing efficient data access by patients and clinicians. Diabetes data management platforms enable availability of data from devices of different manufacturers available to the patient and clinician through a universal portal or smartphone application.29Automated CGM data analysis and treatment advice would be the next step for CGM data platforms.29 


FDA's premarket approval (PMA) database search indicatesover 280 Class III medical devices under the product categories of the continuous glucose monitoring system, glucose controller, glucose sensor, and glucose biographer.37 

  1. Continuous Subcutaneous Infusion (CSI) Pump Functional Component 

The CSI functional component represents the hormonal delivery arm of APDS and performs continuous infusion of short-acting hormoneseither by mechanical force or nonmechanical pumping technology  

The CSI pump functional component and related accessories include: the infusion delivery mechanism; bolus mechanism; drug reservoir; built-in or external pump tubing and connectors; a user-interface consisting of the programming unit, display unit, and audio and tactile notification units; power supply or pump battery and circuitry to charge the battery; a communication interface, including network components and interfaces to other devices and systems; and approved labeling.28   


It is important to recognize that sensor-integrated pumps differ from sensor-augmented pumpssensor-integrated pumps take action in response to CGM sensor data, whereas sensor-augmented pumpsonlydisplay CGM sensor data and donot take anyaction.12Most conventional CSI pumps, in general, are bulky, intrusive, and expensive.38The major reasons for stopping insulin pump use arebody image with wearing the device, discomfort, cost, and trust.36 


Relatively user-friendly patch pumps have emerged on the market, offering flexible insulin delivery options. The patch pump platform systems provide several advantages over conventional insulin pump delivery systems, including being free of tubing, operating discreetly under clothing, and being small and lightweight.39,40 This nonmechanical pumping technology allows for accurate and precise delivery of minimal amounts of exogenous pancreatic hormones, including concentrated insulin.38Patch pumps are usually attached using an adhesive layer to the skin to attain better blood glucose control with more lifestyle flexibility. They are also cheaper than conventional insulin pumps.40,41  


Emerging patch pump platform systems are starting to occupy a noticeable fraction of the insulin delivery market and are expected to be incorporated into the APDS.For example, the Omnipod DASHinsulin management system is a tubeless, wearable insulin pump that holds up to 200 units of insulin and delivers continuous insulin therapy through customizable basal rates and bolus amounts. The system consists of the Pod, which is a waterproof insulin pump worn on the body, and the personal diabetes manager, which is a handheld device used to control the Pod wirelessly.42   

  1. APDS Control Algorithm (APDS-CA) Functional Component 

The glucose sensing CGM functional component,as well as thehormonal-delivering CSI functional component of APDS,are linked by the controller. The controller is an external processor and will use individualized algorithms to control the delivery of hormones. The controller may be embedded in the pump, handheld device, or smartphone. Robustness and stability of an APDS system explicitly depend on its integrated controller. 


The APDS-CA functional component is software and includes all computational steps, control strategies, control parameters, signal processing approaches, verification methodologies, and safety checks that the system performs to ensure that theinsulin pumpdelivers the appropriate dose.28 Algorithms can be integrated or consist of separate module(s) for safety and glucose regulation.With a closed-loop system, an algorithm from aninitial state of glucose level as supplied by a CGM calculates, through a series of equations using a finite sequence of well-defined instructions, the desired end state, including the quantity and way in which exocrine pancreatic hormonemust be infused by a pump to maintain blood glucose within desired limits.43 


Four algorithms are widely used in APDS 

  1. Proportional-integral-derivative (PID) algorithms: measure glucose and modify insulin infusion rate according to the sampled value's difference from the glucose target point as expressed by proportional, integral, and derivative terms; 

  1. Model predictive control (MPC) algorithms: predict future glucose values based on past trends and accordingly modify the insulin infusion rate; 

  1. Fuzzy logic algorithms: take advantage of the user's or clinician's therapeutic input through CGM; 

  1. Bihormonal algorithms:control relies on both insulin and glucagon infusion.20Bio-inspired algorithms construct amathematical model that mimicksinsulin-producing beta cells making insulin in response to changes in blood levels 


Several control techniques and algorithms are reported in the literature. Additionally, modifications of existing models and algorithms and development ofnew modeling algorithms are reported continuously.44–50  


Efficient and effective algorithms are routinely developed and embedded in various formats for closed-loop control of APDS. Structured collaboration between engineers, mathematicians, computer scientists, data scientists, and clinical researchers isrequired for mathematical modeling, simulation, and formal analysis for the development of efficient algorithms. Future work may involve the achievement of greater sensitivity by factoring specific aspects of body physiology, patient statistics to fine-tune control parameters, algorithm self-learning capabilities, and integration of auxiliary sensors for individualized treatment and treatment adaption over time.29Future algorithms may also incorporate physiological time delays and mechanical delays in the system and estimate model parameters with competent statistical methods.  


