Random forest-based prediction of early postoperative pain after pelvic organ prolapse surgery under standardized remimazolam anesthesia.

backgroundPelvic organ prolapse (POP) surgery is a common procedure requiring effective anesthesia. Remimazolam, a novel benzodiazepine-based anesthetic agent, has shown promise in clinical practice. However, predicting its efficacy in managing postoperative pain remains challenging. This research focused on developing and validating artificial intelligence (AI) models to predict the effectiveness of remimazolam anesthesia in patients undergoing POP surgery.

methodsIn this single-centre retrospective observational study, clinical data from 150 women undergoing pelvic organ prolapse surgery under a standardized remimazolam-based anaesthetic protocol were analysed. Candidate predictors included demographic characteristics, comorbidities, baseline pain history, intraoperative haemodynamic and bispectral index (BIS) values, and remifentanil dose. Univariate logistic regression and least absolute shrinkage and selection operator (LASSO) regression were used for feature selection. A random forest (RF) classifier was prespecified as the primary prediction model. Logistic regression (LR) was used as a conventional statistical baseline, and a support vector machine (SVM) model was explored only in supplementary sensitivity analyses. The dataset was randomly split into a training set (70%) and a test set (30%), and model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration, decision curve analysis, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).

resultsAmong all candidate variables, age, duration of illness, diabetes mellitus, pre-anaesthesia (T0) heart rate, pre-anaesthesia (T0) BIS, and BIS at awakening (T3) were retained in the final model and were significantly associated with moderate pain (visual analogue scale > 3) at 2 h postoperatively. In the test set, the random forest model achieved an AUC of 0.9412, outperforming logistic regression (AUC 0.9139) and SVM (AUC 0.8992). The random forest model showed good calibration and net clinical benefit and yielded a sensitivity of 0.88, specificity of 0.85, PPV of 0.87, and NPV of 0.86 for predicting moderate postoperative pain.

conclusionsIn patients undergoing pelvic organ prolapse surgery under a standardized remimazolam anaesthetic regimen, a random forest model based on a small set of routinely available perioperative variables accurately predicted early postoperative pain. This approach may help individualise perioperative pain management and identify patients who require intensified analgesic strategies, although external validation and prospective impact studies are needed before clinical implementation.