Feasibility Study of Automated Overnight Closed-Loop Glucose Control Under MD-Logic Artificial Pancreas in Patients with Type 1 Diabetes: The DREAM Project

Abstract

Background:

Artificial pancreas systems may offer a potential major impact on the normalization of metabolic control and preventing hypoglycemic events. This study aims to establish near-normal overnight glucose control and reduce the risk of nocturnal hypoglycemia using the MD-Logic Artificial Pancreas (MDLAP), an algorithm that was developed by our research group. This inpatient feasibility study is the first step towards implementing an overnight closed-loop MDLAP system at the patient's home.

Subjects and Methods:

Seven patients with type 1 diabetes (three adolescents and four adults; mean±SD age, 20.6±4.7 years; duration of diabetes, 9.6±2.6 years; body mass index, 24.3±3.9 kg/m²; and glycated hemoglobin, 7.8±0.8%) participated in a total of 14 closed-loop overnight sessions. Each participant underwent two closed-loop inpatient sessions starting at dinner alone and at dinner following exercise. The closed-loop inpatient sessions were compared with data derived from nights spent at home with an open-loop system in a similar scenario to the study protocol.

Results:

The mean percentage of time spent in the near normal glucose range of 63–140 mg/dL was 83±16%, and the median (interquartile range) was 85% (78–92%) for the overnight closed-loop sessions compared with 34±31% and 27% (6–57%) in the homecare open-loop setting, respectively. During the overnight closed-loop sessions at dinner alone 92±9% of the sensor values ranged within target range, compared with 73±19% for the sessions following exercise (P=0.03). No hypoglycemic (<63 mg/dL) events occurred during the closed-loop sessions.

Conclusion:

Closed-loop insulin delivery under MDLAP is a feasible and safe solution to control overnight glycemia.

Introduction

N ighttime is a period of concern for patients of all ages and the families of children and adolescents with type 1 diabetes mellitus (T1DM) due to the risk of hypoglycemia. Hypoglycemic continuous glucose monitoring base events occurred in 8.5% out of 36,467 nights as was shown in a study in 176 adults and children with TIDM.¹ Moreover, most severe hypoglycemic events in children occur during the night and account for 75% of all hypoglycemic seizures.² Failure to arouse and respond to nocturnal hypoglycemia can result in coma and even death. In addition, 6% of deaths in patients with T1DM under the age of 40 years are attributed to “death-in-bed” syndrome.³ This risky condition results in increasing parental fear and anxiety,⁴ and approximately one-third of the parents woke to check their children's blood glucose levels during the night.⁵ Thus nocturnal hypoglycemia has a huge impact on parental health and quality of life. Therefore, parents of children with diabetes may choose hypoglycemia avoidance strategies at the expense of good glycemic control.⁴

Approximately one-third of each day is spent in sleep with no meals or exercise; hence the overnight period is a significant proportion of the time requiring good glycemic control for patients with T1DM. The implementation of insulin pump and continuous glucose sensing for the treatment of diabetes has been demonstrated to improve overall glycemic control but not nighttime control. A study by Deiss et al.⁶ showed that changing from multiple daily injections to pump therapy significantly improved glycemic control, but not the nighttime glucose control. The use of continuous glucose monitors in well-controlled children with diabetes demonstrated a significant improvement in daytime glucose control but, again, not overnight control.⁷ The obvious explanation for the limited improvement overnight recorded in both of these studies is the lack of an appropriate intervention for patients who are asleep, even the ones who are most motivated and well controlled. It has also been shown that the glucose sensors alarming for decreasing blood glucose level are not successful in significantly reducing nocturnal hypoglycemic events as 71% of children with diabetes do not respond to the alarms during the night.⁸

Therefore, a feedback closed-loop system for controlling glucose levels throughout the night has the potential to decrease hypoglycemic events and maintain good glycemic control and thus liberates the patients with diabetes, their families, and caregivers from the anxiety and stress associated with this period of time. Moreover, an assessment of parental attitudes regarding automated glucose control revealed that 90% of them would feel confident in using an overnight closed-loop system.⁹