It is worth mentioning that open-protocol systems, open-source initiatives, and the adoption of real-world data will impact future APDS software and control algorithms developed for closed-loop delivery. Open protocol systems will support the development of insulin pump systems and CGM systems that allow for control via secure, openly published communication protocols by third-party cellphone-based applications (apps) and other devices. Open protocol systems will also allow individuals or third-party developers to communicate with the insulin pump and CGM via secure, well-documented, and verified communication protocols. With the advent of open protocol systems, the APDS device itself will not need tohave a built-in algorithm nor even a built-in user interface. JDRF launched its Open-Protocol Automated Insulin Delivery (AID) Systems Initiative in 2017 with the goal to explore ways to overcome potential challenges in the use and adoption of open protocol systems, notably helping to establish clear financial, regulatory and legal frameworks.51 


The thriving community of users has supported a patient-driven ecosystem of do-it-yourself (DIY) approaches. In such systems, continuous glucose monitors and insulin pumps are reverse-engineered, allowing open-protocol efforts, such as OpenAPS, AndroidAPS, and Tidepool Loop, to display data in innovative ways and even to control automated insulin delivery.52OpenAPS code framework is designed to work with interoperable insulin pumps and CGMs from any manufacturer to build and facilitate building DIY closed loop implementation.53As of February 2019, OpenAPS estimates over 1,100 individuals around the world with various types of DIY closed loop implementations.54 


AndroidAPS is an application with a code framework tobuild a closed-loop that can communicate with bluetooth-enabled insulin pumps.55 Tidepool Loop is an open source hybrid closed-loop system currently in development for iPhone and Apple Watch.56 Tidepool has also kicked off a project to build and support an FDA-regulated version of Loop intended to work with commercially available insulin pumps and CGMs. 


The algorithms used in commercially available devices have been developed primarily based on first principles, simulation data, and limited real-world data. JDRF aims to identify the areas of algorithm enhancements through big-data analysis to build or improve mathematical constructs and relationships that may be incorporated into next-generation artificial pancreas algorithms and possibly even personalized algorithms.57 

  1. Communication Pathway (CP) Functional Component 

The CP functional component ensures the reliable passage of information between the various device components of the APDS. It includes: communication hardware necessary to communicate information, communication technologies used to transmitdata (e.g., radio-frequency wireless technology, such as IEEE 802.11, Bluetooth, or Zigbee), and communication software necessary to allow and control the passage of information.28 

  1. User Interface (UI) Functional Component 

The UI functional component represents all components of the device system with which the user interacts and plays a vital role in the performance of the APDS. It includes graphical user interfaces, control mechanisms (e.g., keypads, touchscreens), feedback mechanisms (e.g., alarms, indicators, other messages to users), andAPDS labeling.28  


UI functional components are also dependent on the principal configurations of a wearable APDS,such as embedded configuration (e.g., onboarding insulin pump) or independent mobile device configuration (e.g., smartphone). Irrespective of the configuration, the APDS development process should include human factors/usability testing to ensure that the device system will be safe and effective in the hands of the intended users.Adopting user research (UX) research methodologies will help ensure that new APDS devices are reliable, easytouse, and meet the needs of users.58Proper UI/UX design for AI helps increase patient adherence and allows clinicians to manage the stream of data and insights generated. 


Although the fundamental components described above are common to all APDS, different device designs, control algorithms, and patient management strategies create the potential for different APDS types.Various modes of APDS operation can be discriminated, ranging from fully automated administration systems requiring virtually no user input to hybrid closed-loop systems requiring frequent user input, such as a meal or exercise announcement.29  

  1. Classification based on the APDS technology generation 

JDRF has defined six categories of closed-loop APD technology based on the level of automation: 

  • Stages 1-3: first-generation systems, which arenon-closed-loop systems and focus on preventing unsafe high and low blood sugar levels by maintaining blood sugar between approximately 70 and 180 mg/dL.  

  • Stages 4-5: second-generation systems, which are automated insulin-alone delivery (AID) systems. Stage 4 systems are hybrid closed-loop devices, which are always closed-loop with mealtime manual assist bolus. Stage 5 systems are fully automated AIDs.  

