Research Article: Efficacy and safety of Oliceridine versus Sufentanil in postoperative analgesia for burn skin grafting: a machine learning and SHAP-based cohort study
Abstract:
Balancing profound analgesia with the minimization of opioid-related adverse events, particularly postoperative nausea and vomiting (PONV), remains challenging in burn surgery. This study aimed to use machine learning to evaluate the comparative efficacy and safety of the novel biased agonist Oliceridine versus Sufentanil.
A retrospective cohort of 586 patients undergoing burn skin grafting was analyzed. XGBoost algorithms were developed to predict the need for rescue analgesia (efficacy) and PONV occurrence (safety). The Shapley additive explanations (SHAP) were applied to decipher feature contributions and drug effects.
The XGBoost efficacy model (AUC?=?0.843) identified total burn surface area and surgery duration as dominant predictors, whereas opioid choice had negligible SHAP impact, suggesting a comparable predictive analgesic profile between Oliceridine and Sufentanil in this cohort. Conversely, the XGBoost PONV model (AUC?=?0.788) outperformed logistic regression, identifying Sufentanil administration and female patients as paramount risk drivers. SHAP interaction analysis revealed a strong predictive association in which substituting Sufentanil with Oliceridine correlated with a substantially blunted predicted PONV risk trajectory in female patients.
Machine learning evaluation suggests a distinction between opioid predicted efficacy and safety profiles in this cohort. Oliceridine showed similar predicted analgesic outcomes to Sufentanil while being associated with a significantly lower predicted risk of PONV. These exploratory insights provide hypothesis-generating frameworks for risk stratification in severe burn management, though further prospective trials are required to establish any clinical validity.
Introduction:
Balancing profound analgesia with the minimization of opioid-related adverse events, particularly postoperative nausea and vomiting (PONV), remains challenging in burn surgery. This study aimed to use machine learning to evaluate the comparative efficacy and safety of the novel biased agonist Oliceridine versus Sufentanil.
Read more