Real-time Continuous Glucose Monitoring During a Hyperinsulinemic-Hypoglycemic Clamp Significantly Underestimates the Degree of Hypoglycemia

Farrell, Catriona M.; McNeilly, Alison D.; Hapca, Simona M.; McCrimmon, Rory J.

doi:10.2337/dc20-0882

Cited by 12 publications

(11 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here, unlike the common definition for level 1 hypoglycemia based on the threshold of 70 mg/dL, we instead choose 80 mg/dL as the hypoglycemia threshold. This is because recent results by Farrell et al 57 have revealed a measurement artifact, i.e., that the real-time continuous glucose monitoring (CGM), where we would expect these algorithms to have clinical applicability, underestimates the degree of hypoglycemia by a difference of 10 mg/dL, as shown in Supplementary Fig. 1 .…”

Section: Resultsmentioning

confidence: 99%

“…(%) 21 (52.5) Body compositions Body mass, kg 81.0 (71.3, 94.2), 84.1 ± 18.7 Height, m 1.64 (1.59, 1.73), 1.66 ± 0.10 BMI, kg/m 2 29.7 (26.6, 33.1), 30.1 ± 5.1 Hormone levels Cortisol, μg/dL 15.9 (13.0, 20.2), 16.1 ± 6.0 Leptin, ng/dL 19.8 (9.57, 31.1), 23.0 ± 17.8 Fasting glucose, mg/dL 117.5 ± 17.9 Insulin, μIU/mL 13.33 ± 13.29 HOMA1-IR 3.51 ± 3.47 Blood glucose data brief Data reading length (h) 90 (82, 170), 117 ± 63 Model input BG length (min) 30 Hypoglycemia threshold (mg/dL) 80 Hyperglycemia threshold (mg/dL) 180 HbA1c (%) 7.33 ± 1.31 HOMA1-IR the homeostatic model assessment index for insulin resistance. We choose 80 mg/dL as the hypoglycemia threshold, because recent results by Farrell et al 57 have revealed a measurement artifact, i.e., that the real-time CGM underestimates the degree of hypoglycemia by a difference of 10 mg/dL, as shown in Supplementary Fig. 1 .…”

Section: Methodsmentioning

confidence: 99%

“… HOMA1-IR the homeostatic model assessment index for insulin resistance. We choose 80 mg/dL as the hypoglycemia threshold, because recent results by Farrell et al 57 have revealed a measurement artifact, i.e., that the real-time CGM underestimates the degree of hypoglycemia by a difference of 10 mg/dL, as shown in Supplementary Fig. 1 .…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Deep transfer learning and data augmentation improve glucose levels prediction in type 2 diabetes patients

Deng

Becerra

et al. 2021

npj Digit. Med.

View full text Add to dashboard Cite

Accurate prediction of blood glucose variations in type 2 diabetes (T2D) will facilitate better glycemic control and decrease the occurrence of hypoglycemic episodes as well as the morbidity and mortality associated with T2D, hence increasing the quality of life of patients. Owing to the complexity of the blood glucose dynamics, it is difficult to design accurate predictive models in every circumstance, i.e., hypo/normo/hyperglycemic events. We developed deep-learning methods to predict patient-specific blood glucose during various time horizons in the immediate future using patient-specific every 30-min long glucose measurements by the continuous glucose monitoring (CGM) to predict future glucose levels in 5 min to 1 h. In general, the major challenges to address are (1) the dataset of each patient is often too small to train a patient-specific deep-learning model, and (2) the dataset is usually highly imbalanced given that hypo- and hyperglycemic episodes are usually much less common than normoglycemia. We tackle these two challenges using transfer learning and data augmentation, respectively. We systematically examined three neural network architectures, different loss functions, four transfer-learning strategies, and four data augmentation techniques, including mixup and generative models. Taken together, utilizing these methodologies we achieved over 95% prediction accuracy and 90% sensitivity for a time period within the clinically useful 1 h prediction horizon that would allow a patient to react and correct either hypoglycemia and/or hyperglycemia. We have also demonstrated that the same network architecture and transfer-learning methods perform well for the type 1 diabetes OhioT1DM public dataset.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Deep transfer learning and data augmentation improve glucose levels prediction in type 2 diabetes patients

Deng

Becerra

et al. 2021

npj Digit. Med.

View full text Add to dashboard Cite

show abstract

“…Whether a stricter criterion is necessary for patients unable to master the technology entirely is a question to address in future investigations. It is worth noting that increasing evidence has shown that CGM underestimates the degree of hypoglycemia, suggesting that CGM may worsen the severity and delay the treatment of hypoglycemia [ 19 ].…”

Section: Discussionmentioning

confidence: 99%

Sudden Death Associated with Severe Hypoglycemia in a Diabetic Patient During Sensor-Augmented Pump Therapy with the Predictive Low Glucose Management System

et al. 2021

View full text Add to dashboard Cite

“… 48 – 50 One issue emerging for CGM and FGM is that both devices are less accurate during hypoglycaemia (e.g. Freckmann et al and Farrell et al 51 , 52 ), and may therefore not detect all hypoglycaemic events. This may explain why they have not been shown in clinical trials to improve hypoglycaemia awareness.…”

Section: Technologymentioning

confidence: 99%

Clinical approaches to treat impaired awareness of hypoglycaemia

Farrell

McCrimmon

2021

Therapeutic Advances in Endocrinology

Self Cite

View full text Add to dashboard Cite

Impaired awareness of hypoglycaemia (IAH) affects between 25% and 30% of all people with type 1 diabetes (T1D) and markedly increases risk of severe hypoglycaemia. This greatly feared complication of T1D impairs quality of life and has a recognised morbidity. People with T1D have an increased propensity to hypoglycaemia as a result of fundamental physiological defects in their ability to respond appropriately to a fall in blood glucose levels. With repeated exposure to low glucose, many then develop a condition referred to as IAH, where there is a reduced ability to perceive the onset of hypoglycaemia and take appropriate corrective action. The management of individuals with IAH relies initially on its identification in the clinic through a detailed exploration of the frequency of hypoglycaemia and an assessment of the individual’s ability to recognise these episodes. In this review article, we will address the clinical strategies that may help in the management of the patient with IAH once identified, who may or may not also suffer from problematic hypoglycaemia. The initial focus is on how to identify such patients and then on the variety of approaches involving educational programmes and technological approaches that may be taken to minimise hypoglycaemia risk. No single approach can be advocated for all patients, and it is the role of the health care professional to identify the clinical strategy that best enables their patient to achieve this goal.

show abstract

Real-time Continuous Glucose Monitoring During a Hyperinsulinemic-Hypoglycemic Clamp Significantly Underestimates the Degree of Hypoglycemia

Abstract: Figure 1-Glucose data (mean 6 SEM) during hyperinsulinemic-hypoglycemic clamps. Black circles represent AV plasma glucose; blue squares, AV blood tested via SMBG; red triangles, CGM. ***P , 0.001. care.diabetesjournals.org Farrell and Associates e143

Cited by 12 publications

References 6 publications

Deep transfer learning and data augmentation improve glucose levels prediction in type 2 diabetes patients

Deep transfer learning and data augmentation improve glucose levels prediction in type 2 diabetes patients

Sudden Death Associated with Severe Hypoglycemia in a Diabetic Patient During Sensor-Augmented Pump Therapy with the Predictive Low Glucose Management System

Clinical approaches to treat impaired awareness of hypoglycaemia

Contact Info

Product

Resources

About