Related Experiment Video
Updated: Jul 1, 2026
07:31
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Risk prediction models based on machine learning for emergence delirium in elderly patients undergoing spine surgery:
Qiuyi Wang1,2, Zhilin Zhang1, Weijun Luo3
1Department of Nursing, Naval Medical University, 800 Xiangyin Road, Yangpu District, Shanghai, 200433, China.
BMC Anesthesiology
|June 26, 2026
Summary
This study developed a machine learning model to predict emergence delirium (ED) in elderly patients undergoing spinal surgery. The logistic regression model identified key predictors like age and education, aiding risk stratification.
Area of Science:
- Anesthesiology
- Geriatric Medicine
- Neurosurgery
- Artificial Intelligence in Medicine
Background:
- Emergence delirium (ED) is a frequent complication in elderly patients undergoing surgery for degenerative spinal disease (DSD).
- ED is linked to adverse postoperative outcomes, necessitating improved risk prediction.
- Predictive models can enhance perioperative risk stratification and personalized patient monitoring.
Purpose of the Study:
- To develop and validate a machine learning (ML) model for predicting the risk of emergence delirium (ED).
- To identify key perioperative predictors of ED in elderly patients undergoing DSD surgery.
- To assess the interpretability and performance of ML models for ED risk prediction.
Main Methods:
- A prospective observational cohort study of 430 elderly patients undergoing spinal surgery.
- Data collected across five perioperative domains: patient characteristics, preoperative, intraoperative, anesthesia-related, and postoperative factors.
- Least Absolute Shrinkage and Selection Operator (LASSO) regression for feature selection, followed by ML model development and validation using SHapley Additive exPlanations (SHAP) for interpretability.
Main Results:
- Nine key predictors were identified, including age, education level, postoperative pain, and midazolam exposure.
- A logistic regression model demonstrated the best performance (AUC 0.832, accuracy 0.769).
- Key predictors influencing ED risk included education, age, midazolam exposure, intraoperative bleeding, and operation duration, with potential synergistic effects.
Conclusions:
- An interpretable logistic regression model offers acceptable performance for predicting ED risk in elderly spinal surgery patients.
- The model can aid in anesthesia-informed perioperative risk stratification and individualized monitoring.
- The model's moderate sensitivity suggests its use as an adjunct to clinical judgment, not a standalone tool.