A major concern in using continuous glucose monitoring (CGM) for closed-loop systems is the sensor accuracy over a wide array of glucose levels, especially the hypoglycemic range. Several studies have demonstrated that CGM was less accurate during the night than during the day, leading to overestimation of the frequency of nocturnal hypoglycemia.^10,11 This might be due to decreased movement or to changes in the subcutaneous circulation or oxygen availability overnight. Furthermore, when glucose is falling, interstitial glucose may fall before plasma glucose and reach nadirs lower than corresponding glucose levels.¹² These accuracy issues may be automatically detected by looking at the sensor raw data. Nevertheless, overestimation of hypoglycemia or earlier detection might be safety advancement to a closed-loop system.

The MD-Logic artificial pancreas (MDLAP), an algorithm that was developed by our research group, applies fuzzy logic theory to imitate lines of reasoning of diabetes caregivers in order to establish good glucose control, as demonstrated in previous studies.¹³ This system provides a personalized treatment approach with real-time learning ability.¹⁴ It uses a combination of control-to-range and control-to-target strategies to automatically regulate individual blood glucose levels, in addition to safety algorithms for prevention of hypoglycemia, prolonged hyperglycemia, etc.¹³

The study was carried out within the framework of the Diabetes Wireless Artificial Pancreas Consortium (DREAM) project that was established in order to carry out studies that will evaluate the efficacy and safety of the MDLAP system in controlling the blood glucose levels of patients with T1DM during the night at home. It is a multicenter, multinational consortium with centers in Slovenia, Germany, and Israel. We conducted a validation study, testing MDLAP overnight at inpatient settings under two main challenges: dinner alone and dinner following exercise.

Research Design and Methods

The current study was designed as a prospective feasibility trial to assess overnight closed-loop glucose control under MDLAP. From October 2010 to February 2011, seven patients with T1DM attending the National Center for Childhood Diabetes at Schneider Children's Medical Center of Israel were recruited. Inclusion criteria were >14 years old, disease duration of at least 1 year, and treatment with an insulin pump for at least 3 months. Patients with a concomitant disease affecting metabolic control or another medical condition that could affect their safety during the trial were excluded, as were patients with a known or suspected allergy to the trial products or who had participated in another study involving drugs that could affect glucose measurements or glucose management.

The study was approved by the institutional ethics committee and carried out in accordance with the Declaration of Helsinki. All subjects/surrogates signed an informed consent form.

Seven patients (three adolescents and four adults; four females and three males), 20.6±4.7 years old, with duration of diabetes mellitus of 9.6±2.6 years, duration on insulin pump of 6.1±2.5 years, body mass index of 24.3±3.9 kg/m², total daily insulin dose of 0.96±0.48 units/kg/day, and glycated hemoglobin (HbA1c) of 7.8±0.8% (61.4±9.6 mmol/mol), participated in 14 closed-loop sessions. The Body Mass Index-SD Score for the three adolescents was 1.31, 0.25, and 0.05.

Each patient attended two automated overnight closed-loop sessions on two separate occasions. One session consisted of the subject's having dinner only (“Visit 1”), and the second session consisted of the subject's having dinner after having taken part in exercise (“Visit 2”). Each session was supervised by a diabetologist. In both visits, patients ate dinner comprising approximately 45–50 g of carbohydrates at around 19:00 p.m. followed by a postmeal fast until the following morning.

Study protocol: procedures

Preparation visit

Prior to each closed-loop session, the patients wore a continuous glucose monitor (Sof-sensor^®; Medtronic, Northridge, CA) for a period of 6 days and used an insulin pump that was compatible with the MDLAP system (MiniMed Paradigm^® Veo™ insulin pump+CGM; Medtronic). They were instructed to use the bolus calculator in their insulin pumps in order to record their carbohydrate consumption during this period and to use only their own glucometer throughout the study (FreeStyle^®; Abbott Diabetes Care, Alameda, CA) (coefficient of variance between 2.2% and 2.4%). A short-acting insulin analog (NovoRapid^®; Novo Nordisk, Bagsvaerd, Denmark) was used. The learning algorithm utilized the data downloaded from the sensors, pumps, and glucometers to create the patients' profile for application in the MDLAP system.¹⁴ All patients were healthy during the 6 days at home.

