Background and Aims

The Proportional Integral Derivative controller (PID) is adopted in several applications including the artificial pancreas (AP). An advantage of PID is that only three parameters are needed and is normally subject-tailored with one personalization parameter, the Total Daily Insulin (TDI). No predictive models like in MPC are needed. The slow dynamics of the subcutaneous (SC) route limits PID effectiveness requiring the need of meal announcement. The faster dynamics of intraperitoneal (IP) delivery open the possibility of avoiding meal announcement. In this work, we propose a time-varying PID, test it in silico on the novel IP-Padova simulator (IP-T1DS) which incorporates intra- and inter-day variability of insulin sensitivity.

Methods

We identify the average IP insulin-glucose transfer function starting from data collected on the IP-T1DS, an extended version of the SC FDA accepted simulator. We consider the Insulin-to-Carbohydrate Ratio (CR) as personalization parameter as an alternative to TDI. From the CRs associated to the different portions of the day, we design a time-varying PID. We test the two PIDs simulating a 12-week protocol.

Results

Both controllers are effective without meal announcement, but the time-varying PID is less sensitive to the different day portions.

Conclusions

We propose a time-varying PID for a fully implantable IP AP. IP-PIDs are effective without meals announcement thanks to the faster dynamics of IP sensing and delivery. Personalization through CR faces insulin sensitivity variations, increasing control performance w.r.t. TDI-based personalization.

The work was supported by H2020-FETPROACT Project FORGETDIABETS, n. 951933.

Background and Aims

Background and aims: Evidence from clinical trials have demonstrated the glycemic benefits of hybrid closed-loop insulin delivery systems (HCLS) in children and adults. The primary objective of this study was to provide real-world data on glycemic outcomes and population characteristics in those using HCLS compared to other treatment modalities.

Methods

Methods: Electronic health record data (2019-2021) from the T1D Exchange Quality Improvement (T1DX-QI) Collaborative were analyzed for 15,091 people with T1D. Patients with data on insulin delivery mode [HCLS, sensor augmented pump therapy (SAP), pump only, or multiple daily injections (MDI)], A1c at their most recent clinic visit, and other demographic covariates were included in this analysis. Individuals were classified across insulin delivery groups based on information documented by a healthcare provider at their most recent clinic visit.

Results

Results: In this study population, 744 (5%) were HCLS users, 2,326 (15%) were SAP users, 3,879 (26%) used pump only, and 8,142 (54%) were MDI users. Median (IQR) A1c in the HCLS group was 7.3% (IQR: 6.7,8.0%) compared to 8.0% (IQR: 7.2,9.1%) in the SAP group, 8.3% (IQR: 7.2,10.0%) in the pump only group and 8.7% (IQR: 7.4,10.5%) in the MDI group. Linear mixed models showed HbA1c to be 0.7% lower in the HCLS group compared to SAP, adjusting for age, gender, race/ethnicity and insurance status [Estimated Marginal Mean (95% CI): 7.8% (IQR: 7.3,8.4%) vs 8.7% (IQR: 8.2,9.0%)].

Conclusions

Conclusion: This is a large multi-site real-world study demonstrating significant clinical benefits of HCLS use relative to other insulin treatment options.

Background and Aims

CASE-REPORT

DIABELOOP IN LONG STANDING TYPE 1 DIABETES WITH DEMENTIA

Menzen M¹, Ostrowski-Krause M¹, Otten F², Mohr S¹

1 Gemeinschaftskrankenhaus Bonn, Department of Endocrinology

2 Gemeinschaftskrankenhaus Bonn, Department of Geriatrics

85 years old woman with type 1 diabetes, diagnosed 1946. Now treated with multiple dose injection of insulin lispro to mealtime 1IU at morning, 0,5 IU at lunch and dinner per 10g carbohydrate and 6 Units of detemir once daily. HbA1c 60 mmol/mol, stimulated c-Peptid: 0,046ng/ml.