  • Stage 6: third-generation systems, which are fully automated multihormonal (MH) delivery devices, in which a secondary glucoregulatory hormone, such as glucagon or amylin,are used in addition to insulin.59 

  1. Classification based on the type of computer algorithm 

Another way that APDS can be classified is according to the kind of computer algorithm that the controller uses. There are four main types of control algorithms used in closed-loop APD systems, which were discussed previously: 

  1. Model predictive control (MPC); 

  1. Proportional integral derivative (PID); 

  1. Fuzzy logic (FL); and 

  1. Bio-inspired algorithms. 


In addition to using CGM data, some APD systems measure other biometric/physiological fluctuations (e.g., galvanic skin response), and these are known as multivariable or adaptive systems.60 

  1. Classification based on the approach used for achieving glycemic control 

The third way of classifying APDS is based on the strategy used for achieving glycemic control 

  1. Threshold Suspend Device Systems (Low Glucose Suspend Device Systems): aim to reduce the likelihood or severity of a hypoglycemic event by suspending or reducing insulin delivery temporarily when the sensor value reaches or approaches (reactive or predictive, respectively) a predetermined lower threshold of measured interstitial glucose;28  

  1. Control-to-Range (CTR) systems: reduce the likelihood or severity of a hypoglycemic or hyperglycemic event by adjusting insulin dosing only if a person's glucose levels reach or approach predetermined higher and lower thresholds;28,60 

  1. Control-to-Target (CTT) systems: set target glucose levels and tries to maintain these levels  automatically.28,60 

  1. Classification based on drugs delivered for achieving glycemic control 

Depending on the drug(s) being delivered, APDS can be classified as (or sub-classified of CTR and CTT system subtypes): 

  1. An insulin-only system that delivers insulin; 

  1. A bi-hormonal control system that delivers two different hormonesone hormone (insulin) to lower glucose levels and another hormone (such as glucagon);  

  1. Amulti-hormonal control system delivers three different hormonesinsulin, glucagon, and amylin.   


A fully automated, triple-hormone, closed-loop system that delivers insulin, pramlintide, and glucagon to control glucose levels compared to an insulin-alone closed-loop system with carbohydrate-matched bolus is currentlyinvestigated.61 

  1. Classification based on the mode of implementation 

Research done in the past few years established the two principal implementations of a wearable APDS: an embedded-system with a control algorithm running on the board of an insulin pump, and a mobile- system based on a smartphone as a computational and communication hub.8 Both configurations have their advantages and disadvantages, and both are likely to coexist as a means for automated hormonal delivery.8 


Various closed-loop APDS systems are developed through clinical stages of research, and they employ different combinations of hormonal approaches, control algorithms, and glycemic control strategies.60Many next-generation AP systems are already in development. The modular architecture of AP systemson both the hardware and software levelallows APDS to be assembled from independent but compatible modules, each performing a specific function.6  Also, modular design allows for accommodation of rapidly changing diabetes device technology and algorithms and further streamlines AP use in clinical settings.62  


This modular architecture may also require different types of partnerships or business ecosystemsand mayrepresent new roles of digital health companies to increase the adoption of advanced technologies and accelerate innovation. For example, Bigfoot Biomedical, Inc., hasagreedto integrate Abbott's FreeStyle® Libre glucose sensing technology and Dexcom, Inc., CGM system data with Bigfoot's insulin delivery solutions in the United States.63,64Senseonics Holdings, Inc., and TypeZero Technologies, Inc., are working on integrating SenseonicsEversense®  long-term implantable CGM system with TypeZero’sinControl AP algorithms and Roche Diabetes Care’s Accu-Chek® Insight insulin pump.65Tandem Diabetes Care, Inc., recently approved t:slim X2 Insulin Pump with Basal-IQ Technology consisting of t:slim X2 Insulin Pump and compatible Dexcom G5 mobile CGM.66 A recent study reported the design, development, and testing of the interoperable artificial pancreas system (iAPS) smartphone app that can interface wirelessly with leading CGM, insulin pump devices, and decision-making algorithms while running on an unlocked smartphone.67 


The ultimate metric for success of APDS will be improved outcomes for people with diabetes. APDS accessibility will be driven by the value perceived by patients, health care providers, and payers.13 However, advances in APDS technologies and realizing their full potential aredependent on the structured collaborations among all stakeholders and a streamlined alignment to overcome barriers in widespread implementation. Intensive partnerships among funding bodies, commercial entities, research institutions, regulatory agencies, payors, and not-for-profit organizations are essential. This modular architecture of APDS also might require different types of partnerships or business ecosystems to increase the adoption of advanced technologies and accelerate innovation. 


Figure 2 represents the conceptual framework of the APDS ecosystem. The proposed conceptual framework is to describe a construct representing 

  1. APDS functional components;  

  1. Elements that influence the research and development of APDS functional components;  

  1. Key actors & stakeholders involved in the APDS ecosystem;  

  1. Expression of the complex business environment to translate research into APDS products; 

  1. Digital ecosystem components that are either developments of APDS or impact the future advancements of APDS; and  

  1. Key elements that influence adoption & success of APDS.