SOURCE-LINKED INTELLIGENCE

XGlucoNet++: A Transformer-Based, Explainable and Safe Decision Support System for Personalized Glycemic Control in Intensive Care Units

CORDIS · observation · Publication date unknown

uantification, the system will address the critical challenge of safe insulin therapy through real-time glucose forecasting and hypoglycemia risk prediction. Its hybrid architecture combines temporal deep learning, patient-specific graph representations, and evidence-aware policy learning. Aligned with Horizon Europe’s values, the project embeds ethical AI design, subgroup fairness assessments, and open science practices. Validation will be conducted on both retrospective and prospective ICU datasets, in collaboration with clinical experts, to enhance trust, transparency, and generalizability. The fellowship will accelerate the researcher’s leadership in explainable AI for health while enabling mutual knowledge exchange with the host institution. Explainable AI, Reinforcement Learning, ICU Decision Support, Time-Series Forecasting, Graph Neural Networks, Glycemic Control, Medical AI Ethics, Model Calibration, Precision Medicine.

Read original source ↗ Open in workspace

recordType

award

status

SIGNED

region

EU

value

179006.16

unit

EUR