Visit 1: overnight closed-loop session starting at dinner alone

Patients arrived at the study center at approximately 16:00 h. Intravenous access was obtained, and a venous blood sample was taken to check their blood glucose levels according to the YSI^® 2300 STAT PLUS™ (Yellow Springs Instrument Co. Inc., Yellow Springs, OH), and at the same time the patients used their own glucometers. The patients' blood glucose levels were brought as close as possible within the range of 90–120 mg/dL prior to eating dinner: in order to achieve this, they were given correction boluses or carbohydrates as necessary.

All patients received their dinner around 19:00 h. Patients could choose one of two standard meals containing 45–50 g of carbohydrates (one vegetarian and one with meat). Patients/parents were asked to estimate the carbohydrate content of the meal and to give an insulin bolus 10 min beforehand based on their estimation and the current capillary blood glucose level. From this time onward automated control of all insulin delivery was carried out via the closed-loop MDLAP system. To assure the safety of the patients, blood glucose levels were checked via the patient's home glucometers at the following points: before dinner, 2 h postmeal, prior to bed at around 23:00 h (when they were encouraged to go to sleep), and every 2 h throughout the night until 07:00 h when the patients were awakened and disconnected from the closed-loop system. The CGM system was calibrated before dinner and at bedtime using the glucometer values that were measured at that time.

Visit 2: overnight closed-loop session starting at dinner following exercise

The exercise session took place between 17:00 and 18:00 h and included walking on a treadmill with increasing resistance until a target heart rate of 140 bpm was reached. This was divided into three 10-min segments with 5 min of less intensive activity between them. Prior to exercising, the patients adjusted their insulin treatment and/or consumed carbohydrates as they normally do in preparation for a 40-min exercise session. Blood glucose levels were checked via their glucometers at the start and at the end of exercising, and the sensor levels were monitored throughout. After the exercise was completed, dinner was served at 19:00 h as for the first session followed by the activation of the MDLAP closed-loop system.

Automated MDLAP system

We developed the artificial pancreas link (AP-Link) software that enables our MDLAP system to communicate wirelessly with the Paradigm Veo insulin pump and glucose sensor (Medtronic, Northridge, CA). The AP-Link has a human interface that logs and presents data, incorporates safety rules, facilitates physician overview, and provides alerts to the user in case technical or clinical safety issues arise. The AP-Link software is designed to be simple and easy to operate with a double click start-up procedure. Once the patient's profile has been established and the personal pump identification has been entered, the system is activated and automatically regulates its own safety checks. It operates the closed-loop only when all indicators confirm that it is safe to do so. The MDLAP's algorithm uses fuzzy logic technology, which has been previously described in detail.¹³ The sensor readings are fed automatically into the MDLAP system every 5 min, and the system provides an insulin dose recommendation after each reading, which is then automatically transmitted to the pump.

Data analysis and statistics

The overnight period was defined as the time between bedtime (23:00 h) and wakeup (07:00 h). The primary glycemic control outcome measured was percentage of time glucose levels spent within the target range of 63–140 mg/dL. Secondary outcomes included mean glucose concentration, the percentage of time glucose levels spent below and above the glucose target range, the number of hypoglycemic events (glucose level below 63 mg/dL), glucose levels at bedtime and upon awakening, High and Low Blood Glucose Indexes (variability indexes),^15,16 and insulin infusion rates. The post-dinner results (the period of time between meal [19:00 h] until bedtime [23:00 h]) included peak postprandial glucose level, 2-h postprandial glucose levels, number of hypoglycemic events, and insulin infusion rates.

We compared the glycemic control between the closed-loop session with dinner alone (Visit 1) and the closed-loop session following exercise (Visit 2) to evaluate the effect of physical activity on glycemic control. Glycemic control outcomes were analyzed during the overnight and post-dinner periods using the two-tailed homoscedastic t test.

