Real-time continuous glucose monitoring (CGM), augmented with accurate glucose prediction, offers an effective strategy for maintaining blood glucose levels within a therapeutically appropriate range. This is particularly crucial for individuals with type 1 diabetes (T1D) who require long-term self-management. However, with extensive glycemic variability, developing a prediction algorithm applicable across diverse populations remains a significant challenge. Leveraging meta-learning for domain generalization, we propose GPFormer, a Transformer-based zero-shot learning method designed for multi-horizon glucose prediction. We developed GPFormer on the REPLACE-BG dataset, comprising 226 participants with T1D, and proceeded to evaluate its performance using three external clinical datasets with CGM data. These included the OhioT1DM dataset, a publicly available dataset including 12 T1D participants, as well as two proprietary datasets. The first proprietary dataset included 22 participants, while the second contained 45 participants, encompassing a diverse group with T1D, type 2 diabetes, and those without diabetes, including patients admitted to hospitals. These four datasets include both outpatient and inpatient settings, various intervention strategies, and demographic variability, which effectively reflect real-world scenarios of CGM usage. When compared with a group of machine learning baseline methods, GPFormer consistently demonstrated superior performance and achieved the lowest root mean square error for all the evaluated datasets up to a prediction horizon of two hours. These experimental results highlight the effectiveness and generalizability of the proposed model across a variety of populations, demonstrating its substantial potential to enhance glucose management in a wide range of practical clinical settings.

In type 1 diabetes management, maintaining nocturnal blood glucose within target range can be challenging. Although semi-automatic systems to modulate insulin pump delivery, such as low-glucose insulin suspension and the artificial pancreas, are starting to become a reality, their elevated cost and performance below user expectations is hindering their adoption. Hence, a decision support system that helps people with type 1 diabetes, on multiple daily injections or insulin pump therapy, to avoid undesirable overnight blood glucose fluctuations (hyper- or hypoglycaemic) is an attractive alternative. In this paper, we introduce a novel data-driven approach to predict the quality of overnight glycaemic control in people with type 1 diabetes by analyzing commonly gathered data during the day-time period (continuous glucose monitoring data, meal intake and insulin boluses). The proposed approach is able to predict whether overnight blood glucose concentrations are going to remain within or outside the target range, and therefore allows the user to take the appropriate preventive action (snack or change in basal insulin). For this purpose, a number of popular established machine learning algorithms for binary classification were evaluated and compared on a publicly available clinical dataset (i.e., OhioT1DM). Although there is no clearly superior classification algorithm, this study indicates that, by using commonly gathered data in type 1 diabetes management, it is possible to predict the quality of overnight glycaemic control with reasonable accuracy (AUC-ROC = 0.7).

This paper presents a portable low-power platform that aims to maintain glycaemic control in people with type 1 diabetes. It runs on dedicated hardware, and incorporates a bio-inspired controller that replicates the insulin-secreting physiology of the pancreatic beta cells. The controller runs on an embedded microchip and the system has been optimized for low-power operation. The low-power, hand-held unit (BiAP) interfaces to a continuous glucose sensor (Dexcom G5) and sends the required insulin dose to a subcutaneous pump (Tandem t:slim). An adaptive meal bolus calculator running on an iPhone is used to improve the controller. The system architecture is presented in this paper along with the results of a 4 day hardware-in-the-loop test. The results show that the BiAP platform is robust and capable of achieving good control with no missed glucose values or device disconnections. Finally, the hand-held unit achieves power efficient operation, lasting for 6 days during full control.

