Electrocardiography and heart rate variability in Göttingen Minipigs: Impact of diurnal variation, lead placement, repeatability and streptozotocin-induced diabetes

BACKGROUND

The Göttingen Minipig is widely used in preclinical research and safety pharmacology, but standardisation of porcine electrocardiography (ECG) is lacking. The aim of this study was to investigate diurnal effects, change over time and choice of lead on ECG morphology and heart rate variability (HRV) in healthy and streptozotocin (STZ) induced diabetic Göttingen Minipigs.

METHODS

Diabetes was experimentally induced using STZ in 11 Göttingen Minipigs (DIA). Seven controls (CON) were included. 24-h ECG was recorded at baseline and four months. Morphological parameters (QRS and T wave duration, P- and T-wave amplitude, PR and QT (Bazett's (QTcb) or Fridericia (QTcf) correction) intervals and ST segment), presence of cardiac arrhythmias, heart rate (HR) and HRV (time and frequency domain) were analysed.

RESULTS

Four months after induction, DIA had decreased P-wave amplitude (P < 0.0001) and T-wave duration (P = 0.017), compared to CON. QTcb was lower in DIA, but not in CON. Both groups had decreased HR (P < 0.0001) and QRS duration (lead II, P = 0.04) and length of PR-segment increased (lead I and II, P < 0.01) while selected HRV parameters also increased (all P < 0.01). Time of day influenced HR, QRS duration, PR segment, ST segment, T- and P-wave amplitude and some parameters of HRV. Inter- and intra-observer variability of morphological measurements was low (<6%).

CONCLUSION

ECG parameters were influenced by time setting, diurnal variation and lead. Some ECG and HRV changes were found in diabetic minipigs four months after STZ induction. The findings underline the need for standardisation of ECG and HRV in Göttingen Minipigs.

Hyperinsulinaemic hypoglycaemia in non-anaesthetized Göttingen minipigs induces a counter-regulatory endocrine response and electrocardiographic changes

The potentially fatal cardiovascular effects of hypoglycaemia are not well understood and large animal models of the counter-regulatory responses and cardiovascular consequences of insulin-induced hypoglycaemia are needed to understand the mechanisms in humans. The aim of this study was to develop a human-like minipig model of hypoglycaemia including healthy and diabetic pigs to investigate endocrine, electrocardiographic and platelet effects. Hypoglycaemia was induced using a hyperinsulinaemic, hypoglycaemic clamp and an insulin bolus protocol. Plasma glucose, glucagon, C-peptide, insulin, epinephrine and platelet aggregation responses were measured before, during and after hypoglycaemia. Continuous electrocardiographic recordings were obtained. Hypoglycaemia at a plasma glucose concentration of 0.8–1.0 mM in the clamp induced 25-fold increase in epinephrine and sixfold and threefold increase in glucagon for healthy and diabetic pigs, respectively. The hypoglycaemic clamp induced QTc-interval prolongation and increase in cardiac arrhythmias. In the bolus approach, the non-diabetic group reached plasma glucose target of 1.5 mM and QTc-interval was prolonged after insulin injection, but before glucose nadir. The diabetic group did not reach hypoglycaemic target, but still demonstrated QTc-interval prolongation. These results demonstrate effects of hyperinsulinaemic hypoglycaemia closely resembling human physiology, indicating the minipig as a translational animal model of counter-regulatory endocrine and myocardial effects of hypoglycaemia.

Food intake rather than blood glucose levels affects the pharmacokinetic profile of insulin aspart in pigs

In humans, food intake and glucose infusion have been reported to increase subcutaneous blood flow. Since local blood flow influences the rate of insulin absorption from the subcutaneous tissue, we hypothesised that an increase in blood glucose levels-occurring as the result of glucose infusion or food intake-could modulate the pharmacokinetic properties of subcutaneously administered insulin. The pharmacokinetic profile of insulin aspart was assessed in 29 domestic pigs that were examined in a fed and fasted state or included in hyperinsulinaemic clamp studies of 4 vs. 10 mmol/L glucose prior to subcutaneous (30 nmol) or intravenous (0.1 nmol/kg) insulin administration. Results showed that food intake compared to fasting accelerated absorption and decreased clearance of insulin aspart (P < 0.05). Furthermore, higher c-peptide but also glucagon levels were observed in fed compared to fasted pigs (P < 0.05). The pharmacokinetic profile of insulin aspart did not differ between pigs clamped at 4 vs. 10 mmol/L glucose. Hence, food intake rather than blood glucose levels within normal range modulates the pharmacokinetic properties of insulin aspart upon subcutaneous and intravenous administration in pigs.