Because this was a feasibility study we did not include a control group for comparison. Instead, in order to appreciate the results achieved, we conducted a qualitative comparison of the glycemic control during the overnight inpatient closed-loop sessions with CGM data collected from the home setting for the same group of patients. Patients' home data were matched to the study protocol scenario for the following parameters: premeal blood glucose level, dinner eaten around 19:00 h comprising approximately 45–50 g of carbohydrate, followed by a postmeal fast until the following morning. We were able to find 15 suitable nights.

Results

There were two instances of sensor accuracy problems. In one occasion the sensor was not correlated with capillary measurements for a period of 6 h despite routine calibration. Therefore, this session was stopped per protocol. In the other occasions the sensors began to drift upwards and created a discrepancy of more than 80 mg/dL from the glucometer readings. Only after calibration did glucose sensor readings start to drop. As a result, the system overdosed with insulin, and the study physician's supplemented the patients with carbohydrates. These two instances were excluded from the overall data analysis, and will be presented separately (leaving 12 sessions: six sessions of Visit 1 and six sessions of Visit 2).

Overnight MDLAP closed-loop control

Figure 1A and B details a 12-h period from commencement of the closed-loop system at the start of the meal (19:00 h) until the time the patient woke up the following morning (07:00 h) for Visit 1 and Visit 2, respectively.

FIG. 1.

Overview of glycemic control under the MD-Logic Artificial Pancreas system. (A and B) Overview of glycemic control of the pooled six closed-loop sessions for (A) Visit 1 (dinner alone) and (B) Visit 2 (physical activity followed by dinner), respectively, from the initiation of the closed-loop system at the start of a meal at 19:00 h until 07:00 h the following morning. The graph shows the mean of continuous glucose monitoring readings (dark line), and the range of continuous glucose monitoring readings (shaded range). The black circles are the mean capillary glucometer measurements taken at before dinner, 2 h postmeal, bedtime (∼23:00 h), and every 2 h throughout the night until 07:00 h. The vertical line that crosses the black circles represents the range of capillary glucose values. The small black triangles at the bottom of the graph are the time at which each meal was commenced. (C) Overview of glycemic control of the pooled 12 closed-loop sessions overnight from bedtime 23:00 h until 07:00 h the following morning. (Top) The mean (dark line) and the range (shaded range) of continuous glucose monitoring readings. (Bottom) Average insulin profile (dark line) and the range of insulin delivery (dashed lines) as well as the range of preprogrammed insulin basal rate according to the patients' pump settings at home (shaded range). (D) Overview of glycemic control of the pooled 15 home environment equivalent overnight from bedtime at 23:00 h until 07:00 h the following morning. The graph shows the mean (dark line) and the range (shaded range) of continuous glucose monitoring readings.

Closed-loop control starting at dinner alone versus dinner following exercise

Comparison of closed-loop control with (n=6) and without (n=6) exercise is presented in Table 1.

Table 1.

Summary of Glycemic Control Parameters for Visit 1 (Meal Only) Versus Visit 2 (Meal Post-Exercise) Under MD-Logic Artificial Pancreas Control

	Mean±SD
	Meal only (n=6)	Meal post-exercise (n=6)	p^a
Post-dinner (19:00–23:00 h)
Glucose at 19:00 (mg/dL)	113±24	151±48	NS
Glucose 2 h post-dinner (mg/dL)	138±40	183±16	0.03
Premeal bolus (by the patient) (units)	4.8±2.6	4.1±1.7	NS
Additional correction bolus (by MDLAP) (units)	1±0.7	2.8±0.9	0.003
Overnight (23:00–07:00 h)
Average glucose (mg/dL)	114±11	130±8	0.02
Time<63 mg/dL (%)	0	0	—
Time within 63–140 mg/dL (%)	92±9	73±16	0.03
Number of events<63 mg/dL)	0	0	—
Glucose at 23:00 h (mg/dL)	120±10	170±28	0.002
Glucose at 07:00 h (mg/dL)	107±12	121±6	0.02
HBGI	0.3±0.3	1.5±0.8	0.006
LBGI	0.3±0.5	0.08±0.13	NS
Mean insulin infusion (units/h)	1.02±0.26	1.1±0.26	NS

By two-tailed homoscedastic t test.

