Predictive value of machine learning model based on CT values for urinary tract infection stones

Area of Science:

• Urology
• Medical Imaging
• Artificial Intelligence

Background:

• Preoperative diagnosis of infection stones is clinically challenging.
• Accurate identification is crucial for effective patient management.
• Current diagnostic methods have limitations.

Purpose of the Study:

• To develop and evaluate a machine learning model for preoperative prediction of infection stones.
• To utilize computed tomography (CT) values for in vivo identification.
• To compare the performance of different machine learning algorithms.

Main Methods:

• A retrospective study included 1209 patients undergoing urinary lithotripsy.
• Seven machine learning algorithms were trained using eleven preoperative variables.
• Model performance was assessed using Area Under the Curve (AUC) and Area Under the Precision-Recall Curve (AUPR).

Main Results:

• All seven models showed strong discrimination on the validation set (AUC: 0.687-0.947).
• The XGBoost model demonstrated superior performance compared to the traditional Logistic Regression (LR) model.
• XGBoost achieved the highest predictive accuracy for infection stones.

Conclusions:

• The XGBoost model is the first machine learning approach for preoperative prediction of infection stones based on CT values.
• This model offers rapid and accurate in vitro identification of infection stones.
• It provides valuable guidance for urologists in stone management strategies.