Molecular engineering of safe and efficacious oral basal insulin

Recently, the clinical proof of concept for the first ultra-long oral insulin was reported, showing efficacy and safety similar to subcutaneously administered insulin glargine. Here, we report the molecular engineering as well as biological and pharmacological properties of these insulin analogues. Molecules were designed to have ultra-long pharmacokinetic profile to minimize variability in plasma exposure. Elimination plasma half-life of ~20 h in dogs and ~70 h in man is achieved by a strong albumin binding, and by lowering the insulin receptor affinity 500-fold to slow down receptor mediated clearance. These insulin analogues still stimulate efficient glucose disposal in rats, pigs and dogs during constant intravenous infusion and euglycemic clamp conditions. The albumin binding facilitates initial high plasma exposure with a concomitant delay in distribution to peripheral tissues. This slow appearance in the periphery mediates an early transient hepato-centric insulin action and blunts hypoglycaemia in dogs in response to overdosing.

The counterregulatory response to hypoglycaemia in the pig

The domestic pig is commonly used as animal model in the pharmaceutical development of new therapeutics for treatment of diabetes. Since a formal definition of hypoglycaemia only exists in humans, the purpose of this study was to assess the counterregulatory response in the domestic pig at glucose levels known to induce symptoms of hypoglycaemia in humans. Six pigs were included in hyperinsulinaemic glucose clamps with plasma glucose targets of 2, 3 and 5 mmol/L in a cross-over design, and the associated glucose counterregulatory response was assessed by measuring glucose kinetics and levels of glucagon, c-peptide, catecholamines, cortisol and growth hormone. Results showed that the 2 and 3 vs 5 mmol/L clamps significantly decreased and increased the secretion of c-peptide and glucagon, respectively (P < .05). This finding was associated with increased rate of glucose appearance (R_a ) and decreased rate of glucose disappearance (R_d ) (P < .001). No marked differences in the catecholamine, growth hormone or cortisol response were observed. Consequently, like humans, pigs respond to hypoglycaemia by decreasing the pancreatic output of insulin while increasing that of glucagon, with increased glucose mobilization and decreased glucose disposal as a result. The hypoglycaemic clamps did not result in a marked secretion of the other counterregulatory hormones.

Intraintestinal and Parenteral Administration of an Insulin Analogue Leads to Comparable Activation of Signaling Downstream of the Insulin Receptor in the Small Intestine

BACKGROUND

Oral delivery of insulin was recently demonstrated to have therapeutic relevance in patients with diabetes. Insulin receptors are expressed in the gastrointestinal tract and can be activated by insulin in the bloodstream, but it is not known if the large amount of insulin in the intestinal lumen required for sufficient oral delivery will induce a different effect. The aim of this study was to compare the acute effect in the intestine of insulin administered in the intestinal lumen with that of insulin administered by a parenteral route.

METHOD

Intraintestinal (ii) injection in the mid-jejunum of anaesthetized rats with insulin analogue 106 (I106), formulated with the absorption-enhancer sodium caprate, was used as an animal model of oral insulin administration. As control treatment, rats were treated with I106 by iv infusion according to algorithms which precisely mimicked the pharmacokinetic and pharmacodynamic properties of ii administered I106. Several fold more I106 was administered by ii injection than by iv infusion. Phosphorylated Akt (Ser473) was used as indicator of insulin-stimulated acute effects in the intestine.

RESULTS

Treatment with I106 resulted in activation of Akt in the intestine, with no significant difference between the effects of ii or iv administration.

CONCLUSION

The results from this rat model of orally administered insulin indicate that the unabsorbed insulin in the intestinal lumen after oral administration will not result in an enhanced acute effect in the intestine.

Injection Technique and Pen Needle Design Affect Leakage From Skin After Subcutaneous Injections

BACKGROUND

After a subcutaneous injection fluid might leak out of the skin, commonly referred to as leakage or backflow. The objective was to examine the influence of needle design and injection technique on leakage after injections in the subcutaneous tissue of humans and pigs.

