PhysioSeq2Seq: A Hybrid Physiological Digital Twin and Sequence-to-Sequence LSTM for Long-Horizon Glucose Forecasting in Type 1 Diabetes

In three linesPhysioSeq2Seq combines patient-specific physiological digital twin modeling with Seq2Seq LSTM for 240-minute glucose forecasting in type 1 diabetes. Trained on 348 participants (T1DEXI), evaluated on 74: MAE 39.28 mg/dL at 240-min horizon, reducing bias by 13.89 mg/dL vs recursive LSTM.

Summary generated by Claude — human-verified