Concomitant diseases: Heart failure with preserved ejection fraction, pulmonary arterial hypertension. Frailty with severe sarcopenia (Frailty Share FI 75), incipient dementia (MMST 23/30) with dominant deficits in visuoconstructive skills.

2021 she suffered from two severe hypoglycemia stage III with falls and fracture of os pubis and cranial brain trauma. Both situations caused by unintentional double injection of bolus insulin. Even CGMS with Dexcom G6 supervised by social worker was not able to prevent the rapide drop of blood sugar.

Methods

We decided to switch therapy to automated insulin delivery with AccuChek insight, Dexcom G6 combined with Diabeloop DBLG1 with insulin aspart. No meal annoncement was done. She ate 40g of carbohydrates to each meal.

Results

Tissue glucose in mean was ~180mg/dl with no hypoglycemia and rise after meal up to a maximum of 284 mg/dl for a few minutes with rapid drop to upper range targets. Betahydroxybutrate measured in this situations has been in normal ranges.

Conclusions

This is the first case report of AID use in long standing type 1 diabetes with dementia to our knowledge.

Background and Aims

Despite being an off-label method of insulin delivery, increasing numbers of people with type1 diabetes (T1D) worldwide choose Loop, a form of Do-it-yourself Automated Insulin Delivery. We sought to collect data in an Edmonton cohort of known Loop users, to assess glycemic outcomes, safety and the perceived impact on quality of life (QOL).

Methods

A retrospective observational study of adults with T1D using Loop. Glycemic control (HbA1c andtime in range (TIR)) and safety outcomes (hospital admissions and time below range (TBR)) were assessed pre and post Loop use. QOL outcomes were explored using INSPIRE, Diabetes Impact and Device Satisfaction (DIDS) measures, and semi-structured interviews. Data are presented as mean +SD.

Results

24 adults; 66.7% female, mean age 37.0 +13.0 years, duration of diabetes 24.6 +11.8 years and 19.6 +11.5 months of Loop use. Glycemic control significantly improved with Loop; HbA1c 7.2 +1.0% vs. 7.9 +0.8%, p=0.002, and TIR (3.9-10.0mmol/L) 70.9 +15.6 % vs 57.7 +10.0%, p=0.004. There was a non-significant reduction in TBR (<3.9mmol/L) with Loop, 2.2 +1.3% vs pre-Loop 2.6 +1.8%, p= 0.161. Two episodes of DKA (one associated with concurrent sodium glucose transporter-2 inhibitor use) and no severe hypoglycaemia occurred in a total of 470 months Loop. Positive QOL impact was explored in qualitative analysis and indicated through INSPIRE 84.0 +17.9, Diabetes Impact 2.9 +0.9 and Device Satisfaction 8.9 +0.8 scores.

Conclusions

This local cohort who have chosen and continue to use Loop, demonstrate a beneficial impact on glycemic control and QOL, with no significant increase in hypoglycaemia or safety concerns highlighted.

Background and Aims

To compare glucose control with hybrid closed loop (HCL) when challenged by moderate-intensity exercise (MIE), high-intensity intermittent exercise (HIE) and resistance exercise (RE) while profiling counter-regulatory hormones, lactate, ketones, and kinetic data in adults with type 1 diabetes.

Methods

Open-label multisite randomized crossover trial. Adults with type 1 diabetes undertook 40-min of HIE, MIE, and RE in random order while using HCL (Medtronic 670G) with hypoglycaemia preventative measures including use of a temporary target and supplemental carbohydrates. Primary outcome was median (IQR) continuous glucose monitoring (CGM) time-in-range (TIR, 70-180 mg/dL) for 14-hours post-exercise commencement. Accelerometer data and venous glucose, ketones, lactate, and counter-regulatory hormones were measured for 280-min post-exercise commencement.

