|
چکیده
|
Abstract Aims: Preventive counseling for kidney stone disease requires accurate risk identification; however, traditional interpretation fails to capture complex metabolic interactions. This study aimed to develop a machine learning–based screening tool using routine blood and urine data to identify high-risk individuals and provide actionable insights for patient education. Instrument & Methods: This retrospective stidy analyzed data from 952 individuals (714 stone formers and 238 controls) with 28 features. Particle Swarm Optimization was used to select features while balancing accuracy and test reduction. An XGBoost model was developed using 5-fold cross-validation. SHAP analysis was used to identify influential predictors. Findings: Stone formers had higher body amss index, were more frequently male, and showed significant differences in serum uric acid, creatinine, 24-hour urinary calcium, and citrate compared with healthy controls (all p<0.05). The Particle Swarm Optimization-optimized model reduced the number of tests from 28 to 8 features (71% reduction) while achieving 93.49% accuracy (95% CI: 91.8–95.0%). The final panel included age, blood_phosphorus, blood_uric_acid, blood_creatinine, eGFR, urine_pH, Urine_Calcium_24h, and urine_citrate_24h. SHAP identified modifiable risk factors, including low citrate (potassium citrate), elevated calcium (dietary modification), elevated uric acid (purine reduction), and reduced eGFR (renal evaluation). Conclusion: This tool enables noninvasive risk identification using routine data, supporting targeted preventive counseling in primary care.
|