Machine learning-based prediction of hernia risk in peritoneal dialysis patients: a comparative study of models and SHAP-driven interpretability analysis

Back to news list

Source: Frontiers Medicine

Original: https://www.frontiersin.org/articles/10.3389/fmed.2026.1687055...

Published: 2026-03-04T00:00:00Z

The study aimed to predict hernia risk in peritoneal dialysis (PD) patients using machine learning models and SHAP interpretability analysis. It included 1144 patients with PD from 2010–2024, divided into a training cohort (n=800) and an external validation (n=344). They developed nine ML models, the best being Random Forest (RF) with a training AUC of 97.99% and a validation AUC of 93.66%. RF identified nine major risk factors: abdominal circumference, smoking, history of smoking, history of COPD, and CAPD modality. SHAP analysis explained the non-linear effects of these factors. They developed an online tool on R Shiny that enables real-time risk calculation, risk stratification and personalized recommendations. The RF model achieves high accuracy and interpretability, thus facilitating early intervention and improving patient prognosis.