HBGI, High Blood Glucose Index^19,20; LBGI, Low Blood Glucose Index^19,20; MDLAP, MD-Logic Artificial Pancreas; NS, not significant.

The mean glucose level at dinnertime and the amount of premeal bolus had no statistically significant differences in the two visits. However, the 2-h postprandial glucose level was significantly higher on Visit 2 compared with Visit 1 (P=0.03), causing the MDLAP system to deliver a significant additional amount of correction boluses on Visit 2 compared with Visit 1 (P=0.003). Overall visits, the peak postprandial glucose level ranged between 91 and 208 mg/dL, and the average 2-h postprandial glucose level was 161±37 mg/dL. The meal insulin for Visit 1 and Visit 2 was on average 4.8±2.6 and 4.1±1.8 units, respectively, with no statistical significance between them.

During the overnight period, the average percentage of time that glucose levels spent within target range was significantly higher on the meal-only scenario compared with the meal post-exercise scenario (92% vs. 73%, respectively; P=0.03). Mean blood glucose level was 114±11 mg/dL versus 130±8 mg/dL (P=0.02), respectively. No time was spent below 63 mg/dL in any instance. In addition, the mean insulin infusion rate was similar in the two visits.

Closed-loop versus home equivalent open-loop control

The sensor glucose profiles for the overnight period (23:00–07:00 h) during closed-loop control versus home open-loop control are presented in Table 2 and in Figure 1C and D. The mean±SD percentage of time glucose levels spent in the target range (63–140 mg/dL) was 83±16%, and the median (interquartile range) was 85% (78–92%) for the closed-loop control versus 34±31% and 27% (6–57%) for the homecare setting, respectively. There were no hypoglycemic events and therefore no time spent below 63 mg/dL in the closed-loop control, as opposed to the homecare patients who had a mean of 0.3±0.5 events and had 7±17.5% of the nighttime spent in the hypoglycemic range.

Table 2.

Variables of Glycemic Control for the Overnight Period (23:00–07:00 h) for the MD-Logic Artificial Pancreas and for Home Care with Sensor-Augmented Pump Therapy

	Mean±SD
	MDLAP (n=12) CGM	Capillary ^a	Homecare with SAPT (n = 15)
Time<63 mg/dL (%)	0	—	7±17.5
Time within (%)
63–140 mg/dL	83±16	—	34±31
63–80 mg/dL	0.66±1.5	—	4.5±9.8
Number of events<63 mg/dL	0	0	0.3±0.5
Average glucose (mg/dL)	122±16	129±12	160±57
Glucose (mg/dL) at
23:00 h	135±28	147±35	178±59
07:00 h	114±11	109±12	134±59
HBGI	1±0.9	—	8±9
LBGI	0.2±0.4	—	2±5
Mean insulin infusion (U/h)	1.1±0.25	—	1.2±0.5

Capillary measurement using a glucometer.

CGM, continuous glucose monitoring; HBGI, High Blood Glucose Index^19,20; LBGI, Low Blood Glucose Index^19,20; MDLAP, MD-Logic Artificial Pancreas; SAPT, sensor-augmented pump therapy.

Excluded datasets

Figure 2 shows the two datasets that were excluded from analysis. We had one hypoglycemia event that was related to insulin overdose of the system due to sensor inaccurate readings. Nadir capillary glucose value was 53 mg/dL, whereas CGM nadir reading was 79 mg/dL. The patient was treated with 30 g of fast-acting carbohydrate, and the closed-loop trial was continued. Mean glucose levels according to CGM and capillary values were 123±32 mg/dL and 95±38 mg/dL, respectively.

FIG. 2.

