Neural-Net Artificial Pancreas: A Randomized Crossover Trial of a First-in-Class Automated Insulin Delivery Algorithm

BACKGROUND

Automated Insulin Delivery (AID) is now integral to the clinical practice of Type 1 diabetes (T1D). The objective of this pilot-feasibility study was to introduce a new regulatory and clinical paradigm - a Neural-Net Artificial Pancreas (NAP) - an encoding of an AID algorithm into a neural network that approximates its action, and assess NAP vs the original AID algorithm.

METHODS

The UVA model-predictive control (UMPC) algorithm was encoded into a neural network, creating its NAP approximation. Seventeen AID users with T1D were recruited and 15 participated in two consecutive 20-hour hotel sessions, receiving in random order either NAP or UMPC. Their demographic characteristics were: ages 22-68 years old, duration of diabetes 7-58 years, gender 10/5 female/male, White Non-Hispanic/Black 13/2, and baseline HbA1c 5.4-8.1%.

RESULTS

The time-in-range (TIR) difference between NAP and UMPC, adjusted for entry glucose level, was 1 percentage point, with absolute TIR values of 86% (NAP) and 87% (UMPC). The two algorithms achieved similar times <70 mg/dL of 2.0% vs 1.8% and coefficients of variation of 29.3% (NAP) vs 29.1 (UMPC)%. Under identical inputs, the average absolute insulin-recommendation difference was 0.031 units/hour. There were no serious adverse events on either controller. NAP had 6-fold lower computational demands than UMPC.

CONCLUSION

In a randomized crossover study, a neural-network encoding of a complex model-predictive control algorithm demonstrated similar performance, at a fraction of the computational demands. Regulatory and clinical doors are therefore open for contemporary machine learning methods to enter the AID field.

COVID-19 Vaccine Hesitancy Among US Adults: Safety and Effectiveness Perceptions and Messaging to Increase Vaccine Confidence and Intent to Vaccinate

OBJECTIVE

Public health agencies have a critical role in providing effective messaging about mitigation strategies during a public health emergency. The objectives of this study were (1) to understand perceptions of COVID-19 vaccines, including concerns about side effects, safety, and effectiveness and how these perceptions influence vaccine decision-making among US adults and (2) to learn what messages might motivate vaccine uptake.

METHODS

In April and May 2021, we conducted 14 online focus groups with non-Hispanic English-speaking and English- and Spanish-speaking Hispanic adults (N = 99) not vaccinated against COVID-19. We oversampled adults aged 18-39 years and rural residents and systematically assessed 10 test messages. Researchers used a standardized guide and an a priori codebook for focus group discussions, coding transcripts, and thematic analysis.

RESULTS

Vaccine hesitancy factors included fear of the unknown; long-term side effects, including infertility; and beliefs that the vaccines were developed too quickly and were not sufficiently effective. Motivating factors for receiving vaccination included the ability to safely socialize and travel. Health care providers were considered important trusted messengers. Participants were critical of most messages tested. Messages that came across as "honest" about what is not yet known about COVID-19 vaccines were perceived more positively than other messages tested. Messages were seen as ineffective if perceived as vague or lacking in data and specificity.

CONCLUSIONS

Messages that were simple and transparent about what is unknown about vaccines relative to emerging science were viewed most favorably. Health care providers, friends, and family were considered influential in vaccination decision-making. Findings underscore the benefits of research-informed strategies for developing and disseminating effective messages addressing critical issues in a public health emergency.

