Predicting the Risk of Developing Type 1 Diabetes Using a One-Week Continuous Glucose Monitoring Home Test With Classification Enhanced by Machine Learning: An Exploratory Study

BACKGROUND

Detection of two or more autoantibodies (Ab) in the blood might describe those individuals at increased risk of developing type 1 diabetes (T1D) during the following years. The aim of this exploratory study is to propose a high versus low T1D risk classifier using machine learning technology based on continuous glucose monitoring (CGM) home data.

METHODS

Forty-two healthy relatives of people with T1D with mean ± SD age of 23.8 ± 10.5 years, HbA1c (glycated hemoglobin) of 5.3% ± 0.3%, and BMI (body mass index) of 23.2 ± 5.2 kg/m2 with zero (low risk; N = 21), and ≥2 (high risk; N = 21) Ab, were enrolled in an NIH (National Institutes of Health)-funded TrialNet ancillary study. Participants wore a CGM for a week and consumed three standardized liquid mixed meals (SLMM) instead of three breakfasts. Glycemic features were extracted from two-hour post-SLMM CGM traces, compared across groups, and used in four supervised machine learning Ab risk status classifiers. Recursive Feature Elimination (RFE) algorithm was used for feature selection; classifiers were evaluated through 10-fold cross-validation, using the receiver operating characteristic area under the curve (AUC-ROC) to select the best classification model.

RESULTS

The percent time of glucose >180 mg/dL (T180), glucose range, and glucose CV (coefficient of variation) were the only significant differences between the glycemic features in the two groups with P values of .040, .035, and .028 respectively. The linear SVM (Support Vector Machine) model with RFE features achieved the best performance of classifying low-risk versus high-risk individuals with AUC-ROC = 0.88.

CONCLUSIONS

A machine learning technology, combining a potentially self-administered one-week CGM home test, has the potential to reliably assess the T1D risk.

Predicting Immunological Risk for Stage 1 and Stage 2 Diabetes Using a 1-Week CGM Home Test, Nocturnal Glucose Increments, and Standardized Liquid Mixed Meal Breakfasts, with Classification Enhanced by Machine Learning

Background: Predicting the risk for type 1 diabetes (T1D) is a significant challenge. We use a 1-week continuous glucose monitoring (CGM) home test to characterize differences in glycemia in at-risk healthy individuals based on autoantibody presence and develop a machine-learning technology for CGM-based islet autoantibody classification. Methods: Sixty healthy relatives of people with T1D with mean ± standard deviation age of 23.7 ± 10.7 years, HbA1c of 5.3% ± 0.3%, and body mass index of 23.8 ± 5.6 kg/m2 with zero (n = 21), one (n = 18), and ≥2 (n = 21) autoantibodies were enrolled in an National Institutes of Health TrialNet ancillary study. Participants wore a CGM for a week and consumed three standardized liquid mixed meals (SLMM) instead of three breakfasts. Glycemic outcomes were computed from weekly, overnight (12:00-06:00), and post-SLMM CGM traces, compared across groups, and used in four supervised machine-learning autoantibody status classifiers. Classifiers were evaluated through 10-fold cross-validation using the receiver operating characteristic area under the curve (AUC-ROC) to select the best classification model. Results: Among all computed glycemia metrics, only three were different across the autoantibodies groups: percent time >180 mg/dL (T180) weekly (P = 0.04), overnight CGM incremental AUC (P = 0.005), and T180 for 75 min post-SLMM CGM traces (P = 0.004). Once overnight and post-SLMM features are incorporated in machine-learning classifiers, a linear support vector machine model achieved the best performance of classifying autoantibody positive versus autoantibody negative participants with AUC-ROC ≥0.81. Conclusion: A new technology combining machine learning with a potentially self-administered 1-week CGM home test can help improve T1D risk detection without the need to visit a hospital or use a medical laboratory. Trial registration: ClinicalTrials.gov registration no. NCT02663661.