In Beacon Virtua [1] you can explore the legacy of the shipwrecked VOC ship Batavia by visiting a simulation of Beacon Island (see Figure 1). Beacon Virtua will take you on a tour of the island including its jetties, fishing shacks and several grave sites of Batavia voyagers who were buried on the island after the ship was wrecked and following the uprising. The graves have been reconstructed through a technique called photogrammetric 3D reconstruction, a process which uses multiple photographs of an object to build an accurate and detailed 3D model of it. Beacon Virtua presents the island as it was in 2013, using audio and photography captured during multiple expeditions to the island to preserve this period in its history. [2] In 2013 there were around 15 shacks located across Beacon Island, originally used by the fishing community. These shacks have been recreated as 3D models, which can be explored inside and out. Around the island are photographic panorama bubbles offering 360° views of the island. These bubbles have been captured using a special panoramic photography process - stepping inside a bubble allows you to see the island from that point exactly as it was in 2013.

People with type 1 diabetes (T1D) rely on exogenous insulin in order to bring blood glucose levels back to healthy levels after a meal. The amount of bolus insulin needed depends on various parameters and recent research has been focused on providing decision support for calculating clinically safe insulin doses. In this demonstration we present a mobile `Advanced Bolus Calculator for Diabetes' for insulin dosing decision support and its operation and notification service through a connected smartwatch. The smartwatch can be used to display glucose information, log diabetes related events and request insulin dose recommendations.

This paper presents the architecture and initial usability results of an advanced insulin bolus calculator for diabetes (ABC4D), which provides personalized insulin recommendations for people with diabetes by differentiating between various diabetes scenarios and automatically adjusting its parameters over time. The proposed platform comprises two main components: a smartphone-based patient platform allowing manual input of glucose and variables affecting blood glucose levels (e.g., meal carbohydrate content and exercise) and providing real-time insulin bolus recommendations; and a clinical revision platform to supervise the automatic adaptations of the bolus calculator parameters. The system implements a previously in silico validated bolus calculator algorithm based on case-based reasoning, which uses information from similar past events (i.e., cases) to suggest improved personalized insulin bolus recommendations and automatically learns from new events. Usability of ABC4D was assessed by analyzing the system usage at the end of a six-week pilot study (n = 10). Further feedback on the use of ABC4D has been obtained from each participant at the end of the study from a usability questionnaire. On average, each participant requested 115 ± 21 insulin recommendations, of which 103 ± 28 (90%) were accepted. The clinical revision software proposed a total of 754 case revisions, where 723 (96%) adaptations were approved by a clinical expert and updated in the patient platform.

This paper presents an advanced insulin bolus advisor for people with diabetes on multiple daily injections or insulin pump therapy. The proposed system, which runs on a smartphone, keeps the simplicity of a standard bolus calculator while enhancing its performance by providing more adaptability and flexibility. This is achieved by means of applying a retrospective optimization of the insulin bolus therapy using a novel combination of run-to-run (R2R) that uses intermittent continuous glucose monitoring data, and case-based reasoning (CBR). The validity of the proposed approach has been proven by in-silico studies using the FDA-accepted UVa-Padova type 1 diabetes simulator. Tests under more realistic in-silico scenarios are achieved by updating the simulator to emulate intrasubject insulin sensitivity variations and uncertainty in the capillarity measurements and carbohydrate intake. The CBR(R2R) algorithm performed well in simulations by significantly reducing the mean blood glucose, increasing the time in euglycemia and completely eliminating hypoglycaemia. Finally, compared to an R2R stand-alone version of the algorithm, the CBR(R2R) algorithm performed better in both adults and adolescent populations, proving the benefit of the utilization of CBR. In particular, the mean blood glucose improved from 166 ± 39 to 150 ± 16 in the adult populations (p = 0.03) and from 167 ± 25 to 162 ± 23 in the adolescent population (p = 0.06). In addition, CBR(R2R) was able to completely eliminate hypoglycaemia, while the R2R alone was not able to do it in the adolescent population.

This demonstration presents the first bio-inspired approach to glucose management of Type-I diabetic patients using a real-time closed-loop insulin delivery system. The delivery system consists of a glucose biosensor, used with the silicon beta-cell to drive a motorized pump. Glucose-induced bursting of beta cells in the pancreas are used to control the insulin secretion in our bodies. A low-power implementation of these metabolic cells in silicon is achieved resulting in efficient glucose control.