Automated Insulin Delivery with SGLT2i Combination Therapy in Type 1 Diabetes

Background: Use of sodium-glucose cotransporter 2 inhibitors (SGLT2i) as adjunct therapy to insulin in type 1 diabetes (T1D) has been previously studied. In this study, we present data from the first free-living trial combining low-dose SGLT2i with commercial automated insulin delivery (AID) or predictive low glucose suspend (PLGS) systems. Methods: In an 8-week, randomized, controlled crossover trial, adults with T1D received 5 mg/day empagliflozin (EMPA) or no drug (NOEMPA) as adjunct to insulin therapy. Participants were also randomized to sequential orders of AID (Control-IQ) and PLGS (Basal-IQ) systems for 4 and 2 weeks, respectively. The primary endpoint was percent time-in-range (TIR) 70-180 mg/dL during daytime (7:00-23:00 h) while on AID (NCT04201496). Findings: A total of 39 subjects were enrolled, 35 were randomized, 34 (EMPA; n = 18 and NOEMPA n = 16) were analyzed according to the intention-to-treat principle, and 32 (EMPA; n = 16 and NOEMPA n = 16) completed the trial. On AID, EMPA versus NOEMPA had higher daytime TIR 81% versus 71% with a mean estimated difference of +9.9% (confidence interval [95% CI] 0.6-19.1); p = 0.04. On PLGS, the EMPA versus NOEMPA daytime TIR was 80% versus 63%, mean estimated difference of +16.5% (95% CI 7.3-25.7); p < 0.001. One subject on SGLT2i and AID had one episode of diabetic ketoacidosis with nonfunctioning insulin pump infusion site occlusion contributory. Interpretation: In an 8-week outpatient study, addition of 5 mg daily empagliflozin to commercially available AID or PLGS systems significantly improved daytime glucose control in individuals with T1D, without increased hypoglycemia risk. However, the risk of ketosis and ketoacidosis remains. Therefore, future studies with SGLT2i will need modifications to closed-loop control algorithms to enhance safety.

Evening and overnight closed-loop control versus 24/7 continuous closed-loop control for type 1 diabetes: a randomised crossover trial

Background

Automated closed-loop control (CLC), known as the "artificial pancreas" is emerging as a treatment option for Type 1 Diabetes (T1D), generally superior to sensor-augmented insulin pump (SAP) treatment. It is postulated that evening-night (E-N) CLC may account for most of the benefits of 24-7 CLC; however, a direct comparison has not been done.

Methods

In this trial (NCT02679287), adults with T1D were randomised 1:1 to two groups, which followed different sequences of four 8-week sessions, resulting in two crossover designs comparing SAP vs E-N CLC and E-N CLC vs 24-7 CLC, respectively. Eligibility: T1D for at least 1 year, using an insulin pump for at least six months, ages 18 years or older. Primary hypothesis: E-N CLC compared to SAP will decrease percent time <70mg/dL (3.9mmol/L) measured by continuous glucose monitoring (CGM) without deterioration in HbA1c. Secondary Hypotheses: 24-7 CLC compared to SAP will increase CGM-measured time in target range (TIR, 70-180mg/dL; 3.9-10mmol/L) and will reduce glucose variability during the day.

Findings

Ninety-three participants were randomised and 80 were included in the analysis, ages 18-69 years; HbA1c levels 5.4-10.6%; 66% female. Compared to SAP, E-N CLC reduced overall time <70mg/dL from 4.0% to 2.2% () resulting in an absolute difference of 1.8% (95%CI: 1.2-2.4%), p<0.0001. This was accompanied by overall reduction in HbA1c from 7.4% at baseline to 7.1% at the end of study, resulting in an absolute difference of 0.3% (95% CI: 0.1-0.4%), p<0.0001. There were 5 severe hypoglycaemia adverse events attributed to user-directed boluses without malfunction of the investigational device, and no diabetic ketoacidosis events.

Interpretation

In type 1 diabetes, evening-night closed-loop control was superior to sensor-augmented pump therapy, achieving most of the glycaemic benefits of 24-7 closed-loop.

Continuous Glucose Monitoring and Insulin Informed Advisory System with Automated Titration and Dosing of Insulin Reduces Glucose Variability in Type 1 Diabetes Mellitus

BACKGROUND

Glucose variability (GV) remains a key limiting factor in the success of diabetes management. While new technologies, for example, accurate continuous glucose monitoring (CGM) and connected insulin delivery devices, are now available, current treatment standards fail to leverage the wealth of information generated. Expert systems, from automated insulin delivery to advisory systems, are a key missing element to richer, more personalized, glucose management in diabetes.

METHODS

Twenty four subjects with type 1 diabetes mellitus (T1DM), 15 women, 37 ± 11 years of age, hemoglobin A1c 7.2% ± 1%, total daily insulin (TDI) 46.7 ± 22.3 U, using either an insulin pump or multiple daily injections with carbohydrate counting, completed two randomized crossover 48-h visits at the University of Virginia, wearing Dexcom G4 CGM, and using either usual care or the UVA decision support system (DSS). DSS consisted of a combination of automated insulin titration, bolus calculation, and CHO treatment advice. During each admission, participants were exposed to a variety of meal sizes and contents and two 45-min bouts of exercise. GV and glucose control were assessed using CGM.