Essential Continuous Glucose Monitoring Metrics: The Principal Dimensions of Glycemic Control in Diabetes

Background: With the proliferation of continuous glucose monitoring (CGM), a number of metrics were developed to assess quality of glycemic control. Many of them are highly correlated. Thus, we aim to identify the principal dimensions of glycemic control-a minimal set of metrics, necessary and sufficient for comprehensive assessment of diabetes management. Methods: Seventy-five thousand five hundred sixty-three 2-week CGM profiles recorded in six studies by 790 individuals with type 1 or type 2 diabetes were used to compute mean glucose (MG), percentage time >180 mg/dL (TAR), >250 mg/dL (TAR2), <70 mg/dL (TBR), <54 mg/dL (TBR2), and coefficient of variation (CV). The true dimensionality of the glycemic-metric space was identified in a training set (53,380 profiles) and validated in an independent test set (22,183 profiles). Results: Correlation analysis identified two blocks of metrics-(MG, TAR, TAR2) and (TBR, TBR2, CV)-each with high internal correlation, but insignificant between-block correlation, suggesting that the true dimensionality of the glycemic-metric space is 2. Principal component analysis confirmed two essential metrics quantifying exposure to hyperglycemia (i.e., treatment efficacy) and risk for hypoglycemia (i.e., treatment safety), and explaining ∼90% of the variance in the training and test data. Conclusion: Two essential metrics, treatment efficacy and treatment safety, are necessary and sufficient to characterize glycemic control in diabetes. Thus, quantitatively, diabetes treatment optimization is reduced to a two-dimensional problem, meaning that minimizing both exposure to hyperglycemia and risk for hypoglycemia will lead to improvement in any other metric of glycemic control.

Glucose Prediction under Variable-Length Time-Stamped Daily Events: A Seasonal Stochastic Local Modeling Framework

Accurate glucose prediction along a long-enough time horizon is a key component for technology to improve type 1 diabetes treatment. Subjects with diabetes might benefit from supervision and control systems that accurately predict risks and trigger corrective actions early enough with improved mitigation. However, large intra-patient variability poses big challenges to glucose prediction. In previous works by the authors, clustering and local modeling techniques with seasonal stochastic models proved to be efficient, allowing for good glucose prediction accuracy for long prediction horizons. Continuous glucose monitoring (CGM) data were partitioned into fixed-length postprandial time subseries and clustered with Fuzzy C-Means to collect similar behaviors, enforcing seasonality at each cluster after subseries concatenation. Then, seasonal stochastic models were identified for each cluster and local predictions were integrated into a global prediction. However, free-living conditions do not support the fixed-length partition of CGM data since daily events duration is variable. In this work, a new algorithm is provided to overcome this constraint, allowing better coping with patient's variability under variable-length time-stamped daily events in supervision and control applications. Besides predicted glucose, two real-time indices are additionally provided-a crispness index, indicating good representation of current glucose behavior by a single model, and a normality index, allowing for the detection of an abnormal glucose behavior (unusual according to registered historical data). The framework is tested in a proof-of-concept in silico study with ten patients over four month training data and two independent two month validation datasets, with and without abnormal behaviors, from the distribution version of the UVA/Padova simulator extended with diverse sources of intra-patient variability.

Seasonal Local Models for Glucose Prediction in Type 1 Diabetes

Linear empirical dynamic models have been widely used for blood glucose prediction and risks prevention in people with type 1 diabetes. More accurate blood glucose prediction models with longer prediction horizon (PH) are desirable to enable warnings to patients about imminent blood glucose changes with enough time to take corrective actions. In this study, a blood glucose prediction method is developed by integrating the predictions of a set of seasonal local models (each of them corresponding to different glucose profiles observed along historical data). In the modeling step, the number of sets and their corresponding glucose profiles characteristics are obtained by clustering techniques (Fuzzy C-Means). Then, Box-Jenkins methodology is used to identify a seasonal model for each set. Finally, blood glucose predictions of local models are integrated using different techniques. The proposed method is tested by using 18 60-h closed-loop experiments (including different exercise types and artificial pancreas strategies) and achieving mean absolute percentage error (MAPE) of 2.94%, 3.89%, 5.41%, 6.29% and 8.66% for 15-, 30-, 45-, 60-, and 90-min PHs, respectively.

