Enhancing Glucose Level Prediction of ICU Patients through Hierarchical Modeling of Irregular Time-Series

Accurately predicting blood glucose (BG) levels of ICU patients is critical, as both hypoglycemia (BG < 70 mg/dL) and hyperglycemia (BG > 180 mg/dL) are associated with increased morbidity and mortality. This study presents a proof-of-concept machine learning framework, the Multi-source Irregular Time-Series Transformer (MITST), designed to predict blood glucose (BG) levels in ICU patients. Unlike existing approaches that rely on manual feature engineering or are limited to a small number of Electronic Health Record (EHR) data sources, MITST demonstrates the feasibility of integrating diverse clinical data (e.g., lab results, medications, vital signs) and handling irregular time-series data without predefined aggregation. MITST employs a hierarchical architecture of Transformers, comprising feature-level, timestamp-level, and source-level components, to capture fine-grained temporal dynamics and enable learning-based data integration. This eliminates the need for traditional aggregation and manual feature engineering. In a large-scale evaluation using the eICU database (200,859 ICU stays across 208 hospitals), MITST achieves an average improvement of 1.7% (p < 0.001) in AUROC and 1.8% (p < 0.001) in AUPRC over a state-of-the-art baseline. For hypoglycemia, MITST achieves an AUROC of 0.915 and an AUPRC of 0.247, both significantly outperforming the baseline. The flexible architecture of MITST allows seamless integration of new data sources without retraining the entire model, enhancing its adaptability for clinical decision support. While this study focuses on predicting BG levels, MITST can easily be extended to other critical event prediction tasks in ICU settings, offering a robust solution for analyzing complex, multi-source, irregular time-series data.