RESULTS

The use of DSS significantly reduced GV (coefficient of variation: 0.36 ± 08. vs. 0.33 ± 0.06, P = 0.045) while maintaining glycemic control (average CGM: 155.2 ± 27.1 mg/dL vs. 155.2 ± 23.2 mg/dL), by reducing hypoglycemia exposure (%<70 mg/dL: 3.8% ± 4.6% vs. 1.8% ± 2%, P = 0.018), with nonsignificant trends toward reduction of significant hyperglycemia overnight (%>250 mg/dL: 5.3% ± 9.5% vs. 1.9% ± 4.6%) and at mealtime (11.3% ± 14.8% vs. 5.8% ± 9.1%).

CONCLUSIONS

A CGM/insulin informed advisory system proved to be safe and feasible in a cohort of 24 T1DM subjects. Use of the system may result in reduced GV and improved protection against hypoglycemia.

Overnight Closed-Loop Control Improves Glycemic Control in a Multicenter Study of Adults With Type 1 Diabetes

CONTEXT

Closed-loop control (CLC) for the management of type 1 diabetes (T1D) is a novel method for optimizing glucose control, and strategies for individualized implementation are being developed.

OBJECTIVE

To analyze glycemic control in an overnight CLC system designed to "reset" the patient to near-normal glycemic targets every morning.

DESIGN

Randomized, crossover, multicenter clinical trial.

PARTICIPANTS

Forty-four subjects with T1D requiring insulin pump therapy.

INTERVENTION

Sensor-augmented pump therapy (SAP) at home vs 5 nights of CLC (active from 23:00 to 07:00) in a supervised outpatient setting (research house or hotel), with a substudy of 5 nights of CLC subsequently at home.

MAIN OUTCOME MEASURE

The percentage of time spent in the target range (70 to 180 mg/dL measured using a continuous glucose monitor).

RESULTS

Forty subjects (age, 45.5 ± 9.5 years; hemoglobin A1c, 7.4% ± 0.8%) completed the study. The time in the target range (70 to 180 mg/dL) significantly improved in CLC vs SAP over 24 hours (78.3% vs 71.4%; P = 0.003) and overnight (85.7% vs 67.6%; P < 0.001). The time spent in a hypoglycemic range (<70 mg/dL) decreased significantly in the CLC vs SAP group over 24 hours (2.5% vs 4.3%; P = 0.002) and overnight (0.9% vs 3.2%; P < 0.001). The mean glucose level at 07:00 was lower with CLC than with SAP (123.7 vs 145.3 mg/dL; P < 0.001). The substudy at home, involving 10 T1D subjects, showed similar trends with an increased time in target (70 to 180 mg/dL) overnight (75.2% vs 62.2%; P = 0.07) and decreased time spent in the hypoglycemic range (<70 mg/dL) overnight in CLC vs SAP (0.6% vs 3.7%; P = 0.03).

CONCLUSION

Overnight-only CLC increased the time in the target range over 24 hours and decreased the time in hypoglycemic range over 24 hours in a supervised outpatient setting. A pilot extension study at home showed a similar nonsignificant trend.

Stereotactic body radiotherapy

Extracranial stereotactic body radiotherapy (SBRT) has been developed and refined over the last 25 years as a means to precisely deliver ablative doses of hypofractionated radiotherapy to small targets located outside of the cranial vault. SBRT has armed the radiation oncologist with a therapeutic approach that allows for intensification of both dose delivered and fractionation regimen employed. As a consequence, tumor control rates have improved to levels that previously have been associated only with surgical resection. Several prospective phase I and II studies have evaluated the use of SBRT for non-small cell lung cancer (NSCLC), liver tumors, and spinal metastases. This article will give an overview of SBRT and evidence for its use in the most common sites of disease for which it is employed today.

Adding heart rate signal to a control-to-range artificial pancreas system improves the protection against hypoglycemia during exercise in type 1 diabetes

BACKGROUND

We present a clinical trial establishing the feasibility of a control-to-range (CTR) closed-loop system informed by heart rate (HR) and assess the effect of HR information added to CTR on the risk for hypoglycemia during and after exercise.

SUBJECTS AND METHODS

Twelve subjects with type 1 diabetes (five men, seven women; weight, 68.9 ± 3.1 kg; age, 38 ± 3.3 years; glycated hemoglobin, 6.9 ± 0.2%) participated in a randomized crossover clinical trial comparing CTR versus CTR+HR in two 26-h admissions, each including 30 min of mild exercise. The CTR algorithm was implemented in the DiAs portable artificial pancreas platform based on an Android(®) (Google, Mountainview, CA) smartphone. We assessed blood glucose (BG) decline during exercise, the Low BG Index (LBGI) (a measure of hypoglycemic risk), number of hypoglycemic episodes (BG <70 mg/dL) and overall glucose control (percentage time within the target range 70 mg/dL ≤ BG ≤ 180 mg/dL).

