Deep Learning-Based Hypoglycemia Classification Across Multiple Prediction Horizons

2025-03-25

AI Summary

This study addresses the challenge of multi-horizon hypoglycemia risk prediction for type 1 diabetes patients. We propose a unified modeling framework that, for the first time, jointly covers nine prediction horizons—from ultra-short-term (0–30 min) to 24 hours—within a single model, and additionally derive a clinically deployable simplified four-class variant (≤4 h). Our method integrates continuous glucose monitoring, insulin infusion, and accelerometer time-series data, employing a hybrid LSTM-ResNet architecture trained end-to-end via sliding windows and class-balancing strategies. Personalized LSTM models achieve recall rates of 98%, 72%, and 50% for the 0–5, 5–15, and 15–30 min horizons, respectively; the population-level six-class model attains ≥60% event detection across all horizons—significantly outperforming long-horizon baselines. The framework delivers an interpretable, deployable decision-support tool for dynamic risk stratification and early clinical intervention.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Humans and AI: Human-in-the-loop Machine Learning
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Type 1 diabetes (T1D) management can be significantly enhanced through the use of predictive machine learning (ML) algorithms, which can mitigate the risk of adverse events like hypoglycemia. Hypoglycemia, characterized by blood glucose levels below 70 mg/dL, is a life-threatening condition typically caused by excessive insulin administration, missed meals, or physical activity. Its asymptomatic nature impedes timely intervention, making ML models crucial for early detection. This study integrates short- (up to 2h) and long-term (up to 24h) prediction horizons (PHs) within a single classification model to enhance decision support. The predicted times are 5-15 min, 15-30 min, 30 min-1h, 1-2h, 2-4h, 4-8h, 8-12h, and 12-24h before hypoglycemia. In addition, a simplified model classifying up to 4h before hypoglycemia is compared. We trained ResNet and LSTM models on glucose levels, insulin doses, and acceleration data. The results demonstrate the superiority of the LSTM models when classifying nine classes. In particular, subject-specific models yielded better performance but achieved high recall only for classes 0, 1, and 2 with 98%, 72%, and 50%, respectively. A population-based six-class model improved the results with at least 60% of events detected. In contrast, longer PHs remain challenging with the current approach and may be considered with different models.

Problem

• Classify hypoglycemia across multiple time horizons using deep learning
• Improve early detection of asymptomatic hypoglycemia in T1D patients
• Compare short-term and long-term prediction models for hypoglycemia events

Innovation

• Integrates short and long-term prediction horizons
• Uses ResNet and LSTM models for classification
• Compares subject-specific and population-based models