METHOD

Leakage data were obtained from a post hoc analysis of clinical trial data and from a pig study. Data from the clinical study were used to determine leakage as a function of injection volume, speed and region. Data from the pig study were used to determine leakage as a function of needle wall thickness, needle taper, injection angle, and wait time from end of injection to withdrawal of needle from skin.

RESULTS

Leakage volume was positively related to injection volume. Injections in the abdomen caused less leakage than thigh injections. A 32G needle caused less leakage than a 31G and a 32G tip (tapered) needle, and a "straight in" 90° needle insertion angle caused less leakage than an angled (~45°) insertion. Wait times of minimum 3 seconds caused less leakage than immediate withdrawal of the needle after injection. Needle wall thickness and injection speed did not influence leakage.

CONCLUSIONS

Leakage will be minimized using a thin needle, using 90° needle insertion in the abdomen, injecting maximum 800 µL at a time, and waiting at least 3 seconds after the injection until the needle is withdrawn from the skin.

Pen needle design influences ease of insertion, pain, and skin trauma in subjects with type 2 diabetes

OBJECTIVE

Pen needles used for subcutaneous injections have gradually become shorter, thinner and more thin walled, and thereby less robust to patient reuse. Thus, different needle sizes, alternative tip designs and needles resembling reuse were tested to explore how needle design influences ease of insertion, pain and skin trauma.

RESEARCH DESIGN AND METHODS

30 subjects with injection-treated type 2 diabetes and body mass index 25-35 kg/m² were included in the single-blinded study. Each subject received abdominal insertions with 18 different types of needles. All needles were tested twice per subject and in random order. Penetration force (PF) through the skin, pain perception on 100 mm visual analog scale, and change in skin blood perfusion (SBP) were quantified after the insertions.

RESULTS

Needle diameter was positively related to PF and SBP (p<0.05) and with a positive pain trend relation. Lack of needle lubrication and small 'needle hooks' increased PF and SBP (p<0.05) but did not affect pain. Short-tip, obtuse needle grinds affected PF and SBP, but pain was only significantly affected in extreme cases. PF in skin and in polyurethane rubber were linearly related, and pain outcome was dependent of SBP increase.

CONCLUSIONS

The shape and design of a needle and the needle tip affect ease of insertion, pain and skin trauma. Relations are seen across different data acquisition methods and across species, enabling needle performance testing outside of clinical trials.

TRIAL REGISTRATION NUMBER

NCT02531776; results.

Göttingen minipig model of diet-induced atherosclerosis: influence of mild streptozotocin-induced diabetes on lesion severity and markers of inflammation evaluated in obese, obese and diabetic, and lean control animals

Background

From a pharmacological perspective, readily-available, well-characterized animal models of cardiovascular disease, including relevant in vivo markers of atherosclerosis are important for evaluation of novel drug candidates. Furthermore, considering the impact of diabetes mellitus on atherosclerosis in human patients, inclusion of this disease aspect in the characterization of a such model, is highly relevant. The objective of this study was to evaluate the effect of mild streptozotocin-induced diabetes on ex- and in vivo end-points in a diet-induced atherosclerotic minipig model.

Methods

Castrated male Göttingen minipigs were fed standard chow (CD), atherogenic diet alone (HFD) or with superimposed mild streptozotocin-induced diabetes (HFD-D). Circulating markers of inflammation (C-reactive protein (CRP), oxidized low-density lipoprotein (oxLDL), plasminogen activator inhibitor-1, lipid and glucose metabolism were evaluated together with coronary and aortic atherosclerosis after 22 or 43 diet-weeks. Group differences were evaluated by analysis of variance for parametric data and Kruskal–Wallis test for non-parametric data. For qualitative assessments, Fisher’s exact test was applied. For all analyses, p < 0.05 was considered statistically significant.

Results

Overall, HFD and HFD-D displayed increased CRP, oxLDL and lipid parameters compared to CD at both time points. HFD-D displayed impaired glucose metabolism as compared to HFD and CD. Advanced atherosclerotic lesions were observed in both coronary arteries and aorta of HFD and HFD-D, with more advanced plaque findings in the aorta but without differences in lesion severity or distribution between HFD and HFD-D. Statistically, triglyceride was positively (p = 0.0039), and high-density lipoprotein negatively (p = 0.0461) associated with aortic plaque area.

