Assisting the implementation of screening for type 1 diabetes by using artificial intelligence on publicly available data.

The type 1 diabetes community is coalescing around the benefits and advantages of early screening for disease risk. To be accepted by healthcare providers, regulatory authorities and payers, screening programmes need to show that the testing variables allow accurate risk prediction and that individualised risk-informed monitoring plans are established, as well as operational feasibility, cost-effectiveness and acceptance at population level. Artificial intelligence (AI) has the potential to contribute to solving these issues, starting with the identification and stratification of at-risk individuals. ASSET (AI for Sustainable Prevention of Autoimmunity in the Society; www.asset.healthcare ) is a public/private consortium that was established to contribute to research around screening for type 1 diabetes and particularly to how AI can drive the implementation of a precision medicine approach to disease prevention. ASSET will additionally focus on issues pertaining to operational implementation of screening. The authors of this article, researchers and clinicians active in the field of type 1 diabetes, met in an open forum to independently debate key issues around screening for type 1 diabetes and to advise ASSET. The potential use of AI in the analysis of longitudinal data from observational cohort studies to inform the design of improved, more individualised screening programmes was also discussed. A key issue was whether AI would allow the research community and industry to capitalise on large publicly available data repositories to design screening programmes that allow the early detection of individuals at high risk and enable clinical evaluation of preventive therapies. Overall, AI has the potential to revolutionise type 1 diabetes screening, in particular to help identify individuals who are at increased risk of disease and aid in the design of appropriate follow-up plans. We hope that this initiative will stimulate further research on this very timely topic.

A decade of improved glycemic control in young children with type 1 diabetes: A population-based cohort study.

BACKGROUND: Early-onset type 1 diabetes (T1D) is associated with high risk of early cardiovascular complications and premature death. The strongest modifiable risk factor is HbA1c. Other modifiable factors, such as overweight, also increase the risk of complications. During the last decade, the introduction of continuous glucose monitoring (CGM) has offered new options in the treatment of T1D. OBJECTIVE: To compare treatment outcomes in children younger than 7 years with T1D in Sweden in two separate cohorts: one in 2008 and one in 2018. METHODS: All children in the national pediatric diabetes registry (SWEDIABKIDS) younger than 7 years with T1D were included. Data from 2008 and 2018 were analyzed. RESULTS: Data were available on 666 children (45% girls) in 2008 and 779 children (45% girls) in 2018. Mean age was 5.6 (1.4) versus 5.5 (1.4) years and mean diabetes duration 2.3 (1.4) versus 2.2 (1.4) years. The use of CGM increased from 0% to 98% and the use of an insulin pump from 40% in 2008 to 82% (p < 0.01)in 2018.Mean HbA1c was 58 mmol/mol (7.4%) in 2008 and 50 mmol/mol (6.7%) in 2018 (p < 0.01). The frequency of overweight and obesity was the same in 2008 and 2018(26% vs. 29%). CONCLUSION: During this decade, usage of CGM and insulin pump increased and HbA1c decreased. However, HbA1c remained higher than the physiological level and thus continued to represent a cardiovascular risk, especially in combination with overweight or obesity. The frequency of overweight and obesity remained unchanged.

Changes to care delivery at nine international pediatric diabetes clinics in response to the COVID-19 global pandemic.

BACKGROUND: Pediatric diabetes clinics around the world rapidly adapted care in response to COVID-19. We explored provider perceptions of care delivery adaptations and challenges for providers and patients across nine international pediatric diabetes clinics. METHODS: Providers in a quality improvement collaborative completed a questionnaire about clinic adaptations, including roles, care delivery methods, and provider and patient concerns and challenges. We employed a rapid analysis to identify main themes. RESULTS: Providers described adaptations within multiple domains of care delivery, including provider roles and workload, clinical encounter and team meeting format, care delivery platforms, self-management technology education, and patient-provider data sharing. Providers reported concerns about potential negative impacts on patients from COVID-19 and the clinical adaptations it required, including fears related to telemedicine efficacy, blood glucose and insulin pump/pen data sharing, and delayed care-seeking. Particular concern was expressed about already vulnerable patients. Simultaneously, providers reported 'silver linings' of adaptations that they perceived as having potential to inform care and self-management recommendations going forward, including time-saving clinic processes, telemedicine, lifestyle changes compelled by COVID-19, and improvements to family and clinic staff literacy around data sharing. CONCLUSIONS: Providers across diverse clinical settings reported care delivery adaptations in response to COVID-19-particularly telemedicine processes-created challenges and opportunities to improve care quality and patient health. To develop quality care during COVID-19, providers emphasized the importance of generating evidence about which in-person or telemedicine processes were most beneficial for specific care scenarios, and incorporating the unique care needs of the most vulnerable patients.

