Combined Use of Glucose-Specific Model Identification and Alarm Strategy Based on Prediction-Funnel to Improve Online Forecasting of Hypoglycemic Events

Abstract

Background: Advanced decision support systems for type 1 diabetes (T1D) management often embed prediction modules, which allow T1D people to take preventive actions to avoid critical episodes like hypoglycemia. Real-time prediction of blood glucose (BG) concentration relies on a subject-specific model of glucose-insulin dynamics. Model parameter identification is usually based on the mean square error (MSE) cost function, and the model is usually used to predict BG at a single prediction horizon (PH). Finally, a hypo-alarm is raised if the predicted BG crosses a threshold. This work aims to show that real-time hypoglycemia forecasting can be improved by leveraging: a glucose-specific mean square error (gMSE) cost function in model’s parameters identification, and a “prediction-funnel,” that is, confidence intervals (CIs) for multiple PHs, within the hypo-alarm-raising strategy. Methods: Autoregressive integrated moving average with exogenous input (ARIMAX) models are selected to illustrate the proposed solution (use of gMSE and prediction-funnel) and its assessment against the conventional approach (MSE and single PH). The gMSE penalizes the model misfit in unsafe BG ranges (e.g., hypoglycemia), and the prediction-funnel allows raising an alarm by monitoring if the CIs cross a suitable threshold. The algorithms were evaluated by measuring precision ( P), recall ( R), F1-score ( F1), false positive per day (FP/day), and time gain (TG) on a real dataset collected in 11 T1D individuals. Results: The best performance is achieved exploiting both the gMSE and the prediction-funnel: P = 65%, R = 88%, F1 = 75%, FP/day = 0.29, and mean TG = 15 minutes. Conclusions: The combined use of a glucose-specific metric and an alarm-raising strategy based on the prediction-funnel allows achieving a more effective and reliable hypoglycemia prediction algorithm.

Artificial Intelligence and Machine Learning for Improving Glycemic Control in Diabetes: Best Practices, Pitfalls, and Opportunities

“…Del Favero, Facchinetti and Cobelli [141] described a glucose-range-specific metric to better capture the increased risk of making errors in different ranges of glucose such that errors made in dangerous regions were weighted more heavily than those made in less dangerous regions. This group also showed in Faccioli et al [142] that through use of this glucose-specific weighted error term, they could improve forecasting of hypoglycemia. Cameron et al also report on a risk-based closed-loop algorithm that prioritizes mitigation of the increased risk of hypoglycemia compared with hyperglycemia [143].…”

“…Best practice 19: When designing an algorithm to predict an event such as low glucose (<70 mg/dL), it may be optimal to design a binary classifier rather than a classifier based on a regression algorithm whereby there is a penalty for failing to identify the binary event. Alternatively, it could be optimal to use a glucose-range-specific penalty [141,142].…”