Conclusions

In this model, advanced coronary and aortic atherosclerosis was observed, with increased levels of inflammatory markers, clinically relevant to atherosclerosis. No effect of mild streptozotocin-induced diabetes was observed on plaque area, lesion severity or inflammatory markers.

Electronic supplementary material

The online version of this article (doi:10.1186/s12967-015-0670-2) contains supplementary material, which is available to authorized users.

Impact of injection speed and volume on perceived pain during subcutaneous injections into the abdomen and thigh: a single-centre, randomized controlled trial

AIM

The aim of this study was to assess pain associated with subcutaneous injection into the abdomen and thigh of different combinations of injection speeds and volumes.

METHODS

The study was a single-centre, one-visit, double-blinded, randomized controlled trial in 82 adults with type 1 or type 2 diabetes receiving daily injections of insulin or glucagon-like peptide-1 (GLP-1) agonists. Participants received 17 subcutaneous injections (12 in abdomen, 5 in thigh) of saline at different injection speeds (150, 300 and 450 µl/s), with different volumes (400, 800, 1200 and 1600 µl), and two needle insertions without any injection. Pain was evaluated on a 100-mm visual analogue scale (VAS) (0 mm no pain, 100 mm worst pain) and on a yes/no scale for pain acceptability.

RESULTS

Injection speed had no impact on injection pain (p = 0.833). Injection of larger volumes caused significantly more pain [VAS least square mean differences 1600 vs. 400 µl, 7 · 2 mm (95% confidence interval - CI; 4.6-9.7; p < 0.0001); 1600 vs. 800 µl, 7.2 mm (4.4-10.0; p < 0.0001); 1200 vs. 400 µl, 3.5 mm (0.4-6.6; p = 0.025) and 1200 vs. 800 µl, 3.6 mm (0.4-6.7; p = 0.027)]. Significantly more pain occurred in the thigh versus the abdomen [9.0 mm (6.7-11.3; p < 0.0001)].

CONCLUSIONS

Injection speed had no effect on injection pain, whereas higher injection volumes caused more pain. The results of this study may be of value for guiding patients to use the appropriate injection site and technique to reduce their injection pain. Furthermore, these findings may have important implications for the development of new injection devices and drug formulations for clinical practice.

Skin blood perfusion and cellular response to insertion of insulin pen needles with different diameters

Today most research on pen needle design revolves around pain perception statements through clinical trials, but these are both costly, timely, and require high sample sizes. The purpose of this study was to test if tissue damage, caused by different types of needles, can be assessed by evaluating skin blood perfusion response around needle insertion sites. Three common sized pen needles of 28G, 30G, and 32G as well as hooked 32G needles, were inserted into the neck skin of pigs and then removed. Laser Speckle Contrast Analysis was used to measure skin blood perfusion for 20 minutes after the insertions. Seven pigs were included in the study and a total of 118 randomized needle insertions were conducted. Histology was made of tissue samples inserted with 18G, 28G, and 32G needles, and stained to quantify red and white blood cell response. Based on area under curve, calculated for each individual blood perfusion recording and grouped according to needle type, skin blood perfusion response relates to needle diameter. The response was significantly higher after insertions with 28G and hooked 32G needles than with 30G (P < .05) and 32G (P < .01) needles. Histology results were not significant, but there was a trend of an increased response with increasing needle diameter. Skin blood perfusion response to pen needle insertions rank according to needle diameter, and the tissue response caused by hooked 32G needles corresponds to that of 28G needles. The relation between needle diameter and trauma when analyzing histology was also suggested.

Pharmacokinetics following continuous subcutaneous insulin infusion of insulin aspart with or without initial subcutaneous bolus

AIM

To evaluate time to steady state insulin concentration (C(ss)) following continuous subcutaneous insulin infusion (CSII) of insulin aspart (IAsp) with or without an initial s.c. bolus.

METHODS

In random order 10 healthy volunteers were given a basal insulin infusion rate (0.5 U/h) for 8 h with or without an initial s.c. bolus (1.4 U). Serum IAsp was measured until 3 h after infusion was stopped.

