Predictive Capacities of a Machine Learning Decision Tree Model Created to Analyse Feasibility of an Open or Robotic Kidney Transplant

Int J Med Robot. 2024 Dec;20(6):e70035. doi: 10.1002/rcs.70035.

Abstract

Background: Machine learning has emerged as a potent tool in healthcare. A decision tree model was built to improve the decision-making process when determining the optimal choice between an open or robotic surgical approach for kidney transplant.

Methods: 822 patients (OKT) and 169 (RKT) underwent kidney transplantation at our centre during the study period. A decision tree model was built in a two-step process consisting of: (1) Creating the model on the training data and (2) testing the predictive capabilities of the model using the test data.

Results: Our model correctly predicted an OKT in 148 patients out of 161 test cases who received an OKT (accuracy 91%) and predicted an RKT in 19 out of 25 test cases of patients receiving an RKT (accuracy 76%).

Conclusion: Our model represents the inaugural data-driven model that furnishes concrete insights for the discernment between employing robotic and open surgery techniques.

Keywords: DGF; artificial intelligence; kidney transplantation; machine learning; robotics in transplantation.

MeSH terms

  • Adult
  • Aged
  • Decision Trees*
  • Feasibility Studies*
  • Female
  • Humans
  • Kidney Transplantation* / methods
  • Machine Learning*
  • Male
  • Middle Aged
  • Robotic Surgical Procedures* / methods