Personalized prediction model generated with machine learning for kidney function one year after living kidney donation

TL;DR

A machine learning model predicts kidney function after living donation, offering better accuracy than traditional methods.

Contribution

A novel machine learning model for predicting post-donation kidney function using preoperative clinical data.

Findings

The ML model achieved a median absolute error of 0.079 mg/dL in predicting serum creatinine levels.

The model outperformed conventional methods with higher R2 and lower error metrics.

Preoperative creatinine and remnant kidney volume were key predictive variables.

Abstract

Living kidney donors typically experience approximately a 30% reduction in kidney function after donation, although the degree of reduction varies among individuals. This study aimed to develop a machine learning (ML) model to predict serum creatinine (Cre) levels at one year post-donation using preoperative clinical data, including kidney-, fat-, and muscle-volumetry values from computed tomography. A total of 204 living kidney donors were included. Symbolic regression via genetic programming was employed to create an ML-based Cre prediction model using preoperative clinical variables. Validation was conducted using a 7:3 training-to-test data split. The ML model demonstrated a median absolute error of 0.079 mg/dL for predicting Cre. In the validation cohort, it outperformed conventional methods (which assume post-donation eGFR to be 70% of the preoperative value) with higher R2 (0.58 vs. 0.27), lower root mean squared error (5.27 vs. 6.89), and lower mean absolute error (3.92 vs. 5.8). Key predictive variables included preoperative Cre and remnant kidney volume. The model was deployed as a web application for clinical use. The ML model offers accurate predictions of post-donation kidney function and may assist in monitoring donor outcomes, enhancing personalized care after kidney donation.

Figures