Multiplexing DNA methylation markers to detect circulating cell-free DNA derived from human pancreatic ß cells.

It has been proposed that unmethylated insulin promoter fragments in plasma derive exclusively from ß cells, reflect their recent demise, and can be used to assess ß cell damage in type 1 diabetes. Herein we describe an ultrasensitive assay for detection of a ß cell-specific DNA methylation signature, by simultaneous assessment of 6 DNA methylation markers, that identifies ß cell DNA in mixtures containing as little as 0.03% ß cell DNA (less than 1 ß cell genome equivalent). Based on this assay, plasma from nondiabetic individuals (N = 218, aged 4-78 years) contained on average only 1 ß cell genome equivalent/mL. As expected, cell-free DNA (cfDNA) from ß cells was significantly elevated in islet transplant recipients shortly after transplantation. We also detected ß cell cfDNA in a patient with KATP congenital hyperinsulinism, in which substantial ß cell turnover is thought to occur. Strikingly, in contrast to previous reports, we observed no elevation of ß cell-derived cfDNA in autoantibody-positive subjects at risk for type 1 diabetes (N = 32), individuals with recent-onset type 1 diabetes (<4 months, N = 92), or those with long-standing disease (>4 months, N = 38). We discuss the utility of sensitive ß cell cfDNA analysis and potential explanations for the lack of a ß cell cfDNA signal in type 1 diabetes.

Translating glycated hemoglobin A1c into time spent in glucose target range: A multicenter study.

BACKGROUND: Approximately 90% of children and adolescents with type 1 diabetes in Sweden use continuous glucose monitoring (CGM), either as real-time CGM or intermittently scanned CGM to monitor their glucose levels. Time in target range (TIT) is an easily understandable metric for assessing glycemic control. OBJECTIVE: The aim of this study was to examine the relation between TIT and hemoglobin A1c (HbA1c). SUBJECTS AND METHODS: Subjects were recruited from three diabetes care centers in Sweden. Glucose data were collected for 133 children and adolescents with type 1 diabetes through CGM using Diasend. Subjects with registration time over 80% were included in the analysis. HbA1c was collected from SWEDIABKIDS, the Swedish pediatric diabetes quality registry. TIT was defined as 3.9 to 7.8 mmol/L (70-140 mg/dL) and time in range (TIR) as 3.9 to 10 mmol/L (70-180 mg/dL). RESULTS: During the period of 60 days, 105 subjects provided complete data for analysis. Mean age was 12.2 (±3.3) years, mean HbA1c was 53.9 (±8.2) mmol/mol or 7.1% (±0.7%). Mean sensor glucose value was 8.6 (±1.3) mmol/L, mean coefficient of variation was 42.2% (±7.2%), mean TIT was 40.9% (±SD 12.2%), and mean TIR was 60.8% (±13.1%). There was a significant nonlinear relation between TIT during 60 days and HbA1c, R2 = 0.69. CONCLUSION: This study suggests a nonlinear relation between time spent in glucose target range and HbA1c. The finding implies that time spent in TIT could be a useful metric in addition to HbA1c to assess glycemic control.

Continuous Glucose Monitoring in Healthy Children Aged 2-8 Years.

BACKGROUND: The purpose of this study was to add the missing information on glycemic levels and patterns as measured by continuous glucose monitoring (CGM) in the daily life of healthy children aged 2-8 years. These data are needed when studying glycemic patterns and treatment outcome in children aged 2-8 years with diabetes. METHODS: Each of the 15 healthy children aged 2-7.99 years used a CGM device (Dexcom G4 Platinum) for 7 days. RESULTS: A total of 15 children (10 girls) aged 5.4 ± 1.6 years registered a mean of 1976 ± 15 counts. Mean sensor glucose was 5.3 ± 1.0 mmol/L (95 ± 18 mg/dL) and 89% of values were in the range 4-7.8 mmol/L (72-140 mg/dL), 9% of sensor glucose values were <4.0 mmol/L (72 mg/dL), and 2% of sensor glucose values were >7.8 mmol/L (140 mg/dL). CONCLUSION: We present glycemic data as measured by CGM in healthy children aged 2-8 years.

Insulin pump therapy in children with type 1 diabetes: analysis of data from the SWEET registry.

BACKGROUND: Intensified insulin delivery using multiple daily injections (MDI) or continuous subcutaneous insulin infusion (CSII) is recommended in children with type 1 diabetes (T1D) to achieve good metabolic control. OBJECTIVE: To examine the frequency of pump usage in T1D children treated in SWEET (Better control in Paediatric and Adolescent diabeteS: Working to crEate CEnTers of Reference) centers and to compare metabolic control between patients treated with CSII vs MDI. METHODS: This study included 16 570 T1D children participating in the SWEET prospective, multicenter, standardized diabetes patient registry. Datasets were aggregated over the most recent year of treatment for each patient. Data were collected until March 2016. To assess the organization of pump therapy a survey was carried out. RESULTS: Overall, 44.4% of T1D children were treated with CSII. The proportion of patients with pump usage varied between centers and decreased with increasing age compared with children treated with MDI. In a logistic regression analysis adjusting for age, gender and diabetes duration, the use of pump was associated with both: center size [odd ratio 1.51 (1.47-1.55), P < .0001) and the diabetes-related expenditure per capita [odd ratio 1.55 (1.49-1.61), P < .0001]. Linear regression analysis, adjusted for age, gender, and diabetes duration showed that both HbA1c and daily insulin dose (U/kg/d) remained decreased in children treated with CSII compared to MDI (P < .0001). CONCLUSIONS: Insulin pump therapy is offered by most Sweet centers. The differences between centers affect the frequency of use of modern technology. Despite the heterogeneity of centers, T1D children achieve relatively good metabolic control, especially those treated with insulin pumps and those of younger age.

Detection and treatment efficacy of hypoglycemic events in the everyday life of children younger than 7 yr.

BACKGROUND: Mild hypoglycemia is commonly observed in children treated for type 1 diabetes mellitus (T1DM). Hypoglycemia disturbs cognition and learning. OBJECTIVE: To describe how and to what extent hypoglycemia in young children with T1DM is detected in everyday life. To learn how parents and caregivers treat hypoglycemia and to evaluate how efficient this treatment is. METHODS: Twenty-three children [12 girls, mean age: 4.5 yr, mean HbA1c: 59 mmol/mol (7.5%)], 17 of whom were treated with an insulin pump, underwent blinded continuous glucose monitoring (CGM). Data on symptoms and treatment of hypoglycemia were collected in a logbook. Plasma glucose values were collected through self-monitoring of blood glucose and entered in the logbook, and glucometer memories were uploaded. Data were collected during 1 wk in autumn and 1 wk in spring. RESULTS: Only 32% of all hypoglycemic events were detected despite plasma glucose being checked 10 times per day. Most hypoglycemic events were asymptomatic (90% overall and 98% of those occurring at night). Untreated hypoglycemic events were associated with a relapse into hypoglycemia within 3 h in the majority of events. Compared to treatment of hypoglycemia events with a defined dose of simple carbohydrates, treatment with a mixed meal resulted in a significantly higher glucose value 1 and 2 h after the hypoglycemia. CONCLUSION: For optimum treatment, children younger than 7 yr with T1DM need better strategies and support for detecting hypoglycemia with real-time CGM. Hypoglycemia should be treated with a defined dose of carbohydrates rather than a mixed meal.

Transient neonatal diabetes, ZFP57, and hypomethylation of multiple imprinted loci: a detailed follow-up.

OBJECTIVE: Transient neonatal diabetes mellitus 1 (TNDM1) is the most common cause of diabetes presenting at birth. Approximately 5% of the cases are due to recessive ZFP57 mutations, causing hypomethylation at the TNDM locus and other imprinted loci (HIL). This has consequences for patient care because it has impact on the phenotype and recurrence risk for families. We have determined the genotype, phenotype, and epigenotype of the first 10 families to alert health professionals to this newly described genetic subgroup of diabetes. RESEARCH DESIGN AND METHODS: The 10 families (14 homozygous/compound heterozygous individuals) with ZFP57 mutations were ascertained through TNDM1 diagnostic testing. ZFP57 was sequenced in probands and their relatives, and the methylation levels at multiple maternally and paternally imprinted loci were determined. Medical and family histories were obtained, and clinical examination was performed. RESULTS: The key clinical features in probands were transient neonatal diabetes, intrauterine growth retardation, macroglossia, heart defects, and developmental delay. However, the finding of two homozygous relatives without diabetes and normal intelligence showed that the phenotype could be very variable. The epigenotype always included total loss of methylation at the TNDM1 locus and reproducible combinations of differential hypomethylation at other maternally imprinted loci, including tissue mosaicism. CONCLUSIONS: There is yet no clear genotype-epigenotype-phenotype correlation to explain the variable clinical presentation, and this results in difficulties predicting the prognosis of affected individuals. However, many cases have a more severe phenotype than seen in other causes of TNDM1. Further cases and global epigenetic testing are needed to clarify this.