Excluded dataset results. (A) Dataset in which the trial was stopped. (B) Dataset with the hypoglycemic event due to sensor accuracy problem. (Top) Continuous glucose monitoirng readings (black line) and capillary measurements (black circles). (Bottom) Insulin basal rate delivery (horizontal black line) and insulin bolus (vertical black lines with black circle at the top). In (B) the white circle with black edge is the calibration point where there was a 80 mg/dL discrepancy between the capillary measurement and the continuous glucose monitoring reading.

Conclusions

The results of this current study demonstrate the ability of the MDLAP system to control overnight glucose levels within the near-normal range with no nocturnal hypoglycemia in young patients with T1DM. Our closed-loop system also enhanced the patients' management of blood glucose levels around dinnertime, resulting in improved blood glucose levels at bedtime with fewer overnight variables.

The present study was designed to serve as a feasibility study and therefore did not include a control group. However, we compared nights at home that had similar conditions (i.e., same meal size at a similar time of day and followed by a postprandial fast until the following morning) with closed-loop nights in a clinical research center. The overnight glucose levels at patients' homes remained within the target range for only 34% of the time and within the hypoglycemic range up to 7% of the time. The overnight results at patients' homes are similar to previous inpatient studies we had undertaken with open-loop control using similar conditions (i.e., the overnight blood glucose levels were within target range for 40% of the time). Therefore, although comparing home open-loop nights with hospital closed-loop nights is not ideal, it is reasonably representative. During the night period (23:00–07:00 h), the extent of variability of glucose levels was far less in the closed-loop nights compared with the homecare open-loop nights, possibly as a result of the better pre-bedtime glycemic levels that had been achieved during the evening in the period immediately after dinner. Despite the adjustments made to insulin delivery during the closed-loop session, the actual amount of insulin units was similar in both the closed-loop and homecare nights.

In order to achieve safe overnight control, a system should be able to cope with situations that may have prolonged influence on overnight blood glucose levels, especially those likely to cause hypoglycemia. Therefore, we included in our study overnight sessions following pre-dinner exercise. The patients who exercised prior to their meal had a higher—although not significant—premeal glucose level compared with patients who did not exercise. The 2-h postprandial value was significantly higher in the dinner post-exercise scenario (P=0.03). This finding was somewhat surprising because we expected lower glucose levels in patients who had exercised. This could partly be attributed to the patients' behavior surrounding the exercise. We instructed the patients to follow their usual routine with regards to planned exercise. Some reduced their basal rate during the exercise, and some ate carbohydrates before the exercise. None suspended his or her pump. Patients apparently overestimate the influence of exercise on glycemic control, and this may be the reason for higher post-exercise blood glucose levels. Individual patient's responses to exercise may also have contributed to this blood glucose rise. Whatever the reason could be, the time spent within target range was significantly greater (P=0.03) in the group that did not exercise before dinner, contrary to what we had expected. Indeed, Hovorka et al.¹⁷ showed that closed-loop control provided the greatest amount of time within target range after early evening exercise. Their study included nine patients who undertook early evening exercise in a similar fashion to our protocol. In their study, the preparation for exercise included a snack of 45 g of carbohydrates with a matched insulin bolus, whereas our patients consumed 45 g of carbohydrates with no corresponding pre-exercise bolus. It is of importance to note that even though our patients' pre-exercise actions resulted in higher blood glucose levels before bedtime, the closed-loop control still succeeded in bringing their overnight blood glucose levels within range 73% of the time.

Overnight closed-loop control has been evaluated in several studies,^17

–23 all of which showed increased time spent within-target range, with fewer hypoglycemic events and of shorter duration. It is difficult to compare the results obtained from our overnight MDLAP with similar closed-loop studies because of numerous differences in methodology (e.g., duration of the overnight period, time of commencement of closed-loop control, carbohydrate content of meals, meal/exercise timings, postprandial fasting, age group, and number of participants within each study). Furthermore, the data were analyzed and presented in different ways: for instance, the time within a range was displayed as a median in some studies and as an average in others; some calculated data from sensor glucose levels and some from reference blood glucose results. We believe that these differences highlight the necessity to establish a uniform way to conduct studies on closed-loop systems and determine the appropriate way to evaluate the data for clinical application.