RESULTS

Using HR to inform the CTR algorithm reduced significantly the BG decline during exercise (P=0.022), indicated marginally lower LBGI (P=0.3) and fewer hypoglycemic events during exercise (none vs. two events; P=0.16), and resulted in overall higher percentage time within the target range (81% vs. 75%; P=0.2). LBGI and average BG remained unchanged overall, during recovery, and overnight.

CONCLUSIONS

HR-informed closed-loop control can be implemented in a portable artificial pancreas. Although closed loop has been shown to reduce hypoglycemia, adding HR signal may further limit the risk for hypoglycemia during and immediately after exercise. The most prominent effect of adding HR information is reduced BG decline during exercise, without deterioration of overall glycemic control.

A double-blind trial of Dextran-haemodilution vs. placebo in claudicants

Haemodilution is often recommended for peripheral arterial disease, yet little data is available to support its clinical efficacy. This study was designed to prove or disprove the effectiveness of Dextran-haemodilution in intermittent claudication. Twenty claudicants with long, well-collateralized arterial occlusions were randomized into groups 1 and 2. Group 1 received isovolemic haemodilution with Dextran 40 (500 ml per session) during 3 weeks, which was followed by a wash-out period, followed by placebo treatments for 3 weeks. In group 2 this sequence was reversed. Pain-free and maximal walking distances were measured by standardized treadmill tests along with plethysmographic blood flow, Doppler pressures, haematocrit, blood and plasma viscosity as well as fibrinogen. Walking distances increased significantly by about 50% during haemodilution in both groups. This was paralleled by a fall in haematocrit and blood viscosity. All other variables remained constant. During placebo treatments there were no significant changes of any variable. The treatment was tolerated without complications. Thus Dextran-haemodilution seems safe and effective in selected peripheral occlusive arterial disease (POAD) patients. Potential responders might be identifiable before the start of therapy by angiographic investigations. The clinical effectiveness of Dextran 40 is comparable to that of hydroxyethyl starch 200 as reported in the literature.

Placebo-controlled, double-blind trial of hemodilution in peripheral occlusive arterial disease

A number of uncontrolled trials has suggested that hemodilution is effective in peripheral occlusive arterial disease. This study was aimed at proving or disproving the efficacy of hemodilution by using a randomized, placebo-controlled, double-blind crossover design. Twenty-four stable claudicants with long, collateralized femoropopliteal obstructions were treated with isovolemic hemodilution with 500 mL of 10% hydroxyethyl starch 200 and sham dilutions. Three weeks of hemodilution lowered blood and plasma viscosity, as well as hematocrit, and increased resting blood flow and pain-free walking distance. Placebo treatments produced no such favorable changes. It is concluded that hemodilution therapy can be clinically effective in patients with arterial obstructions in the lower extremities. Hemodilution seems particularly promising under hemodynamic condition of low shear stresses in vivo.

Placebo-controlled, double-blind study of haemodilution in peripheral arterial disease

Twenty-four men who had stable claudication with long, collateralized femoropopliteal obstructions were treated by isovolaemic haemodilution and sham dilutions in a double-blind, placebo-controlled, crossover study. 3 weeks of haemodilution lowered haematocrit and blood viscosity and increased resting blood flow and pain-free walking distance. Sham treatment produced no such favourable changes. It is concluded that haemodilution therapy can be clinically effective in patients with arterial obstructions in the legs. It seems particularly promising when the vascular obstructions are such that low average shear stresses act on the blood in multiple, long collaterals.

Importance of the preoperative haemoglobin concentration in arterial surgery

A study of reconstructive operations for ischaemia of the legs showed a correlation between a high preoperative haemoglobin level and early postoperative reocclusion. The mean preoperative haemoglobin level of patients with early reocclusion (165 +/- 3 g/l, SEM) was significantly higher (p less than 0.001) than that of patients who did not reocclude (153 +/- 3 g/l). There was no significant difference in the total serum protein concentrations, suggesting that dehydration was not the cause of the difference observed in haemoglobin concentration. The mechanism may be the lower flow rate associated with the higher haemoglobin content and increased blood viscosity.