Related Experiment Video
Updated: Jun 30, 2026
Quantification of Oculomotor Responses and Accommodation Through Instrumentation and Analysis Toolboxes
Published on: March 3, 2023
Explainable Machine Learning Analysis of Perioperative Factors Associated with Clinically Significant Emergence
Jung A Lim1, Jonghae Kim2, Minju Kong2
1Department of Anesthesiology and Pain Medicine, School of Medicine, Kyungpook National University, Daegu 41944, Republic of Korea.
Insights
Machine learning models can predict emergence agitation (EA) in children after sevoflurane anesthesia. This study developed an explainable model to identify key risk factors for EA in pediatric ophthalmic surgery patients.
Area of Science:
- Anesthesiology
- Machine Learning
- Pediatric Surgery
Background:
- Emergence agitation (EA) is a frequent complication in pediatric patients recovering from sevoflurane anesthesia, especially after ophthalmic procedures.
- Current risk stratification methods for EA are limited in clinical applicability and validation.
- There is a need for reliable tools to predict individualized EA risk.
Purpose of the Study:
- To develop and internally validate an explainable machine learning model for estimating individualized EA risk in children undergoing ophthalmic surgery.
- To identify key perioperative factors contributing to EA using machine learning interpretability techniques.
Main Methods:
- A retrospective cohort study of 1029 children (3-7 years) undergoing ophthalmic surgery under sevoflurane anesthesia.
- Development and validation of four machine learning algorithms (logistic regression, random forest, XGBoost, CatBoost) using 5-fold cross-validation.
- Performance evaluation included discrimination, calibration, and classification metrics; SHAP analysis was used for interpretability.
Main Results:
- EA occurred in 52.8% of patients.
- The XGBoost model demonstrated strong performance with an AUPRC of 0.827 and sensitivity of 0.796.
- SHAP analysis highlighted airway management and anesthetic-related variables as significant predictors of EA.
Conclusions:
- Machine learning analysis successfully identified perioperative risk factors for EA in pediatric ophthalmic surgery.
- The developed model offers preliminary insights into risk stratification for EA.
- External validation is necessary before clinical implementation of the model.
Abstract:
Background and Objectives: Emergence agitation (EA) is a common neurobehavioral disturbance during recovery from sevoflurane anesthesia in pediatric patients, particularly after ophthalmic surgery. Clinically deployable and rigorously validated risk stratification approaches remain limited. We aimed to develop and internally validate an explainable machine learning model to estimate individualized EA risk after pediatric ophthalmic surgery. Materials and Methods: This retrospective cohort study included 1029 children aged 3-7 years who underwent ophthalmic surgery under sevoflurane anesthesia between 2016 and 2025. EA was defined as clinically significant agitation requiring active management in the post-anesthesia care unit. Four machine learning algorithms (regularized logistic regression, random forest, XGBoost, and CatBoost) were developed using stratified patient-level 5-fold cross-validation. Performance was evaluated using pooled out-of-fold predictions. Discrimination, calibration, and classification metrics at the optimal Youden threshold were assessed. SHAP analysis was applied for interpretability. Results: EA occurred in 543 patients (52.8%). XGBoost showed comparable discrimination with slightly higher AUPRC (0.827) and sensitivity (0.796) compared with other models, while maintaining acceptable specificity (0.728). Calibration demonstrated good agreement between predicted and observed risk. SHAP identified airway management and anesthetic-related variables as key contributors. Conclusions: ML-based analysis identified clinically relevant perioperative factors associated with emergence agitation and may provide preliminary insight into perioperative risk stratification pending external validation. External validation is required before clinical implementation.
