Machine-Learning-Based Predictive Model for Trifecta Achievement in Robot-Assisted Partial Nephrectomy

Abstract

Objective:

To develop and internally validate a machine-learning-based nomogram for predicting trifecta achievement in patients undergoing robot-assisted partial nephrectomy (RAPN).

Materials and Methods:

This retrospective single-center study included 426 patients who underwent RAPN between 2011 and 2025 years. Trifecta was defined as negative surgical margins, warm ischemia time ≤25 minutes, and absence of perioperative complications. Variable selection was performed using the Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression. Selected predictors were incorporated into a multivariate logistic regression model to construct a nomogram. Model performance was assessed using discrimination, calibration, and decision-curve analysis (DCA) with internal bootstrap validation.

Results:

Trifecta was achieved in 87.7% of patients. The total operative time, estimated blood loss, tumor location, collecting system entry, and need for perioperative blood transfusion were identified as independent predictors. The model demonstrated good discrimination (area under the curve = 0.792, 95% confidence interval: 0.738–0.845). Calibration was acceptable (bootstrap–corrected slope 0.91, intercept 0.13), and the DCA showed a higher net benefit than treat-all/none for thresholds of ∼0.10–0.75.

Conclusions:

This LASSO-based nomogram offers individualized and clinically interpretable prediction of trifecta achievement in RAPN. Integrating key surgical and tumor–related parameters has the potential to support perioperative decision-making and patient counseling. External validation through prospective multicenter studies is warranted to confirm its generalizability and facilitate clinical implementation.