Integrating Neural Differential Forecasting with Safe Reinforcement Learning for Blood Glucose Regulation
This addresses the challenge of achieving both personalized and risk-aware glucose control for Type 1 Diabetes patients, representing an incremental improvement by combining existing methods with safety enhancements.
The paper tackled the problem of automated insulin delivery for Type 1 Diabetes by balancing glucose control and safety under uncertain conditions, proposing TSODE, a controller that integrates Thompson Sampling RL with a NeuralODE forecaster and conformal calibration. It achieved 87.9% time-in-range with less than 10% time below 70 mg/dL in simulations, outperforming baselines.
Automated insulin delivery for Type 1 Diabetes must balance glucose control and safety under uncertain meals and physiological variability. While reinforcement learning (RL) enables adaptive personalization, existing approaches struggle to simultaneously guarantee safety, leaving a gap in achieving both personalized and risk-aware glucose control, such as overdosing before meals or stacking corrections. To bridge this gap, we propose TSODE, a safety-aware controller that integrates Thompson Sampling RL with a Neural Ordinary Differential Equation (NeuralODE) forecaster to address this challenge. Specifically, the NeuralODE predicts short-term glucose trajectories conditioned on proposed insulin doses, while a conformal calibration layer quantifies predictive uncertainty to reject or scale risky actions. In the FDA-approved UVa/Padova simulator (adult cohort), TSODE achieved 87.9% time-in-range with less than 10% time below 70 mg/dL, outperforming relevant baselines. These results demonstrate that integrating adaptive RL with calibrated NeuralODE forecasting enables interpretable, safe, and robust glucose regulation.