There were two sensor drift events in two sessions, and both involved a discrepancy of approximately 75 mg/dL between the sensor reading and the glucometer. The episodes occurred during a descent of blood glucose following an insulin bolus. The sensor glucose reading lagged behind the actual glucose level, thus preventing the artificial pancreas system from responding adequately. This demonstrates one of the safety risks faced by the closed-loop system, the reliance on sensors, which needs to be addressed.

The present study shows that the MDLAP system has important built-in layers of restraints and safety mechanisms, especially for hypoglycemia, as well as a fully integrated automatic ability to communicate wirelessly between the components of the closed-loop system. We encountered no major problems with technical communication throughout the study. Our experience in the current study shows that this mode of operation is feasible and easy to use.

In conclusion, closed-loop insulin delivery by the MDLAP system was found to be feasible and safe. Moreover, the use of the MDLAP can improve the control of glucose levels during the night and reduce the risk of nocturnal hypoglycemia in patients with T1DM.

Footnotes

Author Disclosure Statement

R.N., E.A., M.A., S.M., and T.O. declare that no competing financial interests exist. M.P. has served on Advisory Panels for AstraZeneca LP and Sanofi, as a Board Member for C.G.M.3 Ltd., as a consultant for Bristol-Myers Squibb Company, D-medical, AstraZeneca LP, Sanofi, Ferring, and Andromeda, and on Speaker's Bureaus for Animas Corporation, Sanofi, Medtronic, Bayer Health Care, has received research support from Medtronic, Novo Nordisk, Inc., Abbott Diabetes Care, Eli Lilly and Company, Roche, Dexcom, Inc., Sanofi, Insulet Corporation, Animas Corporation, Andromeda, and Macrogenetics, is a stock/shareholder in C.G.M.3 Ltd. and De-medical, and has other relationships with Bayer Health Care, Medtronic, Animas Corporation, Novo Nordisk, Inc., Pfizer, Inc., and Sanofi.

References

Juvenile Diabetes Research Foundation Continuous Glucose Monitoring Study Group. Prolonged nocturnal hypoglycemia is common during 12 months of continuous glucose monitoring in children and adults with type 1 diabetes. Diabetes Care, 2010; 33:1004–1008.

Davis

, Keating

, Byrne

, Russell

, Jones

. Hypoglycemia: incidence and clinical predictors in a large population-based sample of children and adolescents with IDDM. Diabetes Care, 1997; 20:22–25.

Sovik

, Thordarson

. Dead-in-bed syndrome in young diabetic patients. Diabetes Care, 1999; 22,Suppl 2:B40–B42.

Barnard

, Thomas

, Royle

, Noyes

, Waugh

. Fear of hypoglycaemia in parents of young children with type 1 diabetes: a systematic review. BMC Pediatr, 2010; 10:50.

Monaghan

, Hilliard

, Cogen

, Streisand

. Nighttime caregiving behaviors among parents of young children with type 1 diabetes: associations with illness characteristics and parent functioning. Fam Syst Health, 2009; 27:28–38.

Deiss

, Hartmann

, Hoeffe

, Kordonouri

. Assessment of glycemic control by continuous glucose monitoring system in 50 children with type 1 diabetes starting on insulin pump therapy. Pediatr Diabetes, 2004; 5:117–121.

Battelino

, Phillip

, Bratina

, Nimri

, Oskarsson

, Bolinder

. Effect of continuous glucose monitoring on hypoglycemia in type 1 diabetes. Diabetes Care, 2011; 34:795–800.

Buckingham

, Block

, Burdick

, Kalajian

, Kollman

, Choy

, Wilson

, Chase

. Diabetes Research in Children Network: Response to nocturnal alarms using a real-time glucose. sensor Diabetes Technol Ther, 2005; 7:440–447.

Elleri

, Acerini

, Allen

, Hayes

, Pesterfield

, Wilinska

, Dunger

, Hovorka

. Parental attitudes towards overnight closed-loop glucose control in children with type 1 diabetes. Diabetes Technol Ther, 2010; 12:35–39.

10.

