Adverse event prediction in propofol-remimazolam tosilate anesthesia

TL;DR

This study created a machine learning model to predict adverse events during anesthesia using propofol and remimazolam tosilate, finding that surgical duration and hemodynamic stability are key predictors.

Contribution

A novel predictive model using machine learning to forecast anesthesia-related adverse events in patients receiving propofol-remimazolam tosilate.

Findings

The random forest model achieved an AUC of 0.814 in training and 0.777 in validation for predicting adverse events.

Surgical duration and anesthetic drug dosage ratio were identified as the most important predictive features.

Hemodynamic stability and respiratory recovery status were significant predictors of adverse events.

Abstract

This study aimed to develop and validate a predictive model for anesthesia-related adverse events (ARAEs) in patients receiving propofol combined with remimazolam tosilate, based on perioperative clinical indicators.

A retrospective study was conducted on patients who underwent propofol-remimazolam tosilate anesthesia at our hospital from January 2021 to December 2024. The cohort was divided into a training set (n = 218, 70%) and a validation set (n = 94, 30%). Demographic characteristics, vital sign monitoring data, laboratory test results, and anesthesia recovery parameters were collected. Independent predictors of ARAEs were identified through univariate and multivariate logistic regression analyses. Machine learning algorithms, including random forest (RF), support vector machine, and gradient boosting, were employed to construct predictive models. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA). The optimal model was selected, and feature importance was analyzed.

No significant differences were observed in baseline characteristics between the training and validation sets (P > 0.05). Univariate analysis and multivariate logistic regression identified surgical duration, intraoperative hypotension incidence, spontaneous breathing recovery time, serum creatinine, and arterial carbon dioxide partial pressure as independent risk factors for ARAEs (all P < 0.05). Among the machine learning models, the RF model demonstrated the highest discriminative ability in both the training (AUC 0.814, 95% CI: 0.738–0.889) and validation sets (AUC 0.777, 95% CI: 0.640–0.913), along with superior calibration and clinical net benefit. Feature importance analysis showed that surgical duration, and anesthetic drug dosage ratio were the most critical predictive factors.

The RF model, developed using key perioperative indicators, effectively predicts the risk of ARAEs during propofol-remimazolam tosilate anesthesia. Surgical duration, hemodynamic stability, and respiratory recovery status are the most significant predictors.

Figures