Stochastic Seasonal Models for Glucose Prediction in the Artificial Pancreas

BACKGROUND

Linear empirical dynamic models have been widely used for glucose prediction. The extension of the concept of seasonality, characteristic of other domains, is explored here for the improvement of prediction accuracy.

METHODS

Twenty time series of 8-hour postprandial periods (PP) for a same 60g-carbohydrate meal were collected from a closed-loop controller validation study. A single concatenated time series was produced representing a collection of data from similar scenarios, resulting in seasonality. Variability in the resulting time series was representative of worst-case intrasubject variability. Following a leave-one-out cross-validation, seasonal and nonseasonal autoregressive integrated moving average models (SARIMA and ARIMA) were built to analyze the effect of seasonality in the model prediction accuracy. Further improvement achieved from the inclusion of insulin infusion rate as exogenous variable was also analyzed. Prediction horizons (PHs) from 30 to 300 min were considered.

RESULTS

SARIMA outperformed ARIMA revealing a significant role of seasonality. For a 5-h PH, average MAPE was reduced in 26.62%. Considering individual runs, the improvement ranged from 6.3% to 54.52%. In the best-performing case this reduction amounted to 29.45%. The benefit of seasonality was consistent among different PHs, although lower PHs benefited more, with MAPE reduction over 50% for PHs of 60 and 120 minutes, and over 40% for 180 min. Consideration of insulin infusion rate into the seasonal model further improved performance, with a 61.89% reduction in MAPE for 30-min PH and reductions over 20% for PHs over 180 min.

CONCLUSIONS

Seasonality improved model accuracy allowing for the extension of the PH significantly.

Closed-Loop Control of Postprandial Glycemia Using an Insulin-on-Board Limitation Through Continuous Action on Glucose Target

BACKGROUND

Postprandial (PP) control remains a challenge for closed-loop (CL) systems. Few studies with inconsistent results have systematically investigated the PP period.

OBJECTIVE

To compare a new CL algorithm with current pump therapy (open loop [OL]) in the PP glucose control in type 1 diabetes (T1D) subjects.

METHODS

A crossover randomized study was performed in two centers. Twenty T1D subjects (F/M 13/7, age 40.7 ± 10.4 years, disease duration 22.6 ± 9.9 years, and A1c 7.8% ± 0.7%) underwent an 8-h mixed meal test on four occasions. In two (CL1/CL2), after meal announcement, a bolus was given followed by an algorithm-driven basal infusion based on continuous glucose monitoring (CGM). Alternatively, in OL1/OL2 conventional pump therapy was used. Main outcome measures were as follows: glucose variability, estimated with the coefficient of variation (CV) of the area under the curve (AUC) of plasma glucose (PG) and CGM values, and from the analysis of the glucose time series; mean, maximum (C_max), and time to C_max glucose concentrations and time in range (<70, 70-180, >180 mg/dL).

RESULTS

CVs of the glucose AUCs were low and similar in all studies (around 10%). However, CL achieved greater reproducibility and better PG control in the PP period: CL1 = CL2<OL1<OL2 (PG_mean 123 ± 47 and 125 ± 44 vs. 152 ± 53 and 159 ± 54 mg/dL) and C_max OL 217.1 ± 67.0 mg/dL versus CL 183.3 ± 63.9 mg/dL, P < 0.0001. Time-in-range was higher with CL versus OL (80% vs. 64%; P < 0.001). Neither the time below 70 mg/dL (CL 6.1% vs. OL 3.2%; P > 0.05) nor the need for oral glucose was significantly different (CL 40.0% vs. OL 22.5% of meals; P = 0.054).

CONCLUSIONS

This novel CL algorithm effectively and consistently controls PP glucose excursions without increasing hypoglycemia. Study registered at ClinicalTrials.gov : study number NCT02100488.