RESULTS

An overshoot of IAsp was seen before C(ss) was achieved following an initial bolus of insulin as compared to no bolus. The apparent half-life (t((1/2))) with or without bolus did not differ (p = 0.15). Time to steady state (T(ss)) was evaluated in two ways: (1) T(ss) defined as the first point within an interval of C(ss)+/- 2 x CV was 233 vs. 166 min with and without a bolus respectively (p = 0.068). (2) A t-test was performed for each concentration-time point vs. mean C(ss), and the first point with no significance was defined, T(ss). This gave 208 (p = 0.09) and 178 min (p = 0.24) with and without bolus respectively. Mathematical modelling suggests that an ideal mean bolus should be 0.89 U, and that this bolus dose may result in a shorter T(ss).

CONCLUSION

A bolus of 1.4 U resulted in an overshoot of serum IAsp before C(ss) and a longer period before C(ss) is achieved. Mathematical modelling suggests that a mean bolus of 0.89 U would result in a faster achievement of C(ss) compared to no bolus.

A physiological model of the effect of hypoglycemia on plasma potassium

BACKGROUND

Adrenaline release and excess insulin during hypoglycemia stimulate the uptake of potassium from the bloodstream, causing low plasma potassium (hypokalemia). Hypokalemia has a profound effect on the heart and is associated with an increased risk of malignant cardiac arrhythmias. It is the aim of this study to develop a physiological model of potassium changes during hypoglycemia to better understand the effect of hypoglycemia on plasma potassium.

METHOD

Potassium counterregulation to hypokalemia was modeled as a linear function dependent on the absolute potassium level. An insulin-induced uptake of potassium was modeled using a negative exponential function, and an adrenaline-induced uptake of potassium was modeled as a linear function. Functional expressions for the three components were found using published data.

RESULTS

The performance of the model was evaluated by simulating plasma potassium from three published studies. Simulations were done using measured levels of adrenaline and insulin. The mean root mean squared error (RMSE) of simulating plasma potassium from the three studies was 0.09 mmol/liter, and the mean normalized RMSE was 14%. The mean difference between nadirs in simulated and measured potassium was 0.12 mmol/liter.

CONCLUSIONS

The presented model simulated plasma potassium with good accuracy in a wide range of clinical settings. The limited number of hypoglycemic episodes in the test set necessitates further tests to substantiate the ability of the model to simulate potassium during hypoglycemia. In conclusion, the model is a good first step toward better understanding of changes in plasma potassium during hypoglycemia.

Sources of glycemic variability--what type of technology is needed?

People on insulin therapy are challenged with evaluation of numerous factors affecting the blood glucose in order to select the optimal dose for reaching their glucose target. Following medical recommendations precisely still results in considerable blood glucose unpredictability, often resulting in frustration in the short term due to hypoglycemia and hyperglycemia, and, in the long term, will likely result in complications. The kinetics of insulin do indeed vary significantly and have become an important focus when developing new insulin analogues and delivery systems; however, numerous of other factors impact glycemic variability. These have different dependences and interactions and are therefore difficult to characterize. Some of the factors are highly dependent and influenced by the type of insulin and devices used in therapy. Development of future therapy products is therefore highly focused on how to minimize glycemic variability.

Modeling the effect of blood glucose and physical exercise on plasma adrenaline in people with type 1 diabetes

BACKGROUND

Adrenaline is often studied in people with type 1 diabetes during hypoglycemic episodes. Adrenaline is difficult and costly to measure, and therefore a pharmacokinetic model of adrenaline can be a supportive tool that adds information and saves measurements resources.

METHODS

We have developed a compartment model of adrenaline secretion and elimination. It is based on input on physical exercise, blood glucose level, and optional infused adrenaline. The model parameters are identified using least square regression on published data of adrenaline kinetics measured in a number of different clinical studies.

RESULTS

Simulation of published adrenaline measurements shows agreement with data of adrenaline infusion (R(2) = 0.9), exercise (R(2) = 0.97), and hypoglycemic episodes (R(2) = 0.93-0.97). The identified function describing adrenaline secretion during hypoglycemia shows an exponential increase for a blood glucose decreasing below 3.5 mmol/L and an approaching maximum around 1 mmol/L. Exercise intensity increasing above 50% of maximal oxygen uptake maximum causes approximately exponential increase in adrenaline secretion.

CONCLUSION

The model is a simple tool that can be used to simulate and predict adrenaline concentrations in situations of hypoglycemia, physical exercise, and adrenaline infusion. In conclusion, the developed model, although simple, seems to be useful for simulating adrenaline dynamics in situations with hypoglycemic episodes, physical exercise, or infusion.