Results

Median TIR was 81% [67, 93]%, 91% [80, 94]%, and 80% [73, 89]% for 0-14 hours post-exercise commencement for HIE, MIE and RE, respectively (n=30), with no difference between exercise types (MIE v HIE; p=0.11, MIE v RE p=0.11, HIE v RE p=0.90). Time-below-range was 0% for all exercise bouts. For HIE and RE compared with MIE, there were greater increases respectively in noradrenaline (p=0.01, p=0.004), cortisol (p<0.001, p=0.001), lactate (p<0.001, p<0.001) and heart rate (p=0.007, p=0.015). During HIE compared with MIE, there were greater increases in growth hormone (p=0.024).

Conclusions

Under controlled conditions, HCL provided satisfactory glucose control with no difference between exercise type. Lactate, counter-regulatory hormones, and kinetic data, differentiate type and intensity of exercise, and their measurement may help inform insulin needs during exercise. However, their potential utility as insulin dosing modulators will be limited by subcutaneous insulin pharmacokinetics.

Background and Aims

Early evaluation of the Medtronic AHCL pivotal trial in youth demonstrated improved time in target range (TIR, 70-180mg/dL) and reduced time at >180mg/dL, compared with baseline.¹ The present study reports results on the complete pediatric cohort.

Methods

Participants (N=160, aged 7-17 years) with T1D underwent baseline run-in (~2 weeks) with HCL, SAP, PLGM or CSII therapy followed by a three-month AHCL-enabled study period where the 100mg/dL and the 120mg/dL target setpoints were used (~45 days for each). Analyses compared mean HbA1c, sensor glucose (SG), coefficient of variation of SG, time spent at SG ranges and insulin delivery between run-in and study (Wilcoxon signed-rank test or t-test). Safety data were also summarized.

Results

Outcomes from baseline to study end, and by glucose target, are shown (Table). With AHCL, those achieving the target TIR of >70% and HbA1c of <7.0% increased from 15% (N=24/160) to 51% (N=82/160) and 16% (N=25/160) to 26% (N=35/136), respectively. There was one severe hypoglycemic event (Baseline) and no diabetic ketoacidosis events.

Conclusions

Findings demonstrate increased TIR (by ~11%, ~2.6hrs/day) and reduced HbA1c (by -0.5%) with unchanged time in hypoglycemia, versus baseline therapy, in youth with T1D using the Medtronic AHCL system.

Background and Aims

The prevalence of diabetic retinopathy (DR) is increasing at an alarming rate in India. All diabetes patients need regular retina screening. The primary issue is the grading of fundus images by retina specialists, whose numbers are very scarce compared to the load of patients. Many patients are based in rural areas and cannot visit an ophthalmologist. Such limitations created interest in assessment of images using fully automated Artificial Intelligence based grading systems.

The study aims to evaluate the effectiveness of a deep learning algorithm in screening of DR and assess the prevalence of DR screened at multiple centers across Kolkata, West Bengal, India.

Methods

It is a multicentric cross sectional study. A total of 725 diabetes patients were screened for DR using Artificial Intelligence based Diabetic Retinopathy Screening System (AIDRSS) developed by ARTELUS™. The AIDRSS graded fundus images based on the ICDR Scale (DR0, DR1, DR2, DR3, DR4). The fundus images captured were also assessed by retina specialist and reports were compared to evaluate the sensitivity and specificity of the AIDRSS.

Results

The prevalence of DR screened using AIDRSS was 27.17% (DR1=17.38%, DR2=9.52%, Referable DR : DR3 and DR4=0.27%). The AIDRSS performed well with overall 97% sensitivity and 92% specificity; and 100% sensitivity in detecting referable DR when compared against fundus image reported by a retina specialist.

Conclusions

The study establishes a high prevalence of diabetic retinopathy. AIDRSS developed by ARTELUS™ has a high sensitivity and specificity and is an effective solution for routine screening and early detection of diabetic retinopathy in India.