What does this research mean for the field?

Machine learning models using routinely available clinical variables can accurately predict major kidney stone components, and the addition of morphological features further improves predictive performance to excellent levels. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.SUPPORTS_CONSENSUS.

What question did this study set out to answer?

To develop, validate, and explain machine learning models that predict dominant urinary stone composition using routine clinical variables and standardized morphological features, quantifying the added value of morphology.

September 2, 2026BJU International

Kidney stones composition prediction using artificial intelligence ( KiSCAI) study

View Full Paper

Ask AI

Bookmark

Authors

LDLouise DuffautParisTech PSPietro ScilipotiVita-Salute San Raffaele University ZKZ. KheneInserm

Discussion

Member takes

Overview

Retrospective cohort study reveals machine learning models predict stone composition from clinical and morphological data, highlighting tools for surgical planning.

Key Points

To develop, validate, and explain machine learning models that predict dominant urinary stone composition using routine clinical variables and standardized morphological features, quantifying the added value of morphology.
Retrospective cohort study evaluated 442 consecutive patients (2019–2024) undergoing endourological treatment or spontaneous stone expulsion with confirmed laboratory composition (>50% dominant component).
Data split into 80% training and 20% validation sets to evaluate clinical and Daudon-based morphological predictors across five stone types (calcium oxalate monohydrate, calcium oxalate dihydrate, calcium phosphate, uric acid, and cystine).
Assessed multiple fine-tuned classifiers (including CatBoost, XGBoost, and random forest) using repeated-resampling multiclass LASSO, macro-averaged one-vs-rest AUC, and SHAP explainability.
Stone composition across the cohort (N=442) comprised calcium oxalate monohydrate in 41.0% (n=181), calcium oxalate dihydrate in 24.9% (n=110), calcium phosphate in 19.7% (n=87), uric acid in 10.9% (n=49), and cystine in 3.6% (n=16).
Models using clinical variables alone achieved good predictive discrimination with macro-AUC reaching up to 0.809 in the validation cohort.
Integrating Daudon-based morphological features improved discrimination to a macro-AUC of up to 0.983 using CatBoost, with morphological descriptors ranking as top contributors by SHAP analysis.

Cite This Study

Duffaut et al. (2026) studied this question.

synapsesocial.com/papers/6a97e249c562ede874ec64d9 https://doi.org/10.1111/bju.70442

View Full Paper

Ask AI

Bookmark

September 2, 2026BJU International

Kidney stones composition prediction using artificial intelligence ( KiSCAI) study

View Full Paper

Ask AI

Bookmark

Authors

LDLouise DuffautParisTech PSPietro ScilipotiVita-Salute San Raffaele University ZKZ. KheneInserm

Discussion

Member takes

Overview

Retrospective cohort study reveals machine learning models predict stone composition from clinical and morphological data, highlighting tools for surgical planning.

Key Points

To develop, validate, and explain machine learning models that predict dominant urinary stone composition using routine clinical variables and standardized morphological features, quantifying the added value of morphology.
Retrospective cohort study evaluated 442 consecutive patients (2019–2024) undergoing endourological treatment or spontaneous stone expulsion with confirmed laboratory composition (>50% dominant component).
Data split into 80% training and 20% validation sets to evaluate clinical and Daudon-based morphological predictors across five stone types (calcium oxalate monohydrate, calcium oxalate dihydrate, calcium phosphate, uric acid, and cystine).
Assessed multiple fine-tuned classifiers (including CatBoost, XGBoost, and random forest) using repeated-resampling multiclass LASSO, macro-averaged one-vs-rest AUC, and SHAP explainability.
Stone composition across the cohort (N=442) comprised calcium oxalate monohydrate in 41.0% (n=181), calcium oxalate dihydrate in 24.9% (n=110), calcium phosphate in 19.7% (n=87), uric acid in 10.9% (n=49), and cystine in 3.6% (n=16).
Models using clinical variables alone achieved good predictive discrimination with macro-AUC reaching up to 0.809 in the validation cohort.
Integrating Daudon-based morphological features improved discrimination to a macro-AUC of up to 0.983 using CatBoost, with morphological descriptors ranking as top contributors by SHAP analysis.

Cite This Study

Duffaut et al. (2026) studied this question.

synapsesocial.com/papers/6a97e249c562ede874ec64d9 https://doi.org/10.1111/bju.70442

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark