Routine Urinary Biochemistry Does Not Accurately Predict Stone Type Nor Recurrence in Kidney Stone Formers: A Multi-Centre, Multi-Model, Externally Validated Machine-Learning Study

Objectives: Urinary biochemistry is used to detect and monitor conditions associated with recurrent kidney stones. There are no predictive machine learning (ML) tools for kidney stone type or recurrence. We therefore aimed to build and validate ML models for these outcomes using age, gender, 24-hour urine biochemistry and stone composition. Materials and Methods: Data from 3 cohorts were used, Southampton, UK (n=3013), Newcastle, UK (n=5984) and Bern, Switzerland (n=794). Of these 3130 had available 24-hour urine biochemistry measurements (calcium, oxalate, urate, pH, volume), and 1684 had clinical data on kidney stone recurrence. Two predictive models were constructed (UK and Swiss) using two ML techniques (Partitioning and Random Forests [RF]) and validated internally with a subset of the same dataset (e.g UK model/UK test set), and externally with the other dataset (UK model/Swiss test set). Results and Limitations: For kidney stone type, on external validation accuracy of UK RF model=0.79 (95% CI: 0.73-0.84), sensitivity: calcium oxalate=0.99 and calcium phosphate/urate=0.00. Specificity: calcium oxalate=0.00 and calcium phosphate/urate=0.99. For the Swiss RF model accuracy=0.87 (95% CI: 0.83-0.89), sensitivity: calcium oxalate=0.99 and calcium phosphate/urate=0.00. Specificity: calcium oxalate=0.00, calcium phosphate=0.00 and urate=1.00. For stone recurrence, on external validation accuracy of UK RF model=0.22 (95% CI: 0.19-0.25), sensitivity=0.93 and specificity=0.09. Swiss RF model accuracy=0.42 (95% CI: 0.39-0.47), sensitivity=0.03 and specificity=0.97. Conclusions: Neither kidney stone type nor kidney stone recurrence can be accurately predicted using modelling tools built using specific 24-hour urinary biochemistry values alone. Further studies to delineate accurate predictive tools should be undertaken using both known and novel risk factors.