Transformer-based deep learning model for real-time prediction of intraoperative hypotension using dynamic time-series vital signs: A retrospective study - PubMed
3 hours ago
- #Transformer Model
- #Real-time Prediction
- #Intraoperative Hypotension
- Developed Transformer-based deep learning model for real-time intraoperative hypotension (IOH) prediction using dynamic vital sign time-series data.
- Model trained on 319,699 surgical cases from China and externally validated with South Korean data, showing strong performance (AUCs 0.904-0.882).
- Compared to XGBoost, Transformer had higher recall and better calibration, while XGBoost had higher accuracy and specificity.
- IOH burden significantly associated with postoperative acute kidney injury (AKI) and acute kidney disease (AKD).
- Study limited by retrospective design; prospective multicenter validation needed for clinical implementation.