McGowan

, Thomas

, Moran

. Spurious reporting of nocturnal hypoglycemia by CGMS in patients with tightly controlled type 1 diabetes. Diabetes Care, 2002; 25:1499–1503.

11.

Diabetes Research in Children Network (DirecNet) Study Group. The accuracy of the CGMS in children with type 1 diabetes: results of the Diabetes Research in Children Network (DirecNet) accuracy study. Diabetes Technol Ther, 2003; 5:781–789.

12.

Cengiz

, Tamborlane

. A tale of two compartments: interstitial versus blood glucose monitoring. Diabetes Technol Ther, 2009; 11,Suppl 1:S11–S16.

13.

Atlas

, Nimri

, Miller

, Grunberg

, Phillip

. MD-logic artificial pancreas system: a pilot study in adults with type 1 diabetes. Diabetes Care, 2010; 33:1072–1076.

14.

Miller

, Nimri

, Atlas

, Grunberg

, Phillip

. Automatic learning algorithm for the MD-logic artificial pancreas system. Diabetes Technol Ther, 2011; 13:983–990.

15.

Kovatchev

, Otto

, Cox

, Gonder-Frederick

, Clarke

. Evaluation of a new measure of blood glucose variability in diabetes. Diabetes Care, 2006; 29:2433–2438.

16.

McCall

, Cox

, Crean

, Gloster

, Kovatchev

. A novel analytical method for assessing glucose variability: using CGMS in type 1 diabetes mellitus. Diabetes Technol Ther, 2006; 8:644–653.

17.

Hovorka

, Allen

, Elleri

, Chassin

, Harris

, Xing

, Kollman

, Hovorka

, Larsen

, Nodale

, De Palma

, Wilinska

, Acerini

, Dunger

. Manual closed-loop insulin delivery in children and adolescents with type 1 diabetes: a phase 2 randomised crossover trial. Lancet, 2010; 375:743–751.

18.

Elleri

, Allen

, Nodale

, Wilinska

, Mangat

, Larsen

, Acerini

, Dunger

, Hovorka

. Automated overnight closed-loop glucose control in young children with type 1 diabetes. Diabetes Technol Ther, 2011; 13:419–424.

19.

Hovorka

, Kumareswaran

, Harris

, Allen

, Elleri

, Xing

, Kollman

, Nodale

, Murphy

, Dunger

, Amiel

, Heller

, Wilinska

, Evans

. Overnight closed-loop insulin delivery (artificial pancreas) in adults with type 1 diabetes: crossover randomised controlled studies. BMJ, 2011; 13,342:d1855.

20.

Clarke

, Anderson

, Breton

, Patek

, Kashmer

, Kovatchev

. Closed-loop artificial pancreas using subcutaneous glucose sensing and insulin delivery and a model predictive control algorithm: the Virginia experience. J Diabetes Sci Technol, 2009; 3:1031–1038.

21.

Kovatchev

, Cobelli

, Renard

, Anderson

, Breton

, Patek

, Clarke

, Bruttomesso

, Maran

, Costa

, Avogaro

, Dalla Man

, Facchinetti

, Magni

, De Nicolao

, Place

, Farret

. Multinational study of subcutaneous model-predictive closed-loop control in type 1 diabetes mellitus: summary of the results. J Diabetes Sci Technol, 2010; 4:1374–1381.

22.

Bruttomesso

, Farret

, Costa

, Marescotti

, Vettore

, Avogaro

, Tiengo

, Dalla Man

, Place

, Facchinetti

, Guerra

, Magni

, De Nicolao

, Cobelli

, Renard

, Maran

. Closed-loop artificial pancreas using subcutaneous glucose sensing and insulin delivery and a model predictive control algorithm: preliminary studies in Padova and Montpellier. J Diabetes Sci Technol, 2009; 3:1014–1021.

23.

Weinzimer

, Steil

, Swan

, Dziura

, Kurtz

, Tamborlane

. Fully automated closed-loop insulin delivery versus semiautomated hybrid control in pediatric patients with type 1 diabetes using an artificial pancreas. Diabetes Care, 2008; 31:934–939.