Triglyceride-glucose index and kidney stone recurrence in nondiabetic patients: insights from conventional statistical and machine learning models
The triglyceride-glucose (TyG) index, an emerging marker of insulin resistance, is correlated with the presence and recurrence of kidney stones (KS). The nondiabetes mellitus (DM) population is often overlooked but has a relatively high risk. The present study aimed to analyze the association between TyG and KS recurrence in non-DM patients and developed a machine learning model to predict recurrence risk. We retrospectively collected data from KS patients treated at our hospital between January 2020 and January 2025, selected a nondiabetic population, and used univariate analysis, logistic regression, and subgroup analyses to evaluate the association between TyG and KS recurrence. Predictive variables were screened using Pearson correlation, the Boruta algorithm and least absolute shrinkage and selection operator (LASSO) methods. A total of 7 machine learning algorithms were compared, and SHapley Additive exPlanations (SHAP) was employed to interpret the optimal predictive model. The final analysis included 660 patients. This study demonstrated that the TyG index was related to KS recurrence in non-DM individuals. Additionally, age and hypertension were significant interaction factors in the relationship between TyG grade and stone recurrence. The logistic regression model demonstrated stable and superior performance. Furthermore, SHAP analysis revealed that TyG was the most influential predictor in the ML model’s decision process. Our findings revealed a positive relationship between the TyG index and a higher risk of recurrent KS among non-DM adults. We further established an interpretable ML model to precisely identify the risk of KS recurrence in this specific population. Additionally, SHAP analysis deepens the understanding of feature contribution magnitudes and provides empirical support for the formulation of individualized therapeutic strategies.
Authors
- Bingbing Hou (ORCID: https://orcid.org/0000-0002-5352-1622)
- Zongyao Hao (ORCID: https://orcid.org/0000-0003-3399-3689)
- Yuexian Xu (ORCID: https://orcid.org/0000-0003-0986-5505)
- Wanqing Liu (ORCID: https://orcid.org/0000-0001-7727-6900)
- Yun Zhang (ORCID: https://orcid.org/0000-0001-7618-8677)
- Dekai Hu
Institutions
- Anhui Medical University (CN)
- First Affiliated Hospital of Anhui Medical University (CN)
- Hefei First People's Hospital (CN)
Publication Details
- Journal
- BMC Medical Informatics and Decision Making
- Published
- 2026-11-15
- DOI
- https://doi.org/10.1186/s12911-026-03968-2
- Primary Topic
- Kidney Stones and Urolithiasis Treatments
- Type
- article
- Field-Weighted Citation Impact
- 0.00