Using Machine Learning for Non-Invasive Detection of Kidney Stones Based on Laboratory Test Results: A Case Study...

Area of Science:

• Urology
• Medical Informatics
• Machine Learning

Background:

• Kidney stone disease is a prevalent global health issue.
• Current diagnostic methods like CT scans are expensive, time-consuming, and involve radiation exposure.
• Delays in diagnosis can lead to severe complications.

Purpose of the Study:

• To develop and evaluate a machine learning model for early kidney stone detection using routine laboratory test results.
• To explore the efficacy of various machine learning algorithms and data preprocessing techniques for this diagnostic task.

Main Methods:

• Utilized a dataset of 2156 patient records from a Saudi Arabian hospital.
• Applied machine learning algorithms, including ensemble tree-based classifiers (Random Forest, Extra Tree).
• Employed data imputation and class imbalance handling techniques (oversampling, undersampling).

Main Results:

• Random Forest achieved 99% accuracy, 98% recall, and 99% F1 score.
• Extra Tree classifier demonstrated strong performance with a 92% F1 score.
• Ensemble tree-based models proved highly effective in detecting renal stones.

Conclusions:

• Routine laboratory tests, analyzed via machine learning, offer a promising non-invasive and cost-effective method for early kidney stone detection.
• This approach can significantly improve diagnostic speed and patient management in urology.
• Machine learning integration into routine diagnostics can enhance healthcare accessibility and efficiency.