A new machine-learning model to predict long-term renal function impairment after minimally invasive partial...

Area of Science:

• Nephrology
• Urology
• Medical Informatics

Background:

• Partial nephrectomy (PN) is a standard treatment for kidney tumors.
• Assessing long-term renal function after PN is crucial for patient management.
• Predicting chronic kidney disease (CKD) progression post-PN remains a challenge.

Purpose of the Study:

• To develop and validate a predictive model for long-term renal function impairment following minimally invasive partial nephrectomy (PN).
• To identify key predictors of chronic kidney disease (CKD) stage migration after PN.

Main Methods:

• Analysis of data from 381 patients who underwent minimally invasive PN between 2005 and 2022 with at least 12 months follow-up.
• Utilized a classification and regression tree (CART) machine-learning algorithm.
• Identified predictors and created patient clusters based on CKD stage migration and risk factors.

Main Results:

• 103 patients (27%) experienced CKD stage migration.
• Key predictors identified: perioperative loss of renal function, age-adjusted Charlson comorbidity index (ACCI), and baseline CKD stage.
• Four distinct patient clusters were defined, with CKD progression rates ranging from 6.9% to 69.6%.
• The model achieved a c-index of 0.75.

Conclusions:

• A novel model effectively predicts long-term renal function impairment after PN.
• Perioperative renal function loss is a critical factor in predicting recovery.
• The model aids in identifying high-risk patients for intensified functional follow-up to mitigate kidney function decline.