The impact of non-model-related variability on blood glucose prediction

BACKGROUND

Physiological models are frequently used to predict blood glucose values from insulin and meal data of people with diabetes. Obviously, errors in the input data used result in prediction errors. A more complex problem is that no model may include all factors influencing the blood glucose level in any given situation. We have analyzed the influence of five parameters on prediction accuracy with respect to the time horizon.

METHODS

A physiological model, consisting of an insulin model, a meal model, and a glucose metabolism model in combination with a Monte Carlo simulation, was used for this investigation. It was used to examine the change in blood glucose following the intake of carbohydrate and insulin. The intra-individual variability, which was studied, included pharmacokinetic variability of insulin aspart and estimation error of carbohydrate intake, as well as the accuracy of blood glucose meters and insulin pens.

RESULTS

Simulations showed how the coefficient of variance for the different model compartments changes over time. For average people with diabetes the inaccuracies of blood glucose meters and carbohydrate estimates contribute to more than half of the variance.

CONCLUSION

We showed how blood glucose prediction is severely affected by the inaccuracy in the input variables. Metabolic fluctuations, causing variability in insulin dynamics, also display important effects, but these are difficult to change. The inaccuracy of carbohydrate counting and the use of blood glucose meters appear to be the two main sources of error, which can be reduced through better patient education.

A Study of Trained Clinicians’ Blood Glucose Predictions Based on Diaries of People with Type 1 Diabetes

Objectives: How accurate can trained clinicians predict blood glucose concentrations? Good clinical treatment is, among other things, related to understanding the factors influencing blood glucose level. We analyze trained clinician’s prediction accuracy in comparison with selected computer-implemented prediction algorithms and models.

Methods: We have in this study included diaries of 12 people with type 1 diabetes. This test group consists of seven males and five females, ages 24 to 60, HbA1c 6.0 to 8.9 and a BMI between 20 and 28 kg/m2. Eight experienced clinicians tried to predict the blood glucose measurements based on minimum three days of diary history. Selected prediction algorithms and models were used for comparison. The reason we focus on type 1 diabetes is that it has the most critical insulin requirement, so accurate prediction can be more critical than for type 2.

Results: An accuracy of 28.5% and an error of 26.7% were found from predictions made by the clinicians. A physiological model and an artificial intelligence model showed higher accuracy of 32.2% and 34.2% in comparison with the clinicians (p < 0.05). A simple predictor algorithm based on the mean blood glucose history showed significant (p < 0.05) lower total root mean square error compared to predictions made by the clinicians.

Conclusion: To predict blood glucose level from diaries has shown to be profoundly difficult even for experienced clinicians in comparison with predictions from computer algorithms and models. This suggests that computer-based systems incorporating predicting algorithms and models are likely to contribute positively to the day-to-day treatment of people with diabetes.

A study of trained clinicians' blood glucose predictions based on diaries of people with type 1 diabetes

OBJECTIVES

How accurate can trained clinicians predict blood glucose concentrations? Good clinical treatment is, among other things, related to understanding the factors influencing blood glucose level. We analyze trained clinician's prediction accuracy in comparison with selected computer-implemented prediction algorithms and models.

METHODS

We have in this study included diaries of 12 people with type 1 diabetes. This test group consists of seven males and five females, ages 24 to 60, HbA1c 6.0 to 8.9 and a BMI between 20 and 28 kg/m2. Eight experienced clinicians tried to predict the blood glucose measurements based on minimum three days of diary history. Selected prediction algorithms and models were used for comparison. The reason we focus on type 1 diabetes is that it has the most critical insulin requirement, so accurate prediction can be more critical than for type 2.

RESULTS

An accuracy of 28.5% and an error of 26.7% were found from predictions made by the clinicians. A physiological model and an artificial intelligence model showed higher accuracy of 32.2% and 34.2% in comparison with the clinicians (p<0.05). A simple predictor algorithm based on the mean blood glucose history showed significant (p<0.05) lower total root mean square error compared to predictions made by the clinicians.

CONCLUSION

To predict blood glucose level from diaries has shown to be profoundly difficult even for experienced clinicians in comparison with predictions from computer algorithms and models. This suggests that computer-based systems incorporating predicting algorithms and models are likely to contribute positively to the